The 'Add to group' dropdown on /admin/users/{id} silently filtered out
every Google-Workspace-managed group (rightly — the API would 409 on
POST). On deployments where Admin and Everyone are both Workspace-mapped
via AGNES_GROUP_{ADMIN,EVERYONE}_EMAIL and no custom Agnes groups exist
yet (FoundryAI prod + dev today), the picker showed only the literal
'— Pick a group —' option with the 'Add' button disabled. Operator had
no indication that they needed to create a custom group first.
Three states surface a hint below the picker now:
- user is already in every group (literally nothing left)
- every remaining group is Google-Workspace-managed (link to
/admin/groups + admin.google.com explainer)
- no groups exist at all
The skip-google-managed logic stays — POST would still 409 on those
rows, this just stops the empty-state from being a silent dead end.
Adds end-to-end flow for installing and keeping the per-user filtered
Claude Code marketplace in sync with the user's Agnes stack
(admin RBAC grants \ MyAIStack opt-outs U /store installs).
Setup (one-liner in install prompt step 5):
`agnes refresh-marketplace --bootstrap` clones the per-user marketplace
bare repo to ~/.agnes/marketplace, strips PAT from the cloned origin
URL, registers the local path with Claude Code, and installs every
plugin in the served manifest at --scope project. Replaces a 15-line
inline shell sequence that tripped Claude Code's agent-driven `rm -rf`
permission gate.
Auto-refresh (SessionStart hook installed by `agnes init`):
`agnes refresh-marketplace --quiet` runs every Claude Code session,
fetches+resets the clone (server rebuilds as orphan commits, so
pull --ff-only is impossible), and version-aware reconciles:
- missing in workspace -> claude plugin install <name>@agnes --scope project
- version differs -> claude plugin update <name>@agnes
- matches -> skip
Don't auto-uninstall plugins that disappeared from the manifest --
a transient empty manifest from the server would wipe the stack.
Hook output: when --quiet AND something actually changed, emits Claude
Code hook JSON on stdout -- `systemMessage` (transient toast) and
`hookSpecificOutput.additionalContext` (model-side system reminder),
both carrying the change summary plus a "/exit + restart Claude Code"
instruction (Claude only scans plugins at session start).
Windows hook compatibility: the refresh-marketplace hook command is
wrapped in `bash -c "..."` because Claude Code on Windows runs hook
commands directly without invoking a shell, so `2>/dev/null || true`
would otherwise be passed as literal argv tokens.
Cross-cutting:
- cli/lib/marketplace.py: shared CLONE_DIR + MARKETPLACE_NAME constants.
- cli/lib/hooks.py: SessionStart now has two independent entries
(pull + refresh-marketplace) so a failure in one doesn't suppress
the other; legacy `da sync` and prior single-pull layouts upgrade
cleanly on re-init.
- PAT injection on every git fetch via per-invocation credential
helper (token in \$AGNES_TOKEN env, never in argv or .git/config).
- Pre-snapshot of installed plugins captured BEFORE
`claude plugin marketplace update` so silent auto-applied version
bumps still fire notifications.
- scripts/dev/agnes-client-reset.sh: cleans ~/.claude/plugins/marketplaces/agnes,
~/.claude/plugins/cache/agnes, drops uv build cache, documents
workspace-scoped residue that can't be enumerated from the script.
- app/web/setup_instructions.py: legacy AGNES_DEBUG_AUTH path also
uses clone (direct HTTPS marketplace add is broken end-to-end on
every Claude Code distribution -- stores response as single file,
plugin source paths then 404).
28 new tests (test_cli_refresh_marketplace.py) + extended hook + setup
template tests cover bootstrap, fetch+reset ordering, version-aware
reconcile, project-path filtering, hook JSON shape, and the bash-c
Windows wrapper invariant.
Adds X-Agnes-Latest-Version and X-Agnes-Min-Version headers to every
/api/* response. CLI consumes these to hard-stop on incompatible drift.
MIN_COMPAT_CLI_VERSION ships at 0.0.0 — no enforcement until a deliberate
wire-protocol break bumps it.
Also dedupes app version logic: app/main.py:_app_version() helper deleted,
replaced by app/version.py:APP_VERSION as the single source of truth.
test_app_version.py rewritten to target app.version.
`_rewrite_user_sql_for_bigquery_query` does its own bare-name detection
(mirroring the non-RBAC parts of `_bq_guardrail_inputs`). The backtick
masking from #201 was applied to `_bq_guardrail_inputs` and the
forbidden-table loop, but missed this third site — so a registered
local-mode table name appearing as the table segment of a
user-supplied full backtick path (e.g. ``\`prj.ds.orders\`` matching
registered local ``orders``) tripped the cross-source guard and
forced every backtick-path query into the 50-100× slower
ATTACH-catalog fallback.
Mask once at the top of the function, route both the BQ-name
detection (line ~830) and the cross-source check (line ~867) through
the masked copy. New regression test
`test_local_name_inside_backtick_path_does_not_trip_cross_source`
proves the wrapper now wraps when it should.
Adds Pass 3 to `_bq_guardrail_inputs` that scans user SQL for full
backtick paths `<project>.<dataset>.<table>` and gates them
identically to the `bq."<dataset>"."<table>"` pass:
- Project must match the configured BigQuery data project
(`get_bq_access().projects.data`). Mismatch → HTTP 403
`bq_path_cross_project`.
- Path must point at a registered row. Unregistered → HTTP 403
`bq_path_not_registered`.
- Non-admin caller must hold a grant on the registered row's id.
Missing grant → HTTP 403 `bq_path_access_denied`.
Pre-fix, full backtick paths bypassed Agnes RBAC entirely — only the
service account scope limited reach. Post-fix the boundary matches
what `agnes catalog`-driven flows already enforce. Admin still
bypasses the per-id grant check but cannot bypass registration or
project match.
Pass 3 also seeds `dry_run_set` for resolved registered paths so the
cost-cap dry-run runs against the same physical table the user named
— composing cleanly with the Layer 2 fail-fast fallback.
When BQ rejects the rewritten dry-run SQL with `bq_bad_request`, the
cap-guard now retries with the user's ORIGINAL SQL instead of building
a synthetic `SELECT * FROM <table>` per registered table. The
synthetic path threw away user filters / projections / partition
predicates and routinely ballooned the estimate to "full table size",
falsely tripping `remote_scan_too_large` on legitimate narrow queries
(typical issue #201 trace: rewriter corrupts a backtick path → BQ
parse error → synthetic over-estimate → 400).
Behaviour:
- Rewritten SQL succeeds: same as before (issue #171 single-dry-run).
- Rewritten SQL parse-errors, original SQL succeeds: use original
estimate. Common case for users submitting BQ-native input.
- Both fail with `bq_bad_request`: HTTP 400 `remote_estimate_failed`
with a hint pointing at `agnes catalog` / BQ-native syntax. No
silent over-estimate.
- Non-parse BQ error (forbidden, upstream): still 502 as before.
This is a behaviour change for clients matching error kinds — failure
to estimate scan size now surfaces as `remote_estimate_failed`
instead of being masked behind `remote_scan_too_large` from the
synthetic path.
Replaces the existing `test_guardrail_falls_back_to_per_table_estimate_on_bq_parse_error`
(which pinned the old contract) with `test_fallback_tries_original_sql_first`
and `test_fallback_fails_fast_on_pure_duckdb_syntax`.
`agnes query --remote` corrupted user SQL when the request contained a
full BigQuery backtick path (`<project>.<dataset>.<table>`) whose
table segment matched a registered bare-name alias. The bare-name
rewriter used `\b` word-boundary matching against the lower-cased SQL;
both `.` and `` ` `` are non-word characters, so the regex fired
INSIDE the user's backtick path and produced malformed nested-backtick
SQL that BigQuery rejected at parse time.
Fix:
- Add `_mask_backticks(sql)` helper: replace each `…` segment with
spaces of equal length, preserving offsets so word-boundary
searches find positions only outside backticks.
- `_bq_guardrail_inputs` (bare-name pass + forbidden-table pass)
searches against the masked SQL.
- `_rewrite_bq_table_refs_to_native` Pass 1 splits the SQL on
`(\`[^\`]*\`)` and rewrites only the outside-backtick chunks. Pass
2 (`bq."ds"."tbl"` → backtick form) is unchanged — its prefix can't
appear inside backticks.
Adds three regressions covering the rewrite + guardrail paths.
In cross-project BQ setups (where billing != data), the SA typically has
serviceusage.services.use on the billing project but not on the data
project. The rewriter passed bq.projects.data as the first arg to
bigquery_query(), which BQ uses as the execution + billing project →
403 USER_PROJECT_DENIED.
Match the convention used everywhere else in the codebase
(app/api/v2_scan.py, app/api/v2_sample.py, app/api/v2_schema.py,
connectors/bigquery/extractor.py): backtick paths inside the inner SQL
use the **data** project (resolves the actual table location), the
bigquery_query() first arg uses the **billing** project (decides who
pays + which project the job runs under). For single-project deploys
the two are identical so the fix is a no-op there.
Test pins the cross-project case: data-prj for backticks, billing-prj
for the bigquery_query() first arg.
R2 adversarial review surfaced 3 issues, all addressed:
#1 cli/client.py:572-577 outer try/except wrapped both _get_shared_client()
AND the actual download. A 401/403/404/5xx from the server triggered a
full second download attempt with a fresh client — wasted bandwidth on
hard failures, no fail-fast on revoked PAT. Narrowed the try to only
the shared-client construction; the download itself is no longer
retried under the fallback except.
#2 concurrent agnes pull invocations (e.g. SessionStart hook + manual
run) collided on bare <target>.tmp / <target>.partN paths — one process's
in-progress write got yanked by the other's cleanup, manifest hash
check then failed spuriously. Per-process suffix (<target>.{pid}.tmp,
<target>.{pid}.partN) makes intermediate files disjoint; the final
os.replace to the bare target is atomic so last-writer-wins.
#3 _looks_like_bq_rewrite_parse_error patterns 'Syntax error' could
false-positive on a query like WHERE log_msg = 'Syntax error in foo'
that fails for an unrelated reason (quota, network) and has the
literal substring echoed in the error text. Anchored to 'Syntax error: '
(with trailing colon) — BQ always emits the colon in this error
format, user SQL string literals normally don't.
R1 adversarial review surfaced 5 issues, all addressed:
#1 chunked download silently disabled in non-Caddy deployments (HEAD on
GET-only FastAPI route returns 405). _probe_range_support now falls back
to GET with Range: bytes=0-0 when HEAD fails — works against both
Caddy file_server (HEAD-friendly) and dev FastAPI direct (GET-only).
#2 parse-error fallback heuristic too broad — matched on Unrecognized
name / Function not found / No matching signature / Invalid cast,
which BQ surfaces for ordinary user-column typos. That triggered slow
ATTACH-catalog retry on every typo (2× latency tax). Narrowed to just
'Syntax error' / 'syntax error' which are the genuine DuckDB-vs-BQ
dialect mismatch markers.
#3 apply_bq_session_settings was only run on fresh-built pool entries,
not on reuse. An operator's /admin/server-config change to bq_query
_timeout_ms wouldn't propagate to long-lived pooled sessions until
restart. Fixed: re-apply on every pool acquire (idempotent + fail-soft).
#4 content-length sanity bound — a misconfigured proxy returning a
wildly inflated Content-Length would cause overlapping chunked Range
requests against the actual file → corrupt assembled output (caught
by manifest hash check, but only after wasted bandwidth). Cap at 100
GiB; above that, drop to single-stream.
#5 rewriter assumed every BQ row resolves under the single
bq.projects.data project. Bucket containing '.' suggests a project-
qualified bucket (multi-project deployment); rewriter would silently
target the wrong project. Conservative skip with regression test.
Address code-reviewer findings on the bigquery_query() rewrite path:
1. Outer LIMIT wrap — bigquery_query() materialises BQ result into DuckDB
before fetchmany sees it (vs ATTACH-catalog Storage Read API streaming).
A user 'SELECT *' against a billion-row remote table would buffer the
entire result before request.limit applied. Wrap rewritten SQL in an
outer 'LIMIT N+1' so the cap pushes into the BQ job itself.
2. Dollar-quoted inner SQL — naive replace("'", "''") doubling missed
DuckDB backslash-escape sequences (\\, \\n, \\t, …). A predicate
like 'WHERE name = ''O\\'Brien''' was unsafe under the doubling
path. DuckDB $bqq_inner$ … $bqq_inner$ form takes the inner SQL
verbatim with no escapes whatsoever. Falls back to legacy doubling
if user SQL improbably contains the literal tag.
3. Parse-error fallback — when the rewritten path fails with a BQ-side
parse / validation error (DuckDB-only syntax like ::INT cast that
survives identifier rewrite but BQ refuses), retry the user's
original SQL via the legacy ATTACH-catalog path so the request still
succeeds. Mirrors the existing dry-run fallback contract.
4. CHANGELOG — delete duplicate CLI bullets that landed under
already-released [0.38.1] (file corruption from merge — entries are
correctly under [0.39.0]).
User SQL hitting query_mode='remote' BigQuery rows was 50-100x slower
than the equivalent direct bigquery_query() call because DuckDB's master
view (CREATE VIEW … AS SELECT * FROM bigquery.<ds>.<tbl>) does not push
WHERE/SELECT/LIMIT into BQ in ATTACH-catalog mode. The BQ extension opens
a Storage Read API session over the entire upstream table; on >100M-row
sources this was 70-150s and frequently failed with 'Response too large
to return'.
Extract the existing dry-run rewriter's core (table-name → BQ-native
backtick path) into a shared helper. Add an execution-path rewriter
that wraps the whole user SQL in bigquery_query('<project>', '<inner>')
so the BQ planner sees the full query and engages partition pruning +
projection pushdown server-side.
Conservative fall-through: cross-source JOINs (BQ ↔ Keboola/Jira local),
queries already containing bigquery_query(, and unconfigured BQ project
all skip the rewrite and run the original SQL via ATTACH-catalog so
behavior degrades gracefully.
1. instance.yaml overlay path now matches read site under STATE_DIR.
Three sites updated:
- app/api/admin.py:1005 (server-config endpoint writer)
- app/api/admin.py:2610 (configure endpoint writer)
- app/instance_config.py:106 (overlay reader)
All three now go through _state_dir() so under flat-mount layout
(STATE_DIR=/data-state) the irreplaceable instance.yaml overlay
lands on the state disk (sdc) instead of the regenerable data
disk (sdb). Without this fix, .env_overlay correctly went to the
state disk while instance.yaml went to the data disk — config
would be lost if an operator wiped sdb.
2. Strip customer-specific tokens from OSS repo per CLAUDE.md
vendor-agnostic rule:
- docker-compose.host-mount.yml: 'a deployer (Groupon FoundryAI)'
→ 'a deployer in production'
- docker-compose.flat-mount.yml: 'caused 2026-05-05 in the
Groupon FoundryAI deployment' → generic 'production failure
mode'
- docs/state-dir.md: rewrote the incident reference to describe
the failure mode abstractly without naming the deployment;
updated the recommendation table to say 'shadow-mount class'
instead of dating the specific incident.
3. Updated docs/state-dir.md 'What reads STATE_DIR' to list all
read/write sites including the three migrated in this round
(admin.py, instance_config.py, marketplaces.py).
ANALYSIS finding (tls-rotate.sh hardcoded host-mount.yml) deferred
— same operator-side class as auto-upgrade.sh hardcoded host-mount,
documented limitation per the PR body.
1. .env_overlay write paths now match read path under STATE_DIR.
app/main.py:343 reads via _state_dir() (post-PR #194), but two
write sites still hardcoded ${DATA_DIR}/state/.env_overlay:
- app/api/admin.py:2687 — configure endpoint secrets persistence
- app/api/marketplaces.py:152 — marketplace PAT persistence
Under flat-mount layout (STATE_DIR=/data-state) the admin UI wrote
secrets to /data/state/.env_overlay while the app read from
/data-state/.env_overlay, silently dropping the value on next
restart. Both write sites now go through _state_dir().
2. host-mount.yml: caddy inherits data:/srv:ro from base, but with
no service populating the data: named volume (other services
switched to direct /data binds), the inherited mount points at an
empty Docker volume — try_files finds nothing, every parquet
download falls through to uvicorn, defeating the v0.36.0
file_server bypass under the host-mount layout. Added a caddy
override that restates all mounts including a direct /data:/srv:ro
bind. Mirrors the comment + treatment already in flat-mount.yml.
Introduces STATE_DIR as the single source of truth for the writable
state directory path, with backward-compatible default of
${DATA_DIR}/state. Pairs with a new docker-compose.flat-mount.yml
overlay that mounts the state disk in PARALLEL to the data disk
(rather than nested under it).
Why
---
The default deployment topology nests state under data: sdb at /data,
sdc at /data/state. That layout has known fragility documented in
docs/state-dir.md — bind-propagation gotchas, two-writer collisions
on the same prefix, mount-order coupling. The 2026-05-05 incident in
the Groupon FoundryAI deployment was a manifestation of the
propagation gotcha.
The flat layout (sdb at /data, sdc at /data-state — parallel, not
nested) eliminates the nested-mount class entirely. Each disk is its
own bind mount, recursive by default in modern Docker. No volume
options to forget. No two-writer collision (host scripts and
container app share /data-state at the same path, single namespace).
What changes
------------
App code (Python):
- src/db.py: new _get_state_dir() helper. get_system_db() and
schema migration snapshot use it.
- app/secrets.py: new _state_dir() helper. _load_or_generate() uses
it for .session_secret and .jwt_secret.
- app/main.py: .env_overlay loaded from _state_dir().
Host scripts:
- scripts/ops/agnes-auto-upgrade.sh: STATE_DIR drives mount-sanity
check and cert detection. Defaults preserve existing behavior.
- scripts/ops/agnes-tls-rotate.sh: STATE_DIR drives CERT_DIR.
New compose overlay:
- docker-compose.flat-mount.yml: parallel /data and /data-state binds
per service. Mutually exclusive with docker-compose.host-mount.yml;
pick one based on disk topology.
Documentation:
- docs/state-dir.md: layout choice (A nested vs B flat), pros/cons,
migration steps, and which code paths read STATE_DIR.
Backward compatibility
----------------------
STATE_DIR defaults to ${DATA_DIR}/state — current behavior. Existing
deployers that don't set the var see no behavior change. Migration
to flat layout is opt-in per the runbook in docs/state-dir.md.
Validation
----------
- bash -n on both host scripts: pass
- docker compose config -f docker-compose.flat-mount.yml: resolves
cleanly with all 6 services binding /data and /data-state directly
- python3 import + helper exercise: STATE_DIR override works,
default falls back to ${DATA_DIR}/state
Companion to PR #191 (drop named-volume driver_opts in host-mount.yml).
That PR fixes the immutability footgun for Layout A; this PR offers
Layout B as the architectural alternative.
🚩 /api/v2/catalog still async def while now calling sync stat()
`/api/v2/catalog` was left as `async def` when the rest of Tier 1 was
converted, on the assumption it was lightweight. The new
`_materialized_size_hint` populator added in this PR calls
`Path.stat()` / `Path.exists()` for every visible row to bucket the
parquet size — on a local FS that's microseconds, but on a
network-mounted DATA_DIR (NFS / CIFS / GCS-FUSE) those syscalls
can block the event loop. Convert to plain `def` so FastAPI
auto-offloads to the thread pool, mirroring /api/query etc.
🔴 stream_download translates HTTPStatusError as generic transport error
`response.raise_for_status()` inside the retry loop raises
`httpx.HTTPStatusError` on 4xx/5xx. After retries exhaust, the new
`isinstance(last_exc, httpx.HTTPError)` check at line 219 was eating
the status code: HTTPStatusError is a subclass of HTTPError, so the
generic transport translation produced "Unexpected error: HTTPStatusError"
instead of the informative "Client error '401 Unauthorized' for url …"
that callers expect. Fix: short-circuit HTTPStatusError before the
HTTPError branch — it re-raises verbatim so the caller's status-code
handling + the rich server error body (e.g. 401 expired token, 403
cross_project_forbidden) reach the analyst.
api_get / api_post / api_delete / api_patch don't have the same bug:
httpx Client.get/etc. don't raise HTTPStatusError unless the caller
explicitly calls .raise_for_status(), and our wrappers don't.
Only stream_download does, hence the targeted fix there.
Five hottest BQ-touching endpoints were `async def` but invoked synchronous
DuckDB / BQ-extension calls inside the body. Under uvicorn's single event
loop that meant a single heavy `agnes query --remote` (waiting up to
~200 s for BQ's jobs.query) froze EVERY other request — /api/health,
dashboard, auth, even another query — for the full BQ wait. Operators
saw "VM idle, app frozen" during PR #188's testing.
Convert to plain `def` so FastAPI auto-offloads the body to the anyio
thread pool. Event loop stays free for non-BQ requests.
- app/api/query.py:execute_query
- app/api/v2_scan.py:scan_estimate_endpoint, scan_endpoint
- app/api/v2_sample.py:sample
- app/api/v2_schema.py:schema
Audit: 0 `await` statements in any converted handler (verified file-by-
file), so the rename is safe. Tests in tests/test_v2_*.py called the
handlers via `asyncio.run(...)` which now fails on a non-coroutine return;
swapped for direct calls (asyncio.run( -> ( ) — keeps paren balance).
Plus AGNES_THREADPOOL_SIZE env var (default 200, was anyio's stock 40)
in app/main.py:lifespan. Set via
anyio.to_thread.current_default_thread_limiter().total_tokens. 200 is
comfortable headroom for <50 concurrent analysts; bump for more.
480/480 impacted tests pass (the 2 remaining errors are a pre-existing
fixture setup issue in test_reader_smoke_matrix.py unrelated to this
change).
Three concrete changes addressing the "analyst Claude misuses the CLI"
class of bugs (image.png table — issues #3, #5, plus the recurrent
"how big is this table" guesswork):
1. config/claude_md_template.txt — the template agnes init writes to
<workspace>/CLAUDE.md. Surfaces every catalog-row field with a why,
adds a query_mode-based decision tree, explicit --estimate scoping
(snapshot create ONLY — was the #1 first-try error), an agnes fetch
→ agnes snapshot create rename note, and a 6-row failure-mode table
that maps each common error wording to its right next step.
2. app/api/v2_catalog.py — populate rough_size_hint for local +
materialized rows from the on-disk parquet size, bucketed
small/medium/large/very_large. Was hardcoded null with a TODO; AI
couldn't tell "is this 6.8 GB" without a failed --remote round-trip.
3. cli/update_check.py — the [update] banner survived the da→agnes
rename and printed "[update] da X is out of date" on every command,
training analysts to associate the binary with the old name.
Verified by rendering the template against representative contexts
(33/33 tests pass) and running every use case from the original
screenshot through the real CLI against a dev VM.
A single analyst's multi-GB `agnes pull` held the only uvicorn worker
for the duration of the stream, starving UI / /api/health / every other
API endpoint. Container flipped to `unhealthy`. Triggered while a
6.8 GB `order_economics` pull was in-flight on prod 2026-05-05.
Caddy now intercepts `GET /api/data/{table_id}/download` and serves
the parquet directly via sendfile from the data volume (mounted r-o
at /srv inside the caddy container). RBAC enforced by `forward_auth`
to a new lightweight `GET /api/data/{table_id}/check-access` endpoint
(returns 204 / 403) — the bulk transfer never reaches uvicorn.
Path discovery via `try_files` over the known extract.duckdb v2 source
subdirs. Anything not at a static path falls through to the existing
app handler so legacy `src_data/parquet` and future connectors still
work without a Caddyfile change. Non-Caddy deployments are unchanged.
Stage 1 (multi-worker uvicorn) was considered but blocked by the
single-writer DuckDB lock on system.duckdb — workers > 1 would crash
at startup on "Could not set lock on file", the same race that pushed
the scheduler from in-process writes to HTTP-via-app. Multi-reader
workers + single-writer coordination is out of scope for this PR.
DuckDB BigQuery extension defaults `bq_query_timeout_ms` to 90 s, which
is too tight for analyst-scale queries against view-backed BQ datasets.
`agnes query --remote` HTTP 400'd with `Binder Error: Query execution
exceeded the timeout. Job ID: ...` whenever the underlying BQ job ran
longer than 90 s, even though the job itself was healthy.
Add `data_source.bigquery.query_timeout_ms` (default 600 000 ms = 10 min,
sentinel 0 falls through to the extension default). Applied via
`SET bq_query_timeout_ms` after every `LOAD bigquery` on every BQ-touching
DuckDB session: orchestrator's `_remote_attach` ATTACH path, BqAccess
session factory, and the standalone extractor. Configurable via
`/admin/server-config` UI.
Fail-soft: extension versions that don't recognise the setting silently
keep the default rather than poisoning the session.
Backup-orchestration commands were split across two namespaces (pull in
agnes store, push in agnes admin store), which broke the operator
mental model — pull/push are a paired operation and should sit
together.
Move pull + info into agnes admin store so all bulk operations share
one help screen. Add agnes store mine as the user-facing equivalent —
calls the same /api/store/bundle.zip endpoint with ?owner=me, which
the server resolves to the caller's user_id. Authors can archive
their own uploads without admin role; whole-Store bulk reads stay
admin-flavored as a discoverability hint.
Server: 3-line addition to export_bundle handles owner='me' as a
magic alias for the caller. No new endpoint.
Tests updated: pull/info expectations move from agnes store to
agnes admin store; new tests cover agnes store mine and the
?owner=me server resolution. 69/69 store tests green locally.
Adds whole-Store backup/restore primitives so an external CI/CD job can
mirror the Store to a git repo (and restore back from one).
REST:
- GET /api/store/bundle.zip — deterministic ZIP of all (filtered) Store
entities. Layout: manifest.json + entities/<id>/{plugin,assets}/.
Manifest carries owner_email for cross-instance restore. Auth: any
authenticated user (Store is community-open).
- POST /api/store/import-bundle — admin-only restore. Modes
merge|replace|skip; owner resolution by email with stub-disabled-user
fallback when the email is unknown on the target instance.
CLI:
- agnes store update <id> [--description X] [--zip PATH] ... — in-place
edit (server PUT permits owner OR admin per F4). Closes the missing
edit affordance for analysts who want to fix a typo or push a new
ZIP without losing install_count.
- agnes store pull [-o store.zip] [--unpack DIR] — download the bundle.
--unpack streams + extracts so an external git-backup workflow can
drop the tree straight into a repo and `git add .`.
- agnes store info [--json] — counts + size summary.
- agnes admin store push <zip-or-dir> [--mode ...] — admin-only restore.
Auto-zips a directory client-side so a working-tree → server
round-trip is one command.
cli/v2_client.py gains api_get_stream helper for binary downloads.
Tests: 5 new server tests (bundle shape + filters + round-trip + stub
user creation + skip mode + admin-only gate) + 11 new CLI tests
(update, pull/unpack, info, admin push). 66/66 store-related tests
green locally.
The previous gather used `sorted(glob, key=lambda p: p.stat().st_mtime)`.
A transient OSError (race with delete, permission flicker, EBADF on a
weird filesystem) on any single file raised through the lambda and 500-ed
the whole page.
Reworked: stat each path under try/except into a (path, stat) list, sort
the already-statted entries. Bad files drop silently from the listing.
Regression test test_profile_sessions_page_tolerates_stat_failures
patches Path.stat to raise on one of two files, asserts the page returns
200 with the good row rendered and the bad row dropped.
Devin flagged that run_session_collector still had the same audit-skip
gap I fixed in run_verification_detector and run_corporate_memory in
the previous two rounds — a PermissionError walking /home, an OSError
on /data/user_sessions mkdir, or any other unhandled exception from
collector.run() would skip the audit_log row and only show in docker
logs.
Same try/except + unhandled_error pattern as the sibling endpoints.
All three LLM-pipeline run-* endpoints now record their failures the
same way; /admin/scheduler-runs sees them. Regression test in
tests/test_admin_run_endpoints.py::TestRunSessionCollector::test_unhandled_exception_still_audits.
User feedback during e2e of #179: the listing page is nice but I want
to grab the raw jsonl and look at what's inside.
Adds GET /profile/sessions/<filename>:
- Auth via get_current_user (owner-only).
- Path safety: rejects "/", "\", "..", leading ".", and any non-".jsonl"
filename. The served path resolves under
${DATA_DIR}/user_sessions/<caller.id>/; if resolution escapes that
base directory, returns 404 (never 403, so existence of other users'
files isn't leaked).
- FileResponse with Content-Disposition: attachment.
UI: Download button per row in profile_sessions.html.
Tests in test_web_ui.py: path-traversal / nested / dotfile / non-jsonl
all 404 for owner; unauthenticated 302/401/403; authenticated owner
gets 200 + correct Content-Disposition.
Devin flagged that run_corporate_memory still had the same audit-skip
gap I just fixed in run_verification_detector — if collect_all() throws
anything other than the already-translated ValueError (DuckDB lock,
network blip, unexpected SDK error), the audit_log row was never
written and /admin/scheduler-runs missed the failure.
Same try/except + unhandled_error pattern as the verification_detector
fix from 4c4dfee8. Regression test in
tests/test_admin_run_endpoints.py::TestRunCorporateMemory::test_unhandled_exception_still_audits.
Three changes addressing user feedback during e2e test of #179 + Devin Review on e86dd5ed.
1) /profile/sessions — new self-service user page in the user menu.
Lists all session jsonls the caller uploaded via `agnes push` joined
against session_extraction_state. Each row shows uploaded_at, file
size, status badge (pending/processed/extracted), processed_at, and
items_extracted. The page docstring + help text explicitly call out
that items_extracted=0 means the verification detector ran fine but
the LLM found no claims to track — that's the documented "no items"
outcome, not a broken pipeline. Closes the gap surfaced during the
e2e test of #176 where a user could see their sessions on disk and
process them through the LLM but had no UI to inspect what happened.
2) run_verification_detector audits unhandled exceptions (Devin #1).
If detector.run() threw anything other than the already-translated
ValueError, the audit_log row was never written. The endpoint now
wraps detector.run in try/except, records the exception in
audit_params["unhandled_error"], then re-raises as 500 after audit.
The /admin/scheduler-runs page surfaces the failure row with the
error type + message.
3) SCHEDULER_AUDIT_ACTIONS list corrected (Devin #2). Previous list
had "marketplaces_sync_all" (wrong — actual is "marketplace.sync_all")
plus "data_refresh" and "scripts_run_due" which app/api/sync.py and
app/api/scripts.py don't write to audit_log. Fixed to the four
actually-logged strings; comment points at the missing audit calls
as a follow-up.
Tests: tests/test_web_ui.py adds TestAdminRoleGuards::test_profile_sessions_page_no_admin_required and tightens test_admin_scheduler_runs_page_admin_only to assert the correct marketplace.sync_all string.
create_entity + update_entity created the `scratch` temp dir inside one
try/finally but cleaned it up in a separate one. Validation HTTPExceptions
raised by _safe_zip_extract (zip_unsafe_path, zip_too_large_uncompressed)
or the BadZipFile→422 conversion exited the first scope, and the second
finally was never entered → temp dir leaked on every failed upload.
Devin flagged this on the F2 commit. The leak pre-existed (zip_unsafe_path
was the original vector); F2 added zip_too_large_uncompressed to the same
broken cleanup path. Fixed by collapsing scratch creation + cleanup into
one outer try/finally that covers both extraction AND metadata/bake; the
inner try/except/finally still handles BadZipFile→422 + tmp file cleanup.
Same restructure in update_entity. Regression test
`test_scratch_dir_cleaned_up_after_failed_extraction` triggers a
zip_unsafe_path 422 and asserts tmp/agnes_store_* contains no leaked
dirs.
Three independent reviews of PR #180 surfaced four real defects in the new
Store / my-ai-stack surface. CHANGELOG entries detail each; one-liners:
- F1 video_url XSS: any authenticated user could upload a Store entity
with `video_url=javascript:...` and pop XSS in any viewer's session via
the `<a href=...>` "Watch video" link in store_detail.html. Jinja2
autoescape doesn't block URI schemes inside attribute values. Fixed by
scheme-validating to http(s) only on create + update; 400 invalid_video_url.
- F2 ZIP decompression bomb: _safe_zip_extract checked path-traversal but
not declared file_size totals — a 50 MB compressed upload at 1:1000
ratio decompresses to 50 GB and DOS the host disk. Fixed by summing
zinfo.file_size across infolist() and refusing > 200 MB before
extractall touches disk. 413 zip_too_large_uncompressed.
- F4 admin authz parity: PUT /api/store/entities/{id} was owner-only while
DELETE allowed owner OR admin; the store-detail page hid Edit/Delete
buttons from admin even though DELETE was permitted. Fixed by allowing
admin on PUT and passing is_admin to the template; gate is now
is_owner OR is_admin everywhere.
- F5 cross-owner suffix collision: sanitize_username is many-to-one
(alice.smith / alice_smith both → alice-smith). Two such users uploading
entities with the same display name produced identical
`<name>-by-<username>` suffixes, silently colliding in the served
agnes-store-bundle on-disk paths AND the manifest catalog (Claude Code
dedupes by plugin.json `name`). Fixed by enforcing global uniqueness on
the suffixed value at create_entity; 409 conflict_global_suffix.
F3 (ZIP symlink members) was investigated and confirmed to be a
false-positive — Python's stdlib ZipFile.extractall does not honor
symlink mode bits, so no exploit exists.
9 new regression tests in tests/test_store_api.py::TestStoreSecurityFixes
covering all four. Test run locally: 60/60 store-related tests pass.
E2E test on a real BQ deploy showed every verification-extraction call
fails with HTTP 400 invalid_request_error: "output_config.format.schema:
For 'object' type, 'additionalProperties' must be explicitly set to false".
The Anthropic structured-output API now requires the field on every object
node in the json_schema. Fix: connectors/llm/anthropic_provider.py wraps
the caller-supplied schema through a recursive _strict_json_schema()
walker that adds the field where missing (preserving any explicit
override), then passes the strict variant to the API. Six unit tests in
TestStrictJsonSchema pin the recursion across nested objects, array items,
and the no-mutation invariant.
Adds /admin/scheduler-runs — a read-only admin page that surfaces the
last 200 audit-log entries from scheduler-driven actions. New
AuditRepository.query_actions(actions, limit) helper, new admin nav
entry. Failed scheduler ticks (HTTP 401, network errors) don't reach
the audit_log; the page calls that out with a hint to set
SCHEDULER_API_TOKEN if no rows show up.
run_session_collector called collector.main() which did argparse.parse_args()
on uvicorn's sys.argv (['app.main:app', '--host', ...]) → sys.exit(2) →
SystemExit(2), which inherits from BaseException, escapes FastAPI handlers,
and propagates through the thread pool. Every scheduler tick that fired the
endpoint either 500-ed or risked killing the uvicorn worker.
services/session_collector/collector.py now exposes run(dry_run, verbose)
that returns (rc, stats); main() is a thin CLI shim that parses argv and
delegates. The admin endpoint calls run() directly and audit-logs the
per-run stats (users_processed, files_copied, files_skipped) instead of
just the rc. Three regression tests in TestRunHelper.
Closes Devin Review finding on app/api/admin.py:2819 (#179).
Devin BUG: /api/admin/configure seeds an ai: block to the writable
overlay at DATA_DIR/state/instance.yaml, but the three LLM consumers
imported from config.loader.load_instance_config — which reads the
static config dir only. Even if they had read the overlay, the loader
ran yaml.safe_load directly without passing through _resolve_env_refs,
so '${ANTHROPIC_API_KEY}' would have stayed a literal placeholder. The
pipeline appeared to work because the factory falls back to the env
var directly, but the overlay path itself was dead code.
Two fixes, both required:
1. Switched the three LLM consumers to app.instance_config.load_instance_config:
- services/corporate_memory/collector.py:collect_all
- services/verification_detector/__main__.py:main
- app/api/admin.py:run_verification_detector
2. app/instance_config.py runs the loaded overlay through
config.loader._resolve_env_refs *before* the deep-merge, so
'${ANTHROPIC_API_KEY}' resolves at config-load time.
New regression suite tests/test_instance_config_overlay.py pins:
- env-ref resolution against the overlay (resolved when env set,
empty when env missing — never the literal placeholder)
- deep-merge still preserves static-only sections
- the three consumers reach app.instance_config (inspected via
inspect.getsource so a future refactor that reverts the import
fails the test)
- end-to-end: a seeded overlay + ANTHROPIC_API_KEY env reaches the
factory with a resolved api_key
The PR rewrote collect_all() to call the new
create_extractor_from_env_or_config() helper, but the existing tests
still mocked the old direct create_extractor() symbol and the old
silent-skip-on-missing-config behavior. Five tests in
TestCorporateMemoryCollector and one in TestCollectorExtractorIntegration
were red on the PR branch.
Changes:
- Tests now mock connectors.llm.create_extractor_from_env_or_config
(the symbol the collector imports lazily).
- Renamed test_collect_all_no_ai_config_skips ->
test_collect_all_no_ai_config_or_env_raises and
test_collector_handles_invalid_config -> test_collector_raises_on_invalid_config.
Both assert pytest.raises(ValueError) — the explicit fail-fast
semantics defect 5 of #176 was supposed to enforce.
- collect_all() no longer swallows the factory's ValueError into
stats["errors"]; it propagates so the scheduler / admin endpoint
surface the actionable misconfiguration message instead of
pretending the run was a no-op.
- /api/admin/run-corporate-memory translates the propagated ValueError
into a 500 with the factory's message, matching
/api/admin/run-verification-detector.
`_read_agnes_ca_pem()` decides whether the served fullchain.pem needs
trust-bootstrapping in the rendered setup prompt. Pre-fix it only
checked the leaf's *immediate* issuer against `certifi`'s trust store.
For Let's Encrypt that's the intermediate (R13), which `certifi` does
not ship — only roots are in trust stores. So a publicly-trusted LE
chain still tripped the "needs bootstrap" path and the setup prompt
emitted a step-0 TLS trust block + clone-fallback marketplace block
that no client actually needs (Bun-compiled `claude.exe`, system git,
Python via certifi all validate the chain through the bundled ISRG
Root X1).
Now we walk every cert in the fullchain (leaf + intermediates) and
return None the first time any cert's issuer is in the certifi trust
store — that captures the standard "leaf signed by intermediate signed
by publicly-trusted root" shape. Trusted subjects are read once into
a set for O(1) lookup. Self-signed (leaf.issuer == leaf.subject) and
private-CA chains (no chain link's issuer in certifi) keep their
previous "return PEM" behavior, so deployments that genuinely need
the bootstrap still get it.
Validated end-to-end against the live VM at
agnes-marustamyan.groupondev.com (LE R13 → ISRG Root X1):
- Let's Encrypt fullchain → has_ca=False (was True)
- Self-signed cert → has_ca=True
- Corporate-CA chain (private root) → has_ca=True
- Missing fullchain.pem → has_ca=False
Adds a community-driven Store where any authenticated user uploads
skills/agents/plugins as ZIPs, plus /my-ai-stack as the per-user
composition view. The served Claude Code marketplace is now:
(admin_granted ∖ opt_outs) ∪ store_installs
Skill + agent installs are merged into a single `agnes-store-bundle`
plugin in the served marketplace; type=plugin uploads stay standalone.
Names are suffixed with `-by-<owner-username>` at upload time so two
owners can use the same display name without colliding in Claude Code's
flat skill/agent namespace.
Schema v23 → v24 adds three tables:
- store_entities — community-uploaded skills/agents/plugins
- user_store_installs — what each user has chosen to install
- user_plugin_optouts — opt-out overlay on top of admin grants
Admin grant-delete drops every user's opt-out for that plugin so
re-grant resets cleanly to enabled (no sticky personal preference).
UI:
- /store — e-commerce-style listing with type/category/owner
filters, search, pagination, owner-aware [Install]
buttons, clickable cards
- /store/new — 2-step upload wizard with drag & drop, preview
validation (POST /api/store/entities/preview), docs
multi-upload, photo + video URL
- /store/{id} — detail page with hero, file list, docs, owner
actions (Edit/Delete) for the uploader
- /my-ai-stack — Granted plugins (toggle opt-out) + From the Store
(uninstall) sections
- Admin nav: Marketplaces moved into Admin dropdown, renamed to
"Curated Marketplaces"
Validation hardening: type-mismatch guards reject skill ZIP uploaded as
agent (or vice versa), and plugin ZIPs masquerading as skills/agents.
Human-readable error messages mapped client-side from machine codes.
Cross-source naming: Store entity-id-prefixed dirs (`plugins/store-<id>/`)
plus the bundle (`plugins/store-bundle/`) avoid collisions with admin
marketplaces (whose `store` slug is reserved by `is_valid_slug`).
Bundle composition is content-hashed at serve time — install/uninstall
or owner re-upload bumps the bundle's plugin.json `version`, so Claude
Code's auto-update toggle picks up changes.
Tests: 50+ new tests across naming, repositories, filter (admin ∪ store
∪ bundle), API (upload/install/uninstall/delete/preview/docs), end-to-end
marketplace.zip with bundle merging.
GET /api/health/detailed now returns a session_pipeline service entry.
Heuristic:
max(mtime of /data/user_sessions/**/*.jsonl) <=
max(processed_at in session_extraction_state) + grace_seconds
grace_seconds = 2 × verification-detector cadence (default 30 min;
configurable via SCHEDULER_VERIFICATION_DETECTOR_INTERVAL).
When the assert fails, status='warning' (never 'error') with an
actionable detail pointing at the verification-detector scheduler job.
A warning bubbles up to the existing overall='degraded' aggregation —
operators querying /api/health/detailed (or /agnes diagnose system)
get a clear breadcrumb instead of a silently-broken pipeline.
Cold-start case (no session files, or files newer than the grace
window with empty state table) is handled explicitly to avoid noise
on a fresh deploy.
Tests: tests/test_health_session_pipeline.py.
The /corporate-memory page filters status IN ('approved','mandatory')
and showed no hint that pending items exist. With approval_mode set to
'review_queue' (the default in instance.yaml.example), every collection
run would silently funnel new items into the pending bucket where no
operator ever saw them.
For admins (is_km_admin), the page now renders a banner above the
stats bar:
N pending items awaiting review — review them at /corporate-memory/admin
Non-admins see no change (the route zeroes the count server-side
before passing to the template, so the hint is never leaked).
Tests: tests/test_corporate_memory_page.py.
The session-collector, verification-detector, and corporate-memory
services now run on the same scheduler-v2 model that already drives
data-refresh, health-check, script-runner, and marketplaces:
- New admin endpoints in app/api/admin.py:
POST /api/admin/run-session-collector
POST /api/admin/run-verification-detector
POST /api/admin/run-corporate-memory
All admin-gated, sync-def (FastAPI thread pool), with one audit row
per invocation. Same single-writer-of-system.duckdb pattern as the
existing /api/marketplaces/sync-all job.
- services/scheduler/__main__.py JOBS gains three entries with offset
cadences (10m / 15m / 17m, all coprime modulo the 30s tick) so the
three LLM-backed jobs don't fire on the same tick and stack their
API + DB load.
- The verification-detector endpoint surfaces the LLM factory's
fail-fast ValueError as HTTP 500 with the actionable message,
preserving the no-silent-skip contract from the previous commit.
Tests:
- tests/test_admin_run_endpoints.py covers admin gating + scheduler
registration + endpoint contract.
- tests/test_scheduler_sidecar.py existing tests continue to pass.
POST /api/admin/configure now writes a default ai: block into the
instance.yaml overlay when the request leaves it untouched and either
ANTHROPIC_API_KEY or LLM_API_KEY is set in the environment. The block
references the env var via ${VAR} syntax — secrets never land in YAML.
connectors.llm.factory grows create_extractor_from_env_or_config which
falls back to ANTHROPIC_API_KEY / LLM_API_KEY when ai_config is empty
and raises a clear ValueError when neither is available. Both
services/corporate_memory and services/verification_detector switch to
the new helper, replacing the old 'silently skip when ai: missing'
path that was the silent-failure root cause.
Tests:
- tests/test_setup_ai_block.py — overlay seeding contract.
- tests/test_llm_provider_env_fallback.py — fallback + fail-fast.
The iterative bare-name rewriter (one re.sub per name, longest-first)
was vulnerable to cross-contamination when the GCP project ID contained
a registered table name as a hyphen-delimited word.
Concrete repro:
project = 'my-ue-project'
registered = ['orders', 'ue']
user SQL = 'SELECT * FROM orders JOIN ue ON ...'
iter 1 (orders): produces 'FROM `my-ue-project.fin.orders` JOIN ue ...'
iter 2 (ue): '\bue\b' matches 'ue' INSIDE 'my-ue-project' (hyphen
creates word boundary on both sides) — corrupts
the iter-1 path
Fallback at query.py:576 caught the resulting BQ parse error and fell
back to per-table SELECT * estimate, so impact was over-estimation,
not fail-open — but the #171 partition-pruning fix silently degraded
to pre-fix behavior whenever a project name shared a hyphen-segment
with a registered table.
Fix: single re.sub call with an alternation regex sorted longest-first.
Single-pass means each source position is processed exactly once, so
freshly-inserted backticked text from one match isn't re-scanned by
later names in the alternation.
Regression test
test_rewrite_helper_does_not_corrupt_when_project_id_contains_registered_name
covers the exact Devin repro.
Drops the `<nav class="role-tiles">` block (Analyst / Admin tiles),
the `_show_admin_tile` flag, the `const ROLE = {{ role | tojson }};`
JS line, and the role-aware PAT-mint ternary. The setupNewClaude
button now mints a uniform PAT for everyone:
{ name: defaultTokenName(), expires_in_days: 90 }
…against the existing `POST /auth/tokens` endpoint. No new endpoint,
no role-locked TTL clamp. The `bootstrap-analyst` 1-hour scope is no
longer used from /setup (it broke the install flow anyway — saved PATs
expired before the user opened Claude Code; tracked as a separate
cleanup issue).
Also removes the now-unused `.role-tiles` / `.role-tile` CSS rules so
the stylesheet doesn't carry dead selectors.
Plan: docs/superpowers/plans/2026-05-04-unified-setup-prompt.md task 6.
The `/setup` route no longer accepts `?role=analyst|admin`. The route
signature drops the `Literal[...] = Query(...)` parameter and the
silent admin-downgrade block (`if role == "admin" and not is_admin:
role = "analyst"`). The `role` ctx variable threaded into install.html
also goes away — Task 6 cleans up the template's role-tile UI and the
JS PAT-mint ternary.
`?role=` is silently ignored by FastAPI for unknown query params, so
existing bookmarks (none in production — the param was added in this
PR and never shipped) just degrade to the unified layout. No
RedirectResponse shim needed.
Tests: drop the entire `tests/test_setup_page_roles.py` file (eight
role-branching tests that no longer apply) and add
`tests/test_setup_page_unified.py` with three tests:
- `test_setup_page_renders_unified_layout`
- `test_setup_page_ignores_role_query_param`
- `test_setup_page_renders_marketplace_for_user_with_grants`
- `test_install_legacy_path_redirects_to_setup`
Also replace the role-aware `test_install_preview_*` tests in
test_web_ui.py with unified-layout assertions.
Plan: docs/superpowers/plans/2026-05-04-unified-setup-prompt.md task 5.
Removes the `role: Literal["analyst", "admin"] = "admin"` parameter from
`compute_default_agent_prompt`. The same RBAC pass
(`marketplace_filter.resolve_allowed_plugins`) now runs for every user —
admin or not. Users with no `resource_grants` rows get the
no-marketplace layout; users with grants get the marketplace block
inserted. Admin-vs-analyst is no longer a layout branch.
`render_agent_prompt_banner` no longer derives a `role` from
`user.is_admin`; it just delegates to `compute_default_agent_prompt`.
Two `compute_default_agent_prompt(...role=role)` call sites in
`app/web/router.py::setup_page` are updated to drop the keyword so the
route keeps rendering — Task 5 will remove the `?role=` query
parameter and the silent admin-downgrade block from the route signature
itself.
Tests: drop role-aware assertions from test_welcome_template_renderer
and test_welcome_template_api. Both files now assert the unified
default contains `agnes init` + `uv tool install` and bans the legacy
`agnes auth import-token` / `agnes auth whoami` verbs.
Plan: docs/superpowers/plans/2026-05-04-unified-setup-prompt.md task 4.
Renames `_git_check_block` to `_preflight_block` and adds a
`claude --version` check beside `git --version`. Both binaries are
required by the marketplace step — git for the clone fallback,
claude for `claude plugin marketplace add` / `claude plugin install` —
so checking them together gives one clear failure instead of two
confusing downstream errors.
Install hints: `npm i -g @anthropic-ai/claude-code` for Linux / WSL
plus a doc URL (https://docs.claude.com/claude-code) for the native
macOS / Windows installers. We don't try to one-line a native
installer; the canonical instructions live upstream.
Plan: docs/superpowers/plans/2026-05-04-unified-setup-prompt.md task 3.
Adds `_step_numbers(*, has_marketplace, has_skills)` so step numbering
lives in one place instead of being split across three branches in
`resolve_lines`. Pins the unified layout in the tests:
No plugins: 1 install, 2 init, 3 catalog, 4 diagnose, 5 skills, 6 confirm
With plugins: 1, 2, 3, 4 preflight, 5 marketplace, 6 diagnose, 7 skills, 8 confirm
`agnes auth import-token` / `agnes auth whoami` are now banned from the
rendered prompt — `agnes init` subsumes them. The renamed
`test_resolve_lines_no_plugins_unified_six_step_layout` asserts those
strings are absent and that the new step headers (`Bootstrap your Agnes
workspace`, `Verify the data is queryable`) are present.
Plan: docs/superpowers/plans/2026-05-04-unified-setup-prompt.md task 2.
Removes the `role: Literal["analyst", "admin"]` parameter from
`resolve_lines` / `render_setup_instructions` and deletes the
`_resolve_analyst_lines`, `_analyst_init_lines`, `_analyst_finale_lines`
helpers. The unified flow now always emits `agnes init` (the
workspace-rails delivery mechanism) in place of the legacy
`agnes auth import-token` + `agnes auth whoami` pair, and uses
`agnes catalog` as the smoke-verify step.
`agnes init` already verifies the PAT internally, and `agnes catalog`
doubles as a data-plane smoke check, so dropping `agnes auth whoami`
costs no signal.
Drops the now-redundant `tests/test_setup_instructions_analyst.py` and
patches the one ordering test in `tests/test_setup_instructions.py` that
referenced the old "Log in" / "Verify the login" headers. Also strips
the `role=role` kwarg from `compute_default_agent_prompt`'s call into
`resolve_lines` so the welcome-template render path keeps working;
welcome_template.py's own role param is removed in a follow-up task.
Plan: docs/superpowers/plans/2026-05-04-unified-setup-prompt.md task 1.
Bring admin UI, audit-log messages, code comments, and analyst-facing
skill docs in line with the post-bootstrap CLI surface (`agnes pull`,
`agnes push`, `agnes init`, `agnes snapshot create`). The legacy
`_LEGACY_STRINGS` detection tuple in `app/api/claude_md.py` and the hook
upgrade markers in `cli/lib/hooks.py` are intentionally left as-is —
they exist precisely to flag pre-rewrite content for re-authoring.
Strip "(folded from `da metrics list`)" / "(lifted from `da metrics
show`)" / "Replaces the old `da analyst status`" docstring noise — the
rename history is in CHANGELOG.md, not in module docstrings.
Closes#171. The /api/query cost guardrail used to dry-run a synthetic
`SELECT * FROM <table>` for each registered remote-BQ row referenced
by the user SQL — which made BigQuery estimate a full table scan, with
column projection, predicate pushdown, and partition pruning all
disabled. Narrow queries on big partitioned/clustered tables (the
documented happy path for `agnes query --remote`) hit ~30,000×
over-estimates and got rejected with 400 `remote_scan_too_large` even
when BQ's own dry-run reported single-digit MB.
Pavel's report on #171 traced the root cause and proposed the fix:
rewrite the user SQL to BQ-native syntax and dry-run it as a single
job, exactly the way `bq query --dry_run` works.
Implementation:
- New helper _rewrite_user_sql_for_bq_dry_run rewrites bare registered
names (word-boundary, case-insensitive, longest-first to avoid prefix
collisions) + bq."<ds>"."<tbl>" forms to backticked
`<project>.<ds>.<tbl>` paths.
- _bq_quota_and_cap_guard runs ONE dry-run on the rewritten SQL. Cap
check uses the real estimate.
- Fallback path: if BQ rejects with bq_bad_request (e.g. DuckDB-only
syntax like ::INT casts), the guard falls back to the pre-fix
per-table SELECT * approach so non-portable queries still get a
(loose) cap estimate instead of fail-opening. Non-parse BQ errors
(forbidden, upstream) still propagate as 502.
- _bq_guardrail_inputs now also returns name_lookups so the rewriter
has the (registered_name, bucket, source_table) mapping it needs.
- Per-table breakdown is unavailable from a composite dry-run; total
bytes are pinned to dry_run_set[0] for the post-flight
record_bytes(sum(...)) call to keep returning the right total.
Tests (7 new, 3 existing still pass):
- dry-run receives rewritten user SQL with WHERE clause intact (the
load-bearing assertion for #171)
- single dry-run per request even with multiple registered tables
(JOIN, UNION) referenced
- fallback to per-table SELECT * on bq_bad_request
- non-parse BQ errors (forbidden) still 502
- rewriter unit tests: bare + bq.path in same SQL, longest-name-wins
on prefix collision, case-insensitive bare-name match
Three coupled UX fixes for the analyst-onboarding flow:
1. Dashboard "Setup a new Claude Code" CTA was rendering admin paste
prompt for everyone (analysts couldn't actually execute the marketplace
plugin install / skills setup steps). render_agent_prompt_banner now
picks role based on user.is_admin — analysts get the analyst flow.
2. /setup default role changed from admin to analyst. Most visitors are
analysts; admin layout is opt-in via the admin tile or ?role=admin.
3. Admin tile is admin-only on the role-tile nav. Non-admins see only
the analyst tile. Server-side: non-admin requesting ?role=admin is
silently downgraded to analyst (otherwise they'd see admin paste
prompt despite no tile).
Tests:
- New: test_setup_page_admin_tile_hidden_for_non_admin (anonymous client
can't see "Admin CLI" or role=admin link)
- New: test_setup_page_admin_role_downgraded_for_non_admin (anonymous
?role=admin → analyst layout, no marketplace step in clipboard)
- New: test_install_preview_default_role_is_analyst (admin signing in to
bare /setup gets analyst clipboard by default)
- Renamed: test_setup_page_default_role_is_admin → ..._is_analyst
- Updated: test_setup_page_admin_clipboard_renders_admin_layout uses
FastAPI dependency_overrides to inject admin user (admin layout is
now admin-gated)
- Updated: test_install_preview_visible_for_signed_in_user explicitly
passes ?role=admin to exercise admin layout
13 Devin findings across 10 files:
🔴 Critical:
- app/api/v2_catalog.py:42 — `_fetch_hint` returns `da fetch` in /api/v2/catalog
responses (user-visible in every catalog list)
- cli/skills/agnes-data-querying.md — 11 stale `da fetch`/`da sync` refs in the
bundled skill markdown
- config/claude_md_template.txt:38 — referenced `agnes pull --docs-only` flag
that does NOT exist in agnes pull (removed; spec only ships --quiet/--json/
--dry-run)
🟡 Important:
- app/api/admin.py:252 — `da fetch` in bq_max_scan_bytes hint
- cli/commands/auth.py:119 — `da sync` in import-token docstring (--help text)
- cli/commands/tokens.py:48 — "Export it so `da` can use it" prose
- ARCHITECTURE.md — 4 stale rows in CLI commands table
- README.md — stale paragraphs for analysts (da sync, da analyst setup)
🚩 Substantive observations addressed:
- app/api/query.py:249,302,489 — server-side error/help strings still said
`da sync`/`da fetch` (returned in API responses to clients)
- cli/commands/snapshot.py:235-241 — DuckDB existence guard incorrectly
blocked `--estimate` (server-side dry-run that never opens local DB).
Added test ensuring estimate path skips the guard.
Skipped (intentionally historical):
- app/api/admin.py:2377,2429,2437 — historical comments describing past
manifest-vs-sync_state bug; past tense, accurate to keep as `da sync`.
New top-level 'materialize' section, single field (lock_ttl_seconds).
Default 86400 (24h). Backs the file-lock TTL reclaim added in the
per-table-mutex change. Editable via PUT /api/admin/server-config and
the /admin/server-config UI.
When admin registers a materialized BQ row with bucket+source_table but
no source_query, the server generates 'SELECT * FROM `<project>.<ds>.<tbl>`'
from instance.yaml's configured BQ project. Same fallback fires on PUT
when flipping to materialized. The backtick rejection guard, which was
appropriate for DuckDB-flavor source_query, is relaxed for materialized
rows since the new wrapping path (Task 2) runs admin SQL through BQ
jobs API which uses BQ-native syntax (backticks for dashed identifiers).
_run_materialized_pass distinguishes due-check skips from in-flight
skips and never calls state.set_error for either. summary['skipped']
becomes a list of {table, reason} dicts; the end-of-pass log line
breaks out the in_flight subcount.
Hoists is_table_due to module-level import so test monkeypatching of
the symbol intercepts the call (the previous local import made
patches a no-op).
Two Task 4 review fixes for app/web/templates/install.html:
1. JSON-escape `ROLE` JS const via `{{ role | tojson }}` (defense in
depth — removes the dependency on Jinja autoescape semantics for JS
contexts; FastAPI's Literal validator already constrains role values).
2. Verify the analyst tile's clipboard payload is the analyst layout.
The pre-existing role-aware plumbing (compute_default_agent_prompt
threading role into setup_instructions_lines, picked up by the JS
SETUP_INSTRUCTIONS_TEMPLATE array) was correct; adding regression tests
that pin to the JS clipboard block specifically so a future inversion
would fail loudly.
Tests: analyst clipboard contains `agnes init` + `agnes catalog` and
NOT `agnes auth import-token` / `agnes skills`; admin clipboard is the
inverse. Plus an explicit assertion that ROLE is rendered via tojson.
Devin Review iter #5 flagged a pre-existing class of name/id mismatch
in app/api/query.py:131-136 — the SAME root cause as the bq.* RBAC
issue I fixed in iter #3 (line 332/362). Devin called it out as
"NOT introduced by this PR" / "might merit follow-up", but it's
exactly the same security-boundary pattern this PR is hardening, so
fixing here keeps the RBAC story consistent across the handler.
The `forbidden = all_views - set(allowed)` comparison mixed types:
- `all_views` carries DuckDB master view names (= registry display
`name` from the orchestrator's CREATE VIEW)
- `set(allowed)` carries registry IDs (resource_grants.resource_id)
When `id != name` (e.g. id="bq.finance.ue", name="ue"), authorized
users got spurious 403s — the view name landed in `forbidden` even
though the caller had a valid grant on the registry id.
Build a name->id map from the registry, then the forbidden check
compares apples to apples:
allowed_view_names = {r["name"] for r in registry_rows
if r.get("name") and r.get("id") in allowed_ids}
forbidden = all_views - allowed_view_names
107 affected tests pass; 487 pass in wider RBAC/query/access/admin
domain — no regressions.
Devin Review iter #3 found 3 new real bugs after iter #2's fixes landed.
🔴 RBAC check at app/api/query.py:362 used `row["name"]` against
`accessible_set`, but `accessible_set` is keyed by registry IDs
(`get_accessible_tables` returns `resource_grants.resource_id` —
table IDs, not display names). Confirmed by `_table_blocks` projection
at `app/resource_types.py:157-158`. When `id != name` (e.g.
`id="bq.finance.ue", name="ue"`), non-admin users with valid grants
got 403 `bq_path_access_denied`. Switch to `row["id"]`.
🚩 Bare-name pass at app/api/query.py:332 had the same name-vs-id
mismatch (different impact): legitimate accessible rows were skipped
from `dry_run_set`, so the cost guardrail under-counted scan bytes
for non-admin users. Could let an over-cap query through and
under-bill quota. Switch to `row_id` comparison.
🟡 `placeholder_from` for billing_project was dead code.
`_BQ_OPTIONAL_FIELD_DEFAULTS["billing_project"] = ""` seeded an empty
string into every GET payload via `_ensure_bq_optional_fields`. JS
`isUnset = (value === undefined)` evaluated False, so the
`(defaults to <project>)` placeholder NEVER rendered. Drop the seed —
field stays in `known_fields` (UI sees it) but routes through the
unset rendering path on GET, where placeholder_from fires.
Tests: test_get_surfaces_bq_fields_even_when_unset assertion flipped
from "billing_project IS present" to "billing_project NOT auto-seeded"
to lock in the new shape. 67 affected tests pass.
Devin Review iter #2 found 2 new issues (after iter #1's 5 fixes
landed). Both real, both addressed.
🔴 Quota user_id key mismatch defeated shared daily budget. /api/query
computed `user.get("id") or user.get("email")` while /api/v2/scan uses
`user.get("email") or "anon"` (app/api/v2_scan.py:327). Same user → two
different keys in the singleton QuotaTracker. BQ bytes consumed via
/api/query were tracked under UUID; via /api/v2/scan under email; the
`check_daily_budget` pre-flight on either endpoint never saw the
other's recorded bytes — per-user cap was effectively doubled. Match
v2/scan's email-first ordering.
🟡 QuotaExceededError(KIND_CONCURRENT) → 400 instead of 429.
`quota.acquire(user_id)` raises this from __enter__ when the per-user
concurrent-scan slot is at cap. The exception propagated through the
@contextlib.contextmanager generator, the caller's `with guard:`
block, and was caught by execute_query's generic `except Exception`
handler → mapped to 400 with a flattened "Query error: concurrent_scans:
N/M" string, dropping the typed retry_after_seconds field. Wrap the
`with quota.acquire(...)` in a try/except QuotaExceededError that maps
to 429 with the same typed-detail shape used for the daily-budget
rejection — consistent with /api/v2/scan:392-402.
Tests: test_api_query_quota.py user_id strings updated to
"admin@test.com" (the seeded_app admin's email) to match the new
email-first ordering. 40 affected tests pass.
Devin Review on PR #168 found 5 issues — all real, all addressed.
🚩 ANALYSIS_001 (architectural): concurrent-slot guard didn't protect
actual BQ query execution. Earlier `_enforce_remote_bq_quota_and_cap`
ran dry-run + cap check inside `with quota.acquire(user_id):`, then
returned — releasing the slot BEFORE `analytics.execute(...)` ran. Spec
§4.3.3 explicitly designs the slot to wrap execute so the per-user
concurrent cap limits BQ scans, not just dry-runs.
Refactor to a context manager `_bq_quota_and_cap_guard`. Caller's `with`
block now holds the slot through dry-run, cap check, the actual
`analytics.execute(...)` (which is what triggers the BQ scan when DuckDB
resolves the master view), AND the post-flight record_bytes. Slot
released only when caller's `with` body exits.
🟡 BUG_001: placeholder JS walked `original` (full GET payload root)
instead of `original.sections`. `placeholder_from: ["data_source",
"bigquery", "project"]` is a section-relative path, so billing_project
placeholder NEVER rendered. Fix: walk `original.sections` (with fallback
to `original` for safety).
🟡 BUG_002 + BUG_003: admin_tables.html register and edit modals'
operator help text referenced `max_bytes_per_remote_query` (the old
name from the spec) but the actual config key is `bq_max_scan_bytes`
after the fix-up commit `6423888d` moved it. Replace both occurrences.
🟡 BUG_004: CHANGELOG entry said `api.query.bq_max_scan_bytes` (the
old path) but the read at app/api/query.py:53 is
`get_value("data_source", "bigquery", "bq_max_scan_bytes", ...)`. An
operator who set it under `api.query` in their yaml would have no
effect. Correct path in CHANGELOG.
All 95 #160-affected tests pass after the changes.
E2E test on dev VM revealed: spec said "configurable via /admin/server-config"
for the cost guardrail cap, but the underlying read path was
`api.query.bq_max_scan_bytes` and `api` is NOT in `_EDITABLE_SECTIONS`. POST
to /admin/server-config rejected `{"sections":{"api":...}}` as "unknown
section(s): api" — the cap was only adjustable via direct YAML edit.
Move to `data_source.bigquery.bq_max_scan_bytes`:
- `_default_remote_query_cap_bytes()` reads from the new path.
- Add to `_OPTIONAL_FIELDS["data_source"]["bigquery"]["fields"]` with the
same shape as `max_bytes_per_materialize` (kind=int, default 5 GiB, hint).
- Add to `_BQ_OPTIONAL_FIELD_DEFAULTS` so it surfaces in the GET payload
even when YAML omits it.
Convention now mirrors `max_bytes_per_materialize` — both BQ cost
guardrails live under `data_source.bigquery`, both editable in the UI.
Closes the operator-side half of the reporter's loop. The CLI fix in
the previous commit makes USER_PROJECT_DENIED errors readable to
analysts; this commit lets admins verify reachability proactively
from /admin/server-config without waiting for analyst reports.
New endpoint POST /api/admin/bigquery/test-connection
(app/api/admin_bigquery_test.py, ~110 LOC):
- Depends(require_admin); registered in app/main.py.
- Builds BqAccess via existing get_bq_access(), runs `SELECT 1 AS ok`
with a 10s polling timeout.
- 200 with {ok, billing_project, data_project, elapsed_ms} on success.
- 400 for `BqAccessError(not_configured)` (operator config issue).
- 502 for any other typed BqAccessError or unknown upstream exception.
- 504 for concurrent.futures.TimeoutError; best-effort cancel_job
invoked (BQ-side cancel may still run; documented caveat).
Server-config placeholder (app/api/admin.py + admin_server_config.html):
- `data_source.bigquery.billing_project` field-spec gains
`placeholder_from: ["data_source", "bigquery", "project"]`.
- renderLeafInput's text branch reads `opts.spec.placeholder_from`,
walks the loaded `original` config dict, injects
`placeholder="(defaults to <project>)"` into the input HTML at
construction time. Admin sees the access.py:339-340 fallback rule
visible directly in the UI without reading source.
UI button:
- "Test BigQuery connection" button next to data_source's Save button.
- onTestBigQuery() POSTs to the endpoint, renders structured result
inline (green check + elapsed_ms on success; red kind + hint on
failure).
Tests: 6 endpoint cases + 1 placeholder payload test = 7 GREEN. 62
total across the affected admin server-config test files.
Phase 3 review identified an RBAC + cost-cap bypass: `SELECT * FROM
"bq"."ds"."tbl"` (catalog token quoted as a DuckDB identifier) was NOT
matched by the BQ_PATH regex, so direct quoted-form references skipped
both the registry check and the cost-cap dry-run. DuckDB resolves
`"bq"` to the same ATTACHed BQ catalog, so the bypass is real.
Widen the catalog-token alternation: `(?:"bq"|bq)` matches both forms.
Negative lookbehind `(?<![\w.])` still rejects look-alike prefixes
(`other_bq`, `my_bq`); the new "my_bq".ds.tbl negative test locks that
in alongside `other_bq.ds.tbl`.
Tests:
- 2 new positive cases in tests/test_query_bq_regex.py for the quoted
form (`"bq"."finance"."ue"` and uppercase `"BQ"."ds"."tbl"`).
- 1 new negative case rejecting `"my_bq".ds.tbl` so the quoted-form
widening doesn't open a different evasion.
- 1 new RBAC test in tests/test_api_query_rbac_bq_path.py: admin
hitting an unregistered quoted path returns the same
bq_path_not_registered 403 as the unquoted form.
All 33 Phase 3 tests pass after the fix.
The headline implementation for issue #160. POST /api/query now gates
direct `bq."<dataset>"."<source_table>"` references behind the registry
and bounds the BQ scan cost behind a configurable cap. Wired through
the same singleton QuotaTracker as /api/v2/scan so daily-byte budgets
are shared across both BQ-touching paths.
Changes in app/api/query.py:
- Add module-level `BQ_PATH` regex matching the 16 syntax variants
verified empirically (fully-quoted, unquoted, mixed quoting,
case-insensitive, inside CTE bodies, multi-path, …).
- Add `bigquery_query` to the SQL keyword blocklist. Closes the
pre-existing function-call backdoor where a user could run an
arbitrary BQ jobs API call against any reachable dataset, bypassing
the registry and RBAC. Wrap views internal to the BQ extractor still
use bigquery_query() — but those run via DuckDB view resolution at
query time, not via user-submitted SQL, so the blocklist doesn't
break them.
- Add `_bq_guardrail_inputs` helper: walks user SQL twice — once for
bare-name matches against accessible registered remote-BQ names
(contributes to dry_run_set), once for direct `bq.X.Y` matches
(gated against `find_by_bq_path` lookups, returns 403 with
structured detail on miss or grant violation).
- Add `_enforce_remote_bq_quota_and_cap` helper: pre-flight
`check_daily_budget` (over-cap → 429), then `with quota.acquire(...)`
wraps a per-path BQ dry-run, sums bytes, raises 400
`remote_scan_too_large` when total > cap.
- Cap default 5 GiB; configurable via `api.query.bq_max_scan_bytes`
in /admin/server-config (next phase wires the UI).
- Post-flight `record_bytes` against the user's daily counter.
- Module-level imports of `_bq_dry_run_bytes`, `_build_quota_tracker`,
`get_bq_access` so tests can monkeypatch via `app.api.query.<name>`.
Tests:
- All 23 RED tests from the previous commit now pass (regex matrix,
blocklist with detail-string assertion, RBAC unregistered/admin-bypass,
guardrail dry-run-called/over-cap-rejected, quota pre-flight 429).
- mock_dry_run fixture stubs both `_bq_dry_run_bytes` and `get_bq_access`
so guardrail tests don't require a live BQ project.
- Quota test uses `admin1` (the seeded_app fixture's actual user id, not
`admin`).
Smoke: 887 passed across query/bq/admin/extractor/registry/quota
domains. No regressions.
The /api/query cost guardrail (next phase) needs the same singleton
QuotaTracker so its daily-byte and concurrent-slot caps accumulate
across both /api/v2/scan and /api/query BQ-touching paths.
Move `_build_quota_tracker`, `_quota_singleton`, and `_quota_init_lock`
from app/api/v2_scan.py to app/api/v2_quota.py (the natural home; the
factory uses QuotaTracker which already lives there). Re-export the
function from v2_scan.py so the 7 test sites at tests/test_v2_scan.py
(lines 77, 118, 143, 160, 186, 208, 250) keep working without edits.
Crucially do NOT re-export `_quota_singleton` from v2_scan.py — Python
`from X import var` copies the binding at import time, so a re-exported
singleton would freeze at the initial None and never observe the
in-place mutation done inside `_build_quota_tracker()`. Re-export only
the function (which always reads the live module-global through `global`).
Mechanical refactor; no behavior change. 30 quota-related tests pass.
Now that VIEW/MATERIALIZED_VIEW always wrap via bigquery_query() (the
prior `legacy_wrap_views=True` branch behavior, made unconditional in
the previous commit), the toggle has no semantic meaning and is removed
across the codebase.
Production code:
- app/api/admin.py: drop the field from _OPTIONAL_FIELDS["data_source"]
["bigquery"]["fields"] and from _BQ_OPTIONAL_FIELD_DEFAULTS, plus the
comment block above the defaults dict.
- config/instance.yaml.example: drop the example snippet.
- src/orchestrator.py: update the inner-objects skip-branch comment to
reflect the new BQ behavior (the skip itself stays — keboola
use_extension=False still inserts _meta rows without inner views).
- app/web/templates/admin_tables.html: rewrite operator copy in the
register and edit forms to reflect always-wrap.
Tests:
- tests/test_admin_server_config.py (TestServerConfigBigQueryFields):
flip assertions from "field IS present" to "field NOT present" on
legacy_wrap_views. Drop the test_post_persists_legacy_wrap_views test
since the field no longer exists.
- tests/test_admin_server_config_known_fields.py: same flip on the
known-fields registry assertion.
- tests/test_bigquery_extractor.py: drop the obsolete
test_view_entity_does_not_create_master_view_by_default (asserted the
bug we fixed) and test_legacy_wrap_views_toggle_restores_old_behavior
(toggle no longer meaningful). Update remaining test docstrings.
Operators with `legacy_wrap_views: true` set in their overlay get the
new (equivalent) behavior automatically — the unrecognized key is
silently ignored by the YAML loader. Operators with `false` get the
issue-#160 fix as a behavior change, not a regression.
Spec gate updated: production code grep gate
grep -rn 'legacy_wrap_views' connectors app src config cli
must return zero. tests/ excluded — historical "removed in #160"
breadcrumbs and `assert "X" not in fields` regression guards retained
as anti-regression signals.
Finding #1: _build_context now routes through render_agent_prompt_banner when
a DB connection is available, so both /setup and the /dashboard clipboard CTA
always reflect the admin override (or the live default when no override is set).
Previously _build_context unconditionally used resolve_lines(), ignoring the
welcome_template override for the dashboard JS array.
Finding #2: PUT /api/admin/welcome-template now performs a second render pass
with user=None (anonymous stub) after the authenticated-user pass. Templates
that reference user.* fields without an {% if user %} guard are rejected with
a clear 400 error explaining the anon-visitor breakage.
- Fix#1: _detect_existing_project now checks .claude/settings.json for
"da sync" marker instead of deleted CLAUDE.md; update tests accordingly.
- Fix#2: preview endpoint uses autoescape=False to match /setup rendering;
align render_agent_prompt_banner in welcome_template.py to the same.
- Fix#3: apply _sanitize_banner_html to override render path in setup_page
so all render paths sanitize consistently.
- Fix#4: move .setup-link-banner into the existing-user branch where
account_details.last_sync_display is reachable; remove dead copy from
new-user branch.
The /admin/agent-prompt editor now pre-fills with the full bash bootstrap
script from setup_instructions.resolve_lines() instead of being empty.
When an admin saves an override it replaces the default everywhere — the
/setup page display and the dashboard clipboard CTA — rather than adding a
banner above the auto-generated commands.
GET /api/admin/welcome-template now returns a `default` field with the live
computed script so the editor always shows meaningful starting content.
{server_url} and {token} single-brace placeholders survive Jinja2 rendering
and are substituted by JavaScript at clipboard-copy time as before.
Preview pane switches to textContent (not innerHTML) since content is bash.
- admin_welcome.html: update subtitle, description, placeholder cheatsheet
(drop tables/metrics/marketplaces/sync_interval; add user-null note and
security note). Textarea initial value is now empty (no default template
to show). Preview pane uses innerHTML (HTML output). refreshStatus sets
editor to empty when no override. Preview pane styled as light surface.
Reset modal copy updated (no banner shown, not "OSS-shipped template").
- config/claude_md_template.txt: deleted (markdown template is gone;
default is now no banner).
- docs/agent-setup-prompt.md: rewritten for variant C — describes the
/setup banner, smaller placeholder table, security/sanitization notes,
anonymous-user guard, example HTML snippet.
- src/welcome_template.py: rewrite as HTML banner renderer
(render_agent_prompt_banner); drop _list_tables, _metrics_summary,
_marketplaces_for_user, render_welcome, _load_default_template.
build_context now exposes only instance/server/user/now/today.
_sanitize_banner_html strips script/iframe/on*/javascript: post-render.
- app/api/welcome.py: drop get_welcome handler, WelcomeResponse, old
_VALIDATION_STUB_CONTEXT. Admin endpoints stay at same URLs; validation
stub updated to match new slim context. Preview now uses autoescape=True.
- app/web/router.py: setup_page calls render_agent_prompt_banner and passes
banner_html to install.html; admin_agent_prompt_page drops _load_default_template.
- app/web/templates/install.html: add .setup-banner CSS + banner block above hero.
- cli/commands/analyst.py: replace _generate_claude_md with _init_claude_workspace;
no CLAUDE.md written, only .claude/CLAUDE.local.md placeholder + settings.json hooks.
- tests: delete test_cli_analyst_welcome.py (tests deleted endpoint/function);
rewrite TestGenerateClaudeMd → TestInitClaudeWorkspace; update api test to
assert /api/welcome returns 404 and remove welcome-fetch tests.
Rename the welcome prompt editor from /admin/welcome to /admin/agent-prompt
and update all UI labels to "Agent Setup Prompt". API endpoint URLs are
unchanged (PUT/GET/DELETE /api/admin/welcome-template, GET /api/welcome).
- Nav menu: "Welcome prompt" → "Agent Setup Prompt", href updated
- Page title and h2 updated in admin_welcome.html
- Error message hint in app/api/welcome.py updated to /admin/agent-prompt
- Dashboard: replace inline <details> preview of _claude_setup_instructions
with a simple link to /setup (Task C)
- docs/welcome-template.md renamed to docs/agent-setup-prompt.md; internal
references to /admin/welcome updated
- OpenAPI snapshot path updated
- Tests updated to reflect new route and removed inline preview
Remove the setup_banner feature (admin-editable /setup page banner) and
all associated code: API router, repository, renderer, admin template,
tests, and docs. The setup_page handler no longer calls render_setup_banner;
the install.html template no longer renders banner_html. The setup_banner
DuckDB table (v22) is kept intact for forward-compat with already-migrated
instances — only the application code is removed.
CHANGELOG updated: setup_banner bullets removed; Agent Setup Prompt
(welcome-template feature) now stands alone as the single editable prompt.
- Add integrity= + crossorigin= to all 4 cdnjs tags in admin_welcome.html
and admin_setup_banner.html (I-1)
- Add graceful CDN fallback: when CodeMirror is undefined (SRI mismatch or
CDN down), degrade to styled plain textarea with polyfill editor interface
so save/reset/preview still work (I-1)
- Replace fixed 480px editor height with calc(100vh - 320px) for
viewport-relative sizing; add min-height: 480px to .welcome-editor-col (M-8)
- Change /install redirect from 301 to 302 to prevent indefinite browser
caching (I-5)
- Sanitize Jinja2 error detail in /api/welcome 500 response: log full error
server-side, return generic detail pointing at /admin/welcome (M-7)
- Hoist build_context import to module level in app/api/welcome.py (M-11)
Adds an optional Jinja2/HTML banner displayed above the bootstrap
commands on /setup. Empty by default; admin authors it at
/admin/setup-banner. autoescape=True — safe for HTML context.
Render failures return "" so a broken banner never breaks /setup.
Schema v22: setup_banner singleton table, auto-migration v21→v22.
- Add GET /setup serving install.html (CLI + Claude Code setup page)
- Add GET /install → 301 redirect to /setup for backwards compat
- Move first-time setup wizard from /setup to /first-time-setup
- Update nav link: href=/setup, label 'Setup local agent', active on both /setup and /install paths
- Update page <title> to 'Setup local agent — …'
- Update /dashboard and /setup comment in _claude_setup_instructions.jinja
- Update tests and OpenAPI snapshot accordingly
* security(auth): per-IP rate limit on auth endpoints + generalize last-admin guard
Closes#45 and #151.
#45 — every auth endpoint was unthrottled (login, magic-link, token,
bootstrap), leaving us open to password brute-force and SMTP
email-bombing. Wires slowapi (new dep) into the middleware chain with
per-route limits: 10/min on login + token, 5/min on send-link, 3/min on
bootstrap. Returns 429 with Retry-After: 60 once exceeded. Per-IP key
respects the leftmost X-Forwarded-For hop (Caddy in front of the app
strips client-supplied XFF). Operator escape hatch:
AGNES_AUTH_RATELIMIT_ENABLED=0. Test suite disables the limiter via
autouse conftest fixture so existing auth tests that hammer endpoints
in tight loops are unaffected.
#151 — DELETE /api/admin/users/{id}/memberships/{group_id} and the
mirror DELETE /api/admin/groups/{group_id}/members/{user_id} only
guarded against self-removal as last admin. Generalizes to refuse
removing anyone from the seeded Admin group when they are the only
remaining active admin (mirrors the existing
count_admins(active_only=True) <= 1 check on delete_user / update_user).
Recovery from zero admins requires direct DB access, so this closes
a path where a scheduler/bootstrap actor that bypasses normal admin
checks could otherwise empty the group.
* security(auth): throttle remaining email-bombing + token-confirm endpoints
Address code-review gap on PR #165 — the first commit covered /send-link
but missed two endpoints with the IDENTICAL email-bombing surface:
- POST /auth/password/reset — sends reset mail, anti-enum response
- POST /auth/password/setup/request — sends setup mail, anti-enum response
Both now share the 5/min limit with /send-link.
Also add 10/min to the token-confirm surfaces — high-entropy tokens but
partial leaks via logs / referer have surfaced before, and unbounded
guess rate would let an attacker exhaust the keyspace adjacent to a
leaked prefix:
- POST /auth/email/verify
- GET /auth/email/verify — closes the click-through bypass
- POST /auth/password/reset/confirm
- POST /auth/password/setup/confirm
Doc fix: rate_limit.py module docstring + CHANGELOG entry no longer
claim "disable without a redeploy" (misleading). The Limiter constructor
freezes `enabled` from env at import time, matching every other Agnes
env knob — operators set the flag and bounce the container.
Tests: 4 new cases in test_auth_rate_limit.py covering
/reset, /setup/request, /reset/confirm, GET /verify. Full suite:
2583 passed, 32 skipped, 0 failed.
* security(auth): throttle JSON /auth/password/setup — closes form-throttle bypass
Second code-review pass on PR #165 caught a fifth gap: POST /auth/password/setup
(JSON variant, kept for backward compat) consumes the same setup_token as
the web form /setup/confirm but was unthrottled — an attacker brute-forcing
the token just switches from the form path to the JSON path and resumes
at unbounded RPS. Apply the same 10/min limit and signature shape used
on /setup/confirm.
Also extend CHANGELOG note about the JSON-variant bypass for future
operators reading the security entry.
Test: 1 new case (test_password_setup_json_rate_limited_after_10_requests),
9 rate-limit tests + 28 password-flow tests + 41 auth-provider tests pass,
no regressions.
* chore(release): cut 0.30.1 — auth security hardening (rate limit + last-admin guard)
E2E sub-agent finding: `da query --remote "SELECT * FROM <id>"` against a
materialized table that hasn't yet been rebuilt in the server's
analytics.duckdb returns a confusing DuckDB "Table does not exist"
message even though the table is in the registry. Materialized rows
produce parquets at `${DATA_DIR}/extracts/<source>/data/<id>.parquet`,
but the orchestrator's master-view creation is `_meta`-driven — fresh
instances or pre-tick states have the registry row without a
corresponding view, so analysts hit the bare "does not exist" with no
path forward.
Improve the error rendering in `app/api/query.py:execute_query`. When
DuckDB raises a "table does not exist" error, scan the registry for any
`query_mode='materialized'` row whose id or name appears in the failed
SQL. On a hit, return a 400 whose detail names the table, explains the
materialize state, and offers two concrete next steps:
1. Run `da sync` (or wait for the scheduler tick / hit
POST /api/sync/trigger) to materialize the parquet, OR
2. Query the source directly via the catalog alias when the registry row
carries bucket+source_table (e.g. `bq."dataset"."table"` for BigQuery,
`kbc."bucket"."table"` for Keboola).
Detection is bounded — the registry round-trip only fires when DuckDB's
error mentions a missing table, so happy-path queries pay no cost.
Non-materialized unknowns fall through to DuckDB's raw error.
2 new tests: materialized id surfaces the hint with the bucket+source_table
payload; unknown table falls back to the generic error path with no false
positive on the new hint.
The strict source_type-availability validator from the prior commit
broke ~12 existing tests that register tables on the default test
instance (where `data_source.type` resolves to 'local' since no
instance.yaml is loaded).
The intent of the validator is to catch *explicit* misconfig:
`type=bigquery` instance + `source_type=keboola` payload with no
`data_source.keboola.*` block. The bootstrap workflow — admin sets up
a fresh instance and registers a few tables before pointing at a real
source — should not be gated here.
Loosen the check: when `get_data_source_type()` returns 'local' (the
fallback when no `data_source.type` is set), skip the rejection. The
explicit mismatch case still 422s because that path resolves
`configured_primary` to a real source type.
Also adds an autouse keboola_instance fixture to test_journey_sync_query.py
which exercises Keboola registrations through the full sync→query
flow — the fixture documents the test's data-source assumption rather
than relying on the bootstrap escape hatch.
E2E sub-agent finding: instance configured with `data_source.type='bigquery'`
and no `data_source.keboola.*` block. Admin POSTs `{source_type: 'keboola'}`
to /api/admin/register-table → returns 201, row lands in the registry, but
never syncs because the scheduler has no Keboola URL/token to ATTACH
against. Operator only notices the gap when `da catalog` keeps showing
nothing.
The new `_validate_source_type_configured` helper runs immediately after
the id/view-name collision checks in `register_table`. A source_type is
considered configured when:
- it matches `get_data_source_type()` (the instance's primary), OR
- a non-empty `data_source.<source_type>` block exists in the effective
`instance.yaml` (multi-source instance), OR
- it's in `_SOURCE_TYPES_INDEPENDENT_OF_DATA_SOURCE` (Jira / local — both
get data through paths that don't involve `data_source.*`).
Returns 422 with a message that names the configured primary source and
points at `/admin/server-config` for enabling a secondary one. None /
empty source_type is still tolerated for backward compat with legacy CLI
scripts that don't set the field — the route resolves it later.
5 new tests cover: keboola-on-bq rejected, bq-on-keboola rejected,
matching source_type still works, jira allowed regardless, omitted
source_type passes through.
Existing tests that registered Keboola rows on the unconfigured default
test instance now opt into a `keboola_instance` fixture to satisfy the
new validator (tests/test_admin_bq_register.py + .keboola_materialized
+ .unregister_cleanup; the multi-source PUT test in test_admin_bq_register
adds a `keboola` block to its synthetic config).
Pre-existing test_missing_project_returns_error failure in
TestRebuildFromRegistry is unrelated (config-cache leakage from a
previous test in the same class) — confirmed pre-existing on the prior
commit via `git stash` reproduction.
E2E sub-agent finding: register a materialized BQ row → sync to materialize
the parquet at `/data/extracts/bigquery/data/<id>.parquet` → DELETE the
registry row. The DB row goes away but:
- the parquet file stays on disk forever, AND
- the sync_state row stays, so `/api/sync/manifest` keeps advertising the
dropped table to `da sync`, AND
- the orchestrator's next rebuild can resurrect a master view by picking
up the leftover parquet.
Two-part fix in `unregister_table`:
1. For materialized rows on bigquery/keboola, remove
`${DATA_DIR}/extracts/<source_type>/data/<name>.parquet` (and any stale
`<name>.parquet.tmp` from a crashed prior materialize). Filename is
keyed on `table_registry.name` to match sync_state bookkeeping.
File-removal errors are logged but don't fail the DELETE — the registry
row is already gone, and an orphan parquet won't get a master view at
next rebuild because the orchestrator's _meta-driven scan never picks
up bare parquet files.
2. Always clear `sync_state` + `sync_history` rows for the dropped table_id
so the manifest stops advertising the table — applies to all source
types and modes, not just materialized, since any synced row had a
sync_state entry.
Orchestrator-side defensive guard (Finding 2b) is a no-op in the current
implementation: `_attach_and_create_views` only creates master views from
`_meta` rows in each connector's `extract.duckdb`, so a parquet without a
matching `_meta` entry is already invisible to the rebuild. The new
test `test_orchestrator_skips_orphan_parquet_in_extracts` is kept as a
regression guard for that contract.
5 tests cover: BQ + Keboola materialized DELETE removes parquet, remote
DELETE doesn't error trying to remove a non-existent file, sync_state
cleared on DELETE, orchestrator orphan-skip invariant.
E2E testing showed admin POSTs of materialized BQ rows whose source_query
uses BigQuery-native backtick identifiers (`prj.ds.t`) silently no-op'd at
the next sync tick — the materialize path runs the SQL through the DuckDB
BQ extension's COPY which uses DuckDB's parser; backticks aren't recognized
and the query either parse-errors or matches zero rows. No parquet lands at
the canonical path and no error reaches an operator-visible surface.
Two-part fix:
1. RegisterTableRequest's _check_mode_query_coherence model_validator now
rejects any source_query containing a backtick with a 422 + actionable
message pointing at the DuckDB equivalent (bq."dataset"."table"). Same
check is applied in update_table on the merged record so PATCHes that
flip a stored source_query to backtick form are also caught. Covers BQ
AND Keboola materialized rows since both connectors funnel source_query
through DuckDB's COPY.
2. _run_materialized_pass now persists per-row failures via the new
SyncStateRepository.set_error / clear_error methods (existing
sync_state.error / status columns — no schema migration). GET
/api/admin/registry enriches each row with `last_sync_error` from a
single batched SELECT against sync_state, so the admin UI / da admin
status can show "this table failed last sync because: X" instead of
operators having to trawl scheduler logs. Recovered rows have the
error cleared automatically — update_sync's success path resets
status='ok' / error=NULL on the upsert.
The materialized-path test fixture's _materialized_payload helper is
updated to use DuckDB-flavor SQL (the prior backtick example pre-dated the
fix). 6 new tests cover register/update rejection on BQ + Keboola, the
sync_state error persistence, and the registry response surface.
Devin's second review pass on commit 16938ae7 surfaced 2 more issues:
BUG_pr-review-job-58ae3148_0001 — non-BQ materialized via PUT bypasses source_query check
app/api/admin.py update_table only enforces 'query_mode=materialized
requires source_query' for source_type='bigquery' rows (via the
synthetic RegisterTableRequest at line 2129+). Non-BQ source types
(Keboola) skip the check — admin could PUT {query_mode: materialized}
on a Keboola local row without source_query, persist successfully,
then crash at the next sync tick when kb_materialize_query received
sql=None and DuckDB rejected COPY (None) TO '...'.
Fix: generic coherence guard before the BQ-specific block — for ALL
source types, query_mode='materialized' requires non-empty source_query
in the merged record. Returns 422 with a hint about reverting via
query_mode='local'/'remote'.
ANALYSIS_pr-review-job-642ff90f_0007 — diagnose returns 'ok' on BQ resolution failure
app/api/health.py:_check_bq_billing_project caught get_bq_access()
exceptions and returned status='ok' with a 'could not resolve' detail.
Automated alerting keyed on status != 'ok' would silently miss missing
google-cloud-bigquery, auth failures, or malformed config. Fix: return
status='unknown' on resolution failure — surfaces it on operator
dashboards without promoting the overall health to 'degraded' (which
'warning' does, intentionally for the billing==project case).
Tests:
- test_update_keboola_to_materialized_without_source_query_rejected:
PUT {query_mode: materialized} on a Keboola local row returns 422
with 'source_query' in the detail
- test_diagnose_returns_unknown_status_when_bq_resolution_fails:
when get_bq_access raises, the bq_config service entry surfaces
status='unknown' (not 'ok')
Full sweep: 2507 passed, 25 skipped, 0 failed (+2 from previous sweep
because of the 2 new regression tests; 8 pre-existing internal_roles
schema-migration failures still ignored per task brief).
Devin Review on commit 7052a235 flagged 4 real bugs in the Keboola
materialized path. All four are fixed; 3 new regression tests pin the
behavior so future refactors can't quietly regress.
BUG_pr-review-job-3fbd31c9_0001 — _run_materialized_pass gated behind 'if bq_project:'
app/api/sync.py:444-466 wrapped the entire materialized pass (which
dispatches BOTH BigQuery AND Keboola rows by source_type) in a check
for data_source.bigquery.project being non-empty. On Keboola-only
instances this short-circuited and Keboola materialized rows sat in
table_registry forever without their SQL being evaluated — the feature
CHANGELOG advertised was dead code on the most common deployment shape.
Fix: always run the materialized pass; the BQ branch's per-row try/except
catches the typed BqAccessError(not_configured) the sentinel raises
when no BQ project is set, so non-BQ instances incur a per-row error
for any (hypothetical) BQ-tagged row but the Keboola path runs cleanly.
Log line renamed 'Materialized BQ' → 'Materialized SQL' to match.
BUG_pr-review-job-3fbd31c9_0004 — wrong config key 'url' instead of 'stack_url'
app/api/sync.py:149 read get_value('data_source', 'keboola', 'url'),
but the canonical config key documented in instance.yaml.example:111
and used by app/api/admin.py:1503 + 2359 is 'stack_url'. Production
Keboola instances would always see an empty URL and fail with the
'not configured' error. The pre-existing test patched the wrong key
too, so it passed without catching the mismatch. Fix: use stack_url
in both sync.py and the test fixture.
BUG_pr-review-job-3fbd31c9_0003 — no atomic write in Keboola materialize_query
connectors/keboola/extractor.py wrote COPY directly to the final
'<id>.parquet' path. A mid-COPY failure (network, disk full, extension
crash) left a partial parquet that the orchestrator rebuild would
later pick up and serve to analysts. BQ's materialize_query already
uses a '<id>.parquet.tmp' staging path + os.replace() atomic swap
(connectors/bigquery/extractor.py:370-445); Keboola now mirrors that
pattern with the same try/except cleanup on COPY failure.
BUG_pr-review-job-3fbd31c9_0002 — full file read into memory for MD5
Same file:60-62 used parquet_path.read_bytes() for the MD5 hash.
Multi-GB Keboola materialized results would OOM on memory-constrained
containers. BQ's version uses streaming 8 KiB-chunk hashing
(connectors/bigquery/extractor.py:438-442); Keboola now mirrors it.
Tests:
- test_run_sync_runs_materialized_pass_on_keboola_only_instance —
pins BUG_0001's fix; setting bigquery.project='' must NOT skip
Keboola materialized dispatch
- test_keboola_materialize_atomic_write_on_failure — pins BUG_0003;
a mid-COPY RuntimeError leaves no .parquet AND no .parquet.tmp at
the canonical path
- test_keboola_materialize_uses_tmp_path_during_copy — documents the
atomic-write contract: COPY targets .parquet.tmp, final swap to
.parquet (no .tmp suffix on the result['path'])
- existing test_run_materialized_pass_dispatches_keboola_to_keboola_extractor
fixture updated: stack_url instead of url
Full sweep: 2505 passed, 25 skipped, 0 failed (modulo 8 pre-existing
internal_roles schema-migration failures called out in the task brief).
Diagnostic + operator-facing documentation that closes the loop on the work in this PR.
`da diagnose` (via /api/health/detailed):
- New _check_bq_billing_project() helper. When data_source.type='bigquery' and BqProjects.billing == .data, surface a yellow warning: 'BigQuery billing project equals data project'. Hint includes the YAML field path + the /admin/server-config UI shortcut. Diagnose's overall status promotes warning → degraded so the CLI echoes it.
- Non-BQ instances (Keboola-only, etc.) skip the check.
- Implementation hooks into the existing /api/health/detailed surface — no new endpoint, no CLI changes.
config/instance.yaml.example documentation:
- data_source.bigquery.billing_project: USER_PROJECT_DENIED hint, /admin/server-config UI reference
- data_source.bigquery.legacy_wrap_views: analyst-side discipline note (use `da fetch` / `da query --remote`), issue #101 history, view-heavy deployment guidance
- data_source.bigquery.max_bytes_per_materialize: cost guardrail block (NEW — wasn't documented in .example before)
- ai.base_url: provider list + UI hint
- openmetadata + desktop: 'configurable via /admin/server-config UI' headers
- corporate_memory: leading note that the schema is editable via UI
Other docs:
- CHANGELOG.md: comprehensive Unreleased section
- CLAUDE.md: schema chain → v20 + Materialized SQL connector mode + per-connector tab UI mention
- README.md: mode-first source table summary
- docs/architecture.md: per-connector tab UI mention
- cli/skills/connectors.md: bootstrap rails (parallel to #154)
- docs/superpowers/plans/2026-05-01-admin-tables-form-cleanup.md: implementation plan archive (2515 lines)
- scripts/seed_dummy_tables.py: drop is_public after #150 RBAC migration (column gone)
Tests:
- test_diagnose_billing.py — 3 cases (BQ with billing==data warns, BQ with billing!=data clean, non-BQ skips)
Today /admin/server-config renders fields by iterating Object.keys(payload) on the YAML value — if a key isn't in instance.yaml, the operator can't see it. They have to know to type it via the JSON-patch textarea (which only renders for empty sections) or SSH and edit YAML.
Adds a known-fields registry (`_KNOWN_FIELDS` in app/api/admin.py) the UI consumes alongside the YAML payload. Renderer shows BOTH:
- existing fields (from YAML) with current value
- known-but-unset fields with dashed-border placeholder + hint, ready to fill in
Renderer (`renderField`, `renderSection`, `collectSection`):
- kind="string"|"secret"|"bool"|"int"|"select"|"object"|"array"|"map" — picks input type
- kind="object" with `fields` — recursive structured form, arbitrary depth (corporate_memory needs 3-4 levels)
- kind="array" with `item_kind` — vertical stack of typed inputs + add/remove buttons
- kind="map" with `key_kind` + `value_kind` — key:value rows + add/remove (used for confidence.base, domain_owners, entity_resolution.entities)
- data-path encoded as JSON segment array so map keys with embedded dots (e.g. 'user_verification.correction') survive collect → patch round-trip
- .cfg-field.is-unset CSS — dashed border, muted label, italic hint
Sections newly exposed (added to _EDITABLE_SECTIONS):
- openmetadata: url, token (secret), cache_ttl_seconds, verify_ssl
- desktop: jwt_issuer, jwt_secret (secret), url_scheme
Known fields populated for existing sections:
- data_source.bigquery: billing_project (the cause of the 403 USER_PROJECT_DENIED footgun when SA can read but not bill the data project), legacy_wrap_views (bigquery_query() wrap for VIEWs — issue #101 default off, ON for view-heavy deployments), max_bytes_per_materialize (cost guardrail)
- data_source.keboola: stack_url, project_id (hints; values already populated)
- ai: base_url (required for openai_compat), structured_output (select)
- corporate_memory: full schema from instance.yaml.example — distribution_mode, approval_mode, review_period_months, notify_on_new_items, sources.{claude_local_md,session_transcripts}, extraction.{model,sensitivity_check,contradiction_check}, confidence.{base,modifiers,decay.{mode,half_life_months,decay_rate_monthly,floor}}, contradiction_detection.{enabled,max_candidates}, entity_resolution.{enabled,entities}, domain_owners, domains
- Known partial: confidence.modifiers is map<string, map<string, float>> — falls through to JSON-textarea with TODO; structured editor for that one shape needs more renderer work
Tests:
- test_admin_server_config_known_fields — registry envelope shape, smoke fixture
- test_admin_server_config_renderer_depth — 4-level nested objects, arrays of strings, maps of floats, dotted-key safety
- test_admin_server_config_corp_memory — full corporate_memory schema, 12 fields incl. nested
- test_admin_server_config — existing tests adjusted for new shape
Replaces the single mixed Jinja-branched form at /admin/tables with a per-connector tab interface and brings Keboola to capability parity with BigQuery.
Tab structure:
- BigQuery tab: Register modal with two-question radio model (Q1 Live | Synced × Q2 Whole | Custom SQL), Discover datasets / List tables / Use-table-as-base autocomplete buttons, table-vs-view auto-detection hint, per-tab listing filter
- Keboola tab: same two-question radio (Q2 only — no Live mode for Keboola), Custom SQL textarea against kbc."bucket"."table" for materialized rows
- Jira tab: read-only listing (Jira is webhook-driven; no Register form)
- Active tab persists in window.location.hash so refresh keeps the operator in place
Form cleanup (within tabs):
- Drops the misleading 'Sync Strategy' dropdown — runtime never read it (only profiler.is_partitioned() consumes the value for parquet-layout detection); kept in DB for back-compat (Pydantic deprecated)
- Adds Sync Schedule input to Keboola Register/Edit (was missing — scheduler honored per-table cron via is_table_due() for every source but the Keboola UI had no surface)
- Hides Primary Key under <details>Advanced with clarifying hint that it's catalog-metadata only (Agnes does not perform upsert/dedup; every sync is a full overwrite)
- Drops the Strategy column from the registry listing (every Keboola row defaulted to full_refresh after Strategy was hidden — column was noise)
- Removes the legacy out-of-tab #registerModal + the legacy global Discovery panel; each tab now owns its own header + Register button + listing div
Edit modal:
- BigQuery Edit modal physically relocated into <section id="tab-content-bigquery"> (mirrors Phase E Register placement)
- Keboola Edit modal mirrors Register (same Q2 radio, Discover/List buttons via parameterized helpers)
- openEditModal(table) dispatches by source_type to the right modal — fixes a quiet bug where Phase F's openEditKeboolaModal was never wired up and Keboola edits silently used the legacy modal
Per-row Manage access deep link:
- Each row in the per-tab listing has a lock-icon button between Edit and Delete that navigates to /admin/access#table:<table_id>
- admin_access.html bootstrap reads window.location.hash and pre-fills the resource filter, mirroring the existing ?group=<id> deep-link pattern
Tests:
- test_admin_tables_tab_ui.py — tab nav, hash persistence, register-button-per-tab, listing partition by source_type, Manage access deep link
- test_admin_tables_ui_materialized.py — two-question radio (BQ + Keboola), Discover/List/Use-as-base buttons, Edit modal parity, Jira read-only
* feat(rbac): drop dataset_permissions + access_requests + users.role + is_public; v19 migration
BREAKING. Sjednocení datové RBAC vrstvy do per-group resource_grants modelu.
Před PR byla legacy data RBAC vrstva (dataset_permissions + is_public bypass)
de-facto neaktivní — is_public neměl API/UI/CLI surface, default true znamenal
že can_access_table vždycky bypassl. Dnes každý non-admin přístup vyžaduje
explicitní resource_grants(group, "table", id) řádek.
Schema v18 → v19 (src/db.py:_v18_to_v19_finalize):
- DROP TABLE dataset_permissions, access_requests
- DROP COLUMN users.role (NULL artifact since v13)
- DROP COLUMN table_registry.is_public
- Drops přes table-rebuild idiom (rename → create new → INSERT … SELECT
→ drop old) kvůli DuckDB ALTER DROP COLUMN limitacím na tabulkách
s historic FK constraints. INSERT picks intersection sloupců, takže
test fixtures s minimal pre-v19 schemou migrate cleanly.
Runtime:
- src/rbac.py:can_access_table → deleguje na app.auth.access.can_access
- DatasetPermissionRepository, AccessRequestRepository smazány
- AGNES_ENABLE_TABLE_GRANTS env-gate v app/resource_types.py odstraněn
(TABLE je unconditionally enabled)
API drop:
- app/api/permissions.py, app/api/access_requests.py celé soubory
- /admin/permissions web route + admin_permissions.html
- "Request Access" modal v catalog.html + locked-row UI
- ~10 if user.get("role") != "admin" checků nahrazeno (admin shortcut
je uvnitř can_access_table)
- /api/settings: drop permissions field z GET; PUT /api/settings/dataset
gate přepnut na can_access(user_id, "table", dataset, conn)
Auth:
- app/auth/jwt.py:create_access_token: drop role parametr (claim zmizí
z nově vydávaných JWT; staré tokeny zůstávají valid, claim ignored)
- app/api/users.py: drop role z CreateUserRequest / UpdateUserRequest
(admin promotion = explicit add to Admin group via memberships API)
- src/repositories/users.py: drop role z create() / update()
CLI:
- da admin set-role smazán → hard-fail s replacement command
- da admin add-user --role flag pryč
- da auth import-token --role flag pryč
- da auth whoami: drop "Role:" výpis
- cli/config.py:save_token: role parametr now optional, no longer written
(back-compat se starými token.json soubory zachována — pole se ignoruje)
Tests:
- DELETE: test_permissions.py, test_permissions_api.py, test_access_requests_api.py
- REWRITE: test_access_control.py (resource_grants flow), test_rbac.py
(can_access_table over resource_grants), test_journey_rbac.py
(drop access-request flow), test_resource_types.py (drop env-gate
tests, drop is_public from helpers), test_v2_*.py (drop role-based
user dicts in favor of id-based + Admin group membership),
test_settings_api.py (no permissions field, can_access gate)
- TRIVIAL: ~30 souborů — drop role="admin" arg z UserRepository.create
a 3rd positional role z create_access_token
- NEW: test_v18_to_v19 migration test (test_db.py),
test_can_access_table_no_implicit_public (test_rbac.py),
test_admin_set_role_returns_hardfail (test_cli_admin.py)
- OpenAPI snapshot regenerated
Docs:
- CHANGELOG: BREAKING entry pod [Unreleased]
- CLAUDE.md: schema v18 → v19
- docs/architecture.md: schema table + RBAC sekce přepsána
- docs/auth-google-oauth.md: admin promotion přes da admin break-glass
- cli/skills/security.md: kompletně přepsáno na group-based model
- docs/TODO-rbac-data-enforcement.md: smazáno (TODO splněn)
Test results: 2363 passed, 19 failed. Zbývající failures jsou pre-existing
Windows-specific issues (fcntl, charset) nesouvisející s tímto PR —
ověřeno git stash pop.
Plan: ~/.claude/plans/floofy-coalescing-parnas.md
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* chore(release): cut 0.27.0
---------
Co-authored-by: Minas Arustamyan <arustamyan.minas@gmail.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: ZdenekSrotyr <zdenek.srotyr@keboola.com>
* docs(spec): #134 unify BigQuery access behind BqAccess facade
Brainstorm output for issue #134. Captures:
- root cause (incl. correction of the issue's hypothesis about commit 33a9964)
- BqAccess facade API + project resolution rules
- error contract — typed BqAccessError mapped to HTTP 502 for upstream
BQ failures, 500 for deployment/config bugs
- migration plan for v2_scan, v2_sample, RemoteQueryEngine
- test rewrite eliminating _bq_client_factory injection point
- E2E verification protocol on agnes-development as success criterion
* docs(spec): #134 revise after first review
Incorporates code-reviewer findings:
Must-fix:
- Add v2_schema (2 copies of INSTALL/LOAD/SECRET dance) to migration scope.
- Reframe v2_scan headline: missing try/except around BQ calls is the
actual cause of bare 500s, not project resolution (which 33a9964 fixed).
- List two more deferred call sites (extractor.py, register_bq_table)
with explicit rationale.
Important:
- Drop billing != data clause from cross_project_forbidden heuristic;
rely only on 'serviceusage' substring. billing != data is normal
for cross-project setup, was over-classifying.
- Split bq_bad_request into _user (400) and _server (502) variants;
add sql_origin parameter to translate_bq_error so call sites declare
whether SQL contains user input.
- Add @functools.cache to BqAccess.from_config; document tests bypass
via dependency_overrides.
- Replace monkey-patched-classmethod test pattern with
BqAccess(client_factory=...) injection at construction time. Cleaner
than today's _bq_client_factory and 1:1 migration shape.
- Keep BqProjects.data (reviewer assumed registry has source_project;
it doesn't). Multi-project explicitly listed as non-goal with note.
Nice-to-have:
- Add 'Implementation strategy' section: 2 staged commits (bug fix
alone is revertable; refactor follows).
- Extend E2E protocol to cover all three endpoints, not just /sample.
- Note removal of stale docstring at src/remote_query.py:204.
* docs(spec): #134 revision 3 — incorporates second-round review
Must-fix from second review:
- v2_schema split into two migration cases: _fetch_bq_schema translates
errors via translate_bq_error; _fetch_bq_table_options preserves its
swallow-all 'except Exception → return {}' so /schema doesn't 502 on
partition-info failures.
- RemoteQueryEngine.__init__ now resolves BqAccess lazily (in
_get_bq_client, not in __init__). Without this, ~7 DuckDB-only tests
in test_remote_query.py would suddenly fail with not_configured.
- translate_bq_error pass-through for BqAccessError is now load-bearing
(clause 1, before any Google-API branch). bq.client() raises BqAccessError
for bq_lib_missing/auth_failed; without explicit pass-through those
fall to 'unknown' and re-raise as bare 500.
- Commit 1 now emits the SAME structured response shape as commit 2 to
avoid contract churn between commits.
- BIGQUERY_PROJECT env-var precedence is BREAKING for env-only deployments
— flagged in CHANGELOG ### Changed.
Editorial:
- sql_origin renamed to bad_request_status with values 'client_error' /
'upstream_error' (clearer about what the parameter actually decides).
bq_bad_request_user/_server kinds collapsed to bq_bad_request (400)
and bq_upstream_error (502).
- CLI (cli/commands/query.py) noted as external RemoteQueryEngine caller;
unaffected because new bq_access kwarg has default None.
- Added unit/integration tests for the new contracts:
test_translate_passes_through_BqAccessError,
test_v2_scan_returns_500_on_bq_lib_missing,
test_v2_schema_returns_200_with_empty_partition_on_bq_failure,
test_resolve_succeeds_after_config_set.
- E2E protocol now covers /schema as the fourth endpoint.
- Documented functools.cache-doesn't-cache-exceptions semantics and
fixture nullcontext-doesn't-close caveat for nested sessions.
* docs(spec): #134 revision 4 — incorporates third-round review
Third reviewer verdict: 'implementation-ready with two trivial edits';
explicitly noted prior rounds did the heavy lifting.
Edits:
1. get_bq_access() module-level function instead of @classmethod
@functools.cache from_config. Removes the classmethod-cache stacking
footgun (different Python versions wrap differently) and gives FastAPI's
dependency introspection a clean function signature. Drops the
'Do not subclass BqAccess' caveat that no longer applies.
2. Commit 1 strategy explicitly: wrap _fetch_bq_sample (v2_sample),
_bq_dry_run_bytes + _run_bq_scan (v2_scan), and _fetch_bq_schema
(v2_schema strict block). Do NOT touch _fetch_bq_table_options swallow-all
in commit 1 — preserved as-is, then migrated (still preserved) in commit 2.
All three endpoints emit the same structured body shape so client parsers
see one consistent contract throughout the staged rollout. No more
half-rolled-out window where /sample is bare 500 while /scan is
structured 502.
* docs(plan): #134 implementation plan — Phase 1 (atomic bug fix) + Phase 2 (BqAccess refactor) + Phase 3 (verification)
Bite-sized TDD tasks. 3 phases, 16 tasks total:
Phase 1 (Commit 1) — atomic bug fix across all four v2 endpoints:
Tasks 1.1-1.5 wrap _fetch_bq_sample, _bq_dry_run_bytes, _run_bq_scan,
_fetch_bq_schema with structured 502/400 try/except. _fetch_bq_table_options
preserved untouched. CHANGELOG Fixed entries.
Phase 2 (Commit 2) — BqAccess facade extraction + migration:
Tasks 2.1-2.5 build connectors/bigquery/access.py bottom-up
(BqProjects, BqAccessError, translate_bq_error, default factories,
BqAccess class, get_bq_access module-level cached). Task 2.6 adds
conftest.py fixture. Tasks 2.7-2.9 migrate v2_scan, v2_sample, v2_schema
to BqAccess. Tasks 2.10-2.11 migrate RemoteQueryEngine + tests
(lazy bq_access, drop _bq_client_factory). Task 2.12 CHANGELOG
Changed BREAKING + Internal.
Phase 3 — Verification:
3.1 full pytest. 3.2 squash into two PR-shape commits. 3.3 manual
E2E on agnes-development per spec protocol → close#134.
Self-review table maps spec sections to implementing tasks; no gaps.
* fix(v2): #134 structured 502/400 on BQ errors across /scan, /scan/estimate, /sample, /schema
Wraps the BigQuery call sites in v2_scan, v2_sample, and v2_schema (strict
block only) with try/except for google.api_core exceptions, translating to
HTTPException with a structured body shape: {error, message, details}.
Fixes Pavel's report (#134) where these endpoints returned bare HTTP 500
with no body when the SA on agnes-development hit cross-project Forbidden
on serviceusage.services.use.
Also fixes /sample's missing billing_project fallback (the bug 33a9964
fixed for /scan never landed here).
Status code split:
- /scan, /scan/estimate: BadRequest -> 400 (bq_bad_request) since SQL is
user-derived from req.select/where/order_by.
- /sample, /schema: BadRequest -> 502 (bq_upstream_error) since SQL is
server-constructed from validated identifiers.
- All Forbidden -> 502 with cross_project_forbidden if 'serviceusage' in
error message (with hint pointing at data_source.bigquery.billing_project),
else bq_forbidden.
Body shape matches what the upcoming BqAccess refactor (next commit) will
produce, so client-side parsers see one consistent contract throughout
the staged rollout.
_fetch_bq_table_options preserved exactly as-is — its swallow-all-and-return-empty
contract is intentional and survives into the refactor; /schema continues to
return 200 with empty partition info when partition queries fail.
Outer wraps in scan_endpoint, scan_estimate_endpoint, sample, and schema
endpoints exist only to make the test pattern (monkeypatching whole
_fetch_* functions) work, and are tagged TODO(#134 Phase 2) for removal
once BqAccess centralizes translation.
* refactor(bq): #134 BqAccess facade — unify v2_scan, v2_sample, v2_schema, RemoteQueryEngine
Extracts the duplicated BigQuery-access pattern (project resolution +
client construction + DuckDB-extension session + Google-API error
translation) into connectors/bigquery/access.py. Migrates four
call sites to use it:
- app/api/v2_scan.py — _bq_dry_run_bytes, _run_bq_scan
- app/api/v2_sample.py — _fetch_bq_sample
- app/api/v2_schema.py — _fetch_bq_schema (strict translation),
_fetch_bq_table_options (preserves swallow-all best-effort contract)
- src/remote_query.py — RemoteQueryEngine, lazy bq_access kwarg
The new module exposes:
- BqProjects (frozen dataclass: billing + data project IDs)
- BqAccessError (typed exception with HTTP_STATUS class mapping)
- BqAccess (facade with injectable client_factory/duckdb_session_factory
for tests; defaults call the real google-cloud-bigquery + DuckDB extension)
- get_bq_access (module-level @functools.cache; FastAPI Depends target)
- translate_bq_error (Google API exception → BqAccessError mapper, with
BqAccessError pass-through, 'serviceusage'-substring heuristic for
cross_project_forbidden, and bad_request_status param distinguishing
user-derived (400) from server-constructed (502) SQL)
- _default_client_factory, _default_duckdb_session_factory
RemoteQueryEngine.__init__ no longer accepts _bq_client_factory; tests
migrate to bq_access=BqAccess(projects, client_factory=...). DuckDB-only
RemoteQueryEngine tests need no changes — bq_access defaults to None and
get_bq_access() is only invoked on first BQ call (lazy resolution).
BqAccessError raised internally is translated to RemoteQueryError(
error_type="bq_error") in _get_bq_client to preserve the engine's
existing public contract — CLI and /api/query/hybrid callers see no change.
Endpoint tests (test_v2_scan, test_v2_scan_estimate, test_v2_sample,
test_v2_schema) migrate from monkey-patching whole _fetch_* functions
to using the new bq_access fixture in tests/conftest.py — which
exercises the REAL translation path through BqAccess + translate_bq_error,
closing the test gap flagged in Task 1.1's review.
Side-effect behavior change: v2_sample's FROM clause now uses the data
project (instance.yaml data_source.bigquery.project), not the conflated
billing_project from Phase 1. Documented in CHANGELOG ### Internal.
BREAKING for deployments combining BIGQUERY_PROJECT env var with
data_source.bigquery.project in instance.yaml — env var now overrides
data project too. See CHANGELOG ### Changed.
Two known-duplicate BQ-access sites (connectors/bigquery/extractor.py,
scripts/duckdb_manager.register_bq_table) explicitly out of scope;
tracked as follow-up.
Removed stale docstring at the previous src/remote_query.py:204
that referenced scripts.duckdb_manager._create_bq_client as the default
BQ client factory (RemoteQueryEngine never actually used that function).
Test counts: tests/test_bq_access.py +27 (new), tests/test_v2_*.py +
tests/test_remote_query.py migrated to bq_access fixture (counts unchanged
or +1-2 per file). Full suite: 2086 passed, 8 pre-existing failures
(DB migration tests with unrelated internal_roles DependencyException —
not introduced by this PR).
* fix(bq_access): translate DefaultCredentialsError to BqAccessError(auth_failed)
CI on PR #138 caught: bigquery.Client(...) resolves Application Default
Credentials at construction time; without ADC (CI without SA key, dev
laptop without 'gcloud auth application-default login') it raises
google.auth.exceptions.DefaultCredentialsError synchronously.
Pre-fix _default_client_factory only caught ImportError, so DefaultCredentialsError
propagated as raw exception — and from production endpoints would surface
as bare 500 (the exact failure mode #134 sets out to fix).
Now translates to BqAccessError(kind='auth_failed', details.hint='Run
gcloud auth application-default login...'). Endpoint catch chain returns
HTTP 502 with structured body. Adds unit test
test_raises_auth_failed_on_default_credentials_error.
Third-round spec review flagged this case in passing; the fix didn't land.
CI's auth-less environment surfaced it.
* fix(bq_access): get_bq_access() returns sentinel instead of raising when not configured
Devin BUG_0001 on PR #138 review: 'get_bq_access() as FastAPI Depends
breaks all v2 endpoints for non-BigQuery instances'.
Pre-fix: get_bq_access() raised BqAccessError(not_configured) when
neither BIGQUERY_PROJECT env nor data_source.bigquery.project was set.
Because FastAPI resolves Depends() BEFORE the endpoint body runs, this
exception fires during dep-injection — the endpoint's try/except
BqAccessError clause never gets a chance to catch it. Result: every
v2 request on Keboola-only or CSV-only instances returned bare HTTP
500, even for local-source tables that never touch BigQuery.
Fix: get_bq_access() now returns a sentinel BqAccess with empty
BqProjects and factories that raise BqAccessError(not_configured)
on actual use. Construction succeeds, FastAPI's dep-injection cleanly
yields the sentinel, the endpoint runs. The local-source code path
in build_sample / build_schema / etc. never calls bq.client() or
bq.duckdb_session() (it reads parquet directly), so non-BQ tables
return 200 as before. Only when an endpoint actually tries to query
BQ (source_type == 'bigquery') does the sentinel raise — and the
endpoint's existing except BqAccessError catches it normally,
returning structured 502 with hint.
Test get_bq_access::test_raises_not_configured_when_neither_set
renamed and rewritten to test_returns_sentinel_when_neither_set:
asserts BqAccess is returned, then asserts client() and
duckdb_session() each raise BqAccessError(not_configured) on call.
Test test_does_not_cache_exceptions removed (no longer applicable)
and replaced with test_sentinel_is_cached_per_process documenting
the operator-restart-on-config-change contract.
* docs(spec+plan): #134 genericize customer-specific tokens (CLAUDE.md OSS rule)
Devin BUG_0001/0002 round 3 on PR #138: spec and plan docs contained
customer-specific deployment hostnames, deployment names, and a GCP
project ID that violated CLAUDE.md's vendor-agnostic OSS rule
('Nothing customer-specific belongs in code, configuration defaults,
comments, docs, commit messages, PR titles, or PR bodies').
Replacements:
agnes-development.groupondev.com -> <your-agnes-host>
agnes-development -> <your-dev-instance>
prj-grp-dataview-prod-1ff9 -> <your-data-project>
s1_session_landings -> <bq_table_id>
E2E verification semantics unchanged — operators still run the same
four curls + config flip + retry, just substituting their own host /
deployment name / project / table.
* fix(bq_access): hook get_bq_access.cache_clear into instance_config.reset_cache
Devin ANALYSIS_0004 on PR #138: get_bq_access is @functools.cache'd at
process level, so it captures BigQuery project IDs at first call and
ignores subsequent instance.yaml changes. Pre-Phase-2 the v2 endpoints
re-read get_value() on every request, so admin /api/admin/server-config
saves (which call instance_config.reset_cache()) hot-reloaded the BQ
project. Without this fix, my refactor silently regresses that contract
— operators editing instance.yaml via the admin UI would see no effect
on v2 endpoints until container restart.
instance_config.reset_cache() now also calls
connectors.bigquery.access.get_bq_access.cache_clear() (lazy import,
swallowed if connectors module isn't loaded — keeps instance_config
usable in isolated unit tests).
Adds test_instance_config_reset_cache_invalidates_get_bq_access as
regression guard. Updates CHANGELOG Internal entry to mention the
hot-reload contract + the not-configured sentinel behavior (round-3
fix from Devin BUG_0001 was previously only in commit message).
* fix(bq_access): surface not_configured before identifier validation + plan path genericize
Devin BUG_0001 + BUG_0002 round 5 on PR #138.
BUG_0001 (plan doc): personal filesystem path violated CLAUDE.md
vendor-agnostic rule. Replaced with '<worktree-root>' placeholder.
BUG_0002 (sentinel error path): when get_bq_access() returns the sentinel
BqAccess (BQ not configured), the empty bq.projects.data was reaching
validate_quoted_identifier first and raising ValueError -> endpoint
mapped to HTTP 400 'unsafe_identifier' instead of structured 500
'not_configured' with hint.
Each fetch helper now checks 'if not bq.projects.data: bq.client()' as
the first step, which triggers the sentinel's BqAccessError(not_configured).
Endpoint catches the typed error and returns HTTP 500 with hint pointing
at data_source.bigquery.project. Best-effort _fetch_bq_table_options
returns {} silently in this case (preserves the swallow-all contract).
* fix(bq_access): classify DuckDB-native exceptions from bigquery_query() via string match
Devin ANALYSIS on PR #138 review (latest round). The DuckDB bigquery
extension is a C++ plugin making its own HTTP calls — when BQ returns
403, it throws duckdb.IOException with the BQ error embedded as text,
not gax.Forbidden. translate_bq_error's isinstance checks would miss
these, falling to case 7 → bare 500 in production for v2_scan, v2_sample,
and v2_schema (the bigquery_query() paths).
Fix: last-resort string-match heuristic before the re-raise. 'Forbidden'
/ '403' / 'Bad Request' / '400' in the lowercased message classifies via
the same kind hierarchy. The 'serviceusage' substring still distinguishes
cross_project_forbidden from bq_forbidden. Specific enough that random
exceptions without HTTP-error keywords still re-raise.
Adds 4 unit tests covering the new heuristic + the 'don't swallow random
exceptions' invariant.
* chore(release): cut 0.22.0
PR #138 contains issue #134 user-visible behavior changes:
- BREAKING: BIGQUERY_PROJECT env var now overrides instance.yaml
data_source.bigquery.project for v2 endpoints (previously
RemoteQueryEngine billing only).
- Fixed: structured 502/400 on /api/v2/sample, /scan, /scan/estimate,
/schema when BigQuery raises Forbidden/BadRequest (was bare 500).
- Internal: BqAccess facade refactor unifying four duplicate BQ-access
call sites; instance_config.reset_cache() now invalidates BqAccess
cache too so admin server-config saves hot-reload BQ project IDs.
Bumps to 0.22.0 because PR #137 merged first and took 0.21.0.
Bootstraps the Agnes Claude Code marketplace + RBAC-allowed plugins from
the dashboard CTA, and inlines the server's TLS cert when the chain isn't
publicly trusted (self-signed / private CA). Cross-platform setup prompt
covers Windows Git Bash, macOS, Linux. Includes Bun-compiled `claude` fix
(macOS goes via git-clone fallback, same as Windows), PAT stripping after
clone, explicit error handling, and four rounds of Devin Review fixes
(phantom step references, $PLATFORM re-detection, heredoc/awk line-count
sync). Cuts 0.21.0.
See CHANGELOG.md [0.21.0] section for details.
Closes the /plugin UI 'Plugin <X> not found in marketplace' bug. Synth marketplace.json catalog 'name' now reads from <plugin_dir>/.claude-plugin/plugin.json (with fallback to upstream marketplace.json name). On-disk plugins/<slug>-<plugin>/ layout preserved so cross-marketplace files don't collide. /marketplace/info exposes both name and prefixed_name (BREAKING — downstream tooling parsing 'name' for the slug-prefixed form must switch to prefixed_name).
Three new env vars wire the Google OAuth callback to a configurable Workspace prefix and route admin/everyone Workspace groups onto the seeded system rows: AGNES_GOOGLE_GROUP_PREFIX, AGNES_GROUP_ADMIN_EMAIL, AGNES_GROUP_EVERYONE_EMAIL. Login gate redirects users with no prefix-matching group to /login?error=not_in_allowed_group. BREAKING: auto-Everyone membership for new users removed. Admin UI/API are read-only on Google-managed groups. See docs/auth-groups.md.
Issue #108 Milestone 1. Adds BigQuery table registration via /admin/tables UI and `da admin register-table` CLI without hand-editing table_registry. POST /api/admin/register-table/precheck for round-trip validation. --dry-run flag on CLI. Audit-log entries on register/update/unregister. PUT /api/admin/registry/{id} now preserves registered_at (closes#130).
* fix(scheduler): HTTP marketplaces job + SCHEDULER_API_TOKEN shared secret
Two scheduler-reliability bugs surfaced after the v0.12.1 USER-agnes flip:
1. The marketplaces job called src.marketplace.sync_marketplaces() in-process
from the scheduler container, racing the app's long-lived system.duckdb
handle. DuckDB rejects cross-process writers — every cron tick 500-ed on
"Could not set lock on file ... PID 0".
2. The data-refresh + new marketplaces jobs both 401-ed on the API because
SCHEDULER_API_TOKEN was never propagated by the Terraform startup script.
The scheduler had no credential to authenticate with.
Fix:
- New POST /api/marketplaces/sync-all (admin-only) drives the nightly refresh
through the app process so it inherits the existing DB connection.
- Scheduler swaps fn->http for marketplaces; all jobs are now plain HTTP and
the scheduler is reduced to a cron clock.
- New app/auth/scheduler_token.py adds a shared-secret auth path. The
startup script generates a 256-bit secret on first boot, persists it
across reboots, and writes it to /opt/agnes/.env. Both containers source
the same .env. The app validates incoming Bearer tokens against the env
var (constant-time, length-floored) and resolves matches to a synthetic
scheduler@system.local user that's a member of the Admin system group.
Audit-log entries from the scheduler are attributed to this user.
- app/main.py seeds the synthetic user at startup so the first cron tick
has a valid actor; lazy seed in get_scheduler_user covers token rotation
before the next app restart.
Tests: 5 new in tests/test_auth_scheduler_token.py covering empty/short
secret rejection, exact-match comparison, idempotent user seeding, and
lazy provisioning. 142 marketplace + scheduler tests + 96 auth tests
remain green.
Existing VMs with .env from before this change need a one-time
re-provisioning (re-run startup-script or rotate via openssl rand);
documented in CHANGELOG.
* fix(audit): use '_all' sentinel for bulk marketplace sync — Devin review #127
Avoids the literal string 'marketplace:None' in the audit_log resource
column when the bulk sync endpoint writes its summary row.
* fix(scheduler): unblock event loop + per-job timeouts — Devin review #127
Two findings from Devin re-review on commit 5fbad15:
1. BUG: trigger_sync_all was async def, so FastAPI ran it on the asyncio
event loop. sync_marketplaces() does blocking I/O (subprocess git
clones up to GIT_TIMEOUT_SEC=300 each, threading.Lock, DuckDB writes)
and would freeze every concurrent request for the duration of a bulk
sync. Switched to plain def so FastAPI auto-routes to the thread pool.
2. ANALYSIS: scheduler used a fixed 120s httpx timeout for every POST.
Bulk marketplace sync iterates the registry under a single lock with
up to 300s per repo — easily exceeds 120s on 2-3 slow repos. The
scheduler then sees a timeout, doesn't update last_run, and re-fires
on the next 30s tick, queueing redundant work. Per-job timeout
override added to the JOBS tuple; marketplaces gets 900s (15 min),
data-refresh keeps 120s, health-check 30s.
* fix(auth): require_session_token rejects scheduler shared secret — Devin review #127
require_session_token gates /auth/tokens (PAT minting). Pre-fix it only
rejected JWTs with typ=pat — but the scheduler shared secret is an opaque
string, so verify_token() returns None, payload becomes {}, and the
PAT-claim check silently passed. A caller bearing SCHEDULER_API_TOKEN
could mint persistent PATs that survive a secret rotation.
Added explicit is_scheduler_token() check before the PAT-claim check;
new regression test in tests/test_auth_scheduler_token.py.
Devin's other note (pre-existing async def trigger_sync at marketplaces.py:392
also calls blocking sync_one) — Devin flagged it as out-of-scope for this PR
and I agree; tracking separately.
* release(0.17.0): cut + clean up CHANGELOG duplicates
Cuts 0.17.0 (minor: scheduler shared-secret auth + sync-all endpoint
plus the deploy-shape fixes that landed since the last release tag).
Bumps pyproject from 0.15.0 — also corrects the missed bump from PR #120
(v0.16.0 was tagged on GitHub and shipped as :stable, but pyproject
stayed at 0.15.0, so /api/version, /cli/latest, and `da --version` had
been under-reporting the running release).
Removes the long-form duplicate entries for 0.13.0 / 0.14.0 / 0.15.0
above [0.16.0] — the canonical short summaries (with GitHub-release
links) already exist below 0.16.0, the long forms were leftover state
from before those versions were cut and have been silently shadowed
ever since.
v13 RBAC migration nulled users.role and moved admin authority onto user_group_members. Header still gated on session.user.role == 'admin', so admin menu was hidden for everyone. Inject user['is_admin'] via is_user_admin in get_current_user; header reads session.user.is_admin.
Move {% include '_app_header.html' %} out of .container-memory (max-width: 1000px)
in corporate_memory.html and corporate_memory_admin.html so the header spans the
viewport, matching dashboard.html. Page content stays constrained by the container.
Adds /me/debug HTML page rendering the logged-in user's own session state — decoded JWT claims (no raw token, sha256[:12] fingerprint for log correlation), group memberships with sources and bound external_id when present, resource grants effective via those memberships, and a Refetch from Google (dry-run) button that diffs a fresh fetch_user_groups call against the cached user_group_members snapshot. Gated by AGNES_DEBUG_AUTH env var (default off → 404, route existence undetectable in production). Self-only by construction: user_id is read from the validated session, never echoes raw JWT / password hash / full PAT. Tolerates v13 + v14 schemas via information_schema check on users.external_id.
Replaces the BigQuery wrap-view pattern with a discovery + scoped-fetch toolkit driven by the analyst's Claude session. Adds /api/v2/{catalog,schema,sample,scan,scan/estimate}, da catalog/schema/describe/fetch/snapshot/disk-info CLI commands, sqlglot-backed WHERE validator, process-local quota tracker, agent rails skill (cli/skills/agnes-data-querying.md). BREAKING: BQ wrap views off by default — set data_source.bigquery.legacy_wrap_views=true for one cycle. Backward-compat field_validator on primary_key. Catalog cache now matches documented 300s TTL with RBAC fresh per request. Cuts release v0.14.0.
Adds /admin/server-config UI for editing instance.yaml from the web. Hardening: SSRF gate on data_source URLs, narrow-overlay write strategy, atomic writes, audit log with secret masking on shape changes, threading lock on read-modify-write, corrupt-overlay refusal on write side + louder log on read side, modal Promise resolution on backdrop dismiss, sentinel scrub on save (defense-in-depth client+server). Bundles Windows PowerShell wrapper from #80. Cuts release v0.13.0.
* fix(security+ops): #82#85#87 — auth hardening, API validation, deploy posture
Security and operational hardening across three issue groups:
- M23: docker-compose.override.yml → docker-compose.dev.yml (BREAKING, prod foot-gun)
- C13: Container runs as non-root user 'agnes' (USER directive in Dockerfile)
- M21: Docker resource limits (mem_limit, cpus) on app + scheduler
- M22: Caddyfile security headers (X-Frame-Options, X-Content-Type-Options, Referrer-Policy, -Server)
- M17: /api/health split into minimal (unauth) + /api/health/detailed (auth) (BREAKING)
- M26: release.yml restricts build-and-push to main + workflow_dispatch; paths-ignore for docs
- C2: table_id traversal validation on /api/data/{table_id}/download
- M4: Upload streaming (chunk-read + temp file) instead of full-buffer; /local-md hashed filename
- C5: reset_token removed from POST /api/users/{id}/reset-password response
- C8: Startup WARNING when no user has password_hash (bootstrap window visible)
- M9: Audit log on failed web form login (mirrors /auth/token endpoint)
- M10: Atomic magic-link consume via compare-and-swap (CONSUMED: marker + DuckDB conflict catch)
Also: SSRF protection on /api/admin/configure (#46), memory stats SQL aggregation (#90)
Generated with [Devin](https://cli.devin.ai/docs)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(review): SSRF 169.254.x.x + IPv6 multicast; M10 marker cleanup safety
Review fixes:
- Add 169.254.0.0/16 (link-local, cloud metadata) to SSRF regex — was
missing, allowing requests to AWS/GCP/Azure metadata endpoints
- Add ff[0-9a-f]{2}: (IPv6 multicast) to SSRF regex
- M10: wrap Step 3 (CONSUMED marker cleanup) in try-except with
warning log — prevents unhandled exception if DB write fails after
successful token consumption
- Add test for 169.254.169.254 SSRF rejection
Generated with [Devin](https://cli.devin.ai/docs)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(review): SSRF IPv6 bypass, CLI health endpoint, upload FD leak
Address Devin Review findings on PR #104:
1. SSRF IPv6 bypass: Replace hostname regex with DNS resolution +
ipaddress module checks. The old regex patterns like `fe80:` only
matched up to the first colon, missing real IPv6 addresses like
`fe80::1`, `fc00::1`, `ff02::1`. The new approach resolves the
hostname via getaddrinfo and checks each resulting IP against
ipaddress.is_private/is_loopback/is_link_local/is_reserved/is_multicast.
2. CLI commands broken: `da setup test-connection`, `da setup verify`,
`da diagnose`, `da status` all called /api/health expecting the old
format (status=="healthy", services dict). Now they call
/api/health/detailed for service-level checks (with graceful fallback
to the minimal endpoint when auth is not configured).
3. Temp file handle leak: _stream_to_temp returns an open
NamedTemporaryFile; callers now close it before shutil.move() to
prevent FD leaks until GC.
Also adds IPv6 SSRF test cases (loopback, link-local, unique-local,
multicast) with mocked DNS resolution for test environment independence.
Generated with [Devin](https://cli.devin.ai/docs)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(review): download regex blocks hyphenated IDs; document health split
Address Devin Review round-3 findings on PR #104:
1. _SAFE_IDENTIFIER regex blocked hyphenated table IDs: The download
endpoint used the strict SQL-identifier regex which does not allow
dots or hyphens, but Keboola table IDs like in.c-crm.orders
contain both. Switched to _SAFE_QUOTED_IDENTIFIER which allows dots
and hyphens while still blocking path-traversal chars (/, .., \)
and quote/control characters. Added test for hyphenated/dotted IDs.
2. Documented health endpoint split in DEPLOYMENT.md: Added Health
checks & external monitoring section explaining both endpoints
(minimal unauth /api/health vs authenticated /api/health/detailed)
and how to wire external monitoring tools to the detailed endpoint
with a PAT.
Generated with [Devin](https://cli.devin.ai/docs)
Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* release(0.12.1): cut hotfix for snapshot integrity + #82/#85/#87 hardening
* fix(security): apply CAS pattern to password reset confirm (#82/M10 follow-up)
Devin review on the rebased PR flagged the asymmetry: magic-link verify
got the atomic compare-and-swap pattern in the original M10 fix, but
password reset confirm at /auth/password/reset/confirm was still using
read-validate-clear. Two concurrent POSTs with the same valid reset
token could both succeed in setting different new passwords (last-write-
wins). Lower severity than the magic-link race because the attacker
would need the reset token AND to race the legitimate user, but the
asymmetry was a polish gap.
Mirrors app/auth/providers/email.py::_consume_token CAS exactly: write
unique CONSUMED:<random> marker via UPDATE...WHERE token=old_token, then
SELECT to verify our marker won, then proceed. Only the winner clears
the marker and applies the password change.
New regression test_concurrent_reset_only_one_wins in
tests/test_password_flows.py::TestResetConfirm pins the contract: two
ThreadPoolExecutor workers + Barrier hit /reset/confirm with the same
token; exactly one gets 302 (password applied), the other gets 200 with
'Invalid or expired'. Sanity-checked against the pre-CAS code — both
POSTs got 302 (race confirmed).
---------
Co-authored-by: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Two Devin-flagged regressions on the squashed PR #106 head:
1) PATCH /api/admin/groups/{id} blanket-rejected on system groups.
The repository guard at src/repositories/user_groups.py was already
narrowed to "rename only" by 7147bac (PR #110 follow-up), but the
endpoint at app/api/access.py:331-343 still short-circuited with
409 "System groups are immutable" for any mutation. A description-only
payload like {"description": "..."} returned 409 instead of 200 even
though the repo would have accepted it. CHANGELOG entry promised the
fix but the code didn't match.
Endpoint now mirrors the repo contract: 409 only when payload.name
is set AND differs from existing name. Same-name no-op renames are
dropped before the repo call. Description-only updates flow through.
2) Google OAuth callback wiped google_sync memberships on transient
API failure.
fetch_user_groups is fail-soft and returns [] for both "user has no
groups" and "Cloud Identity API error". The callback fed that empty
list into replace_google_sync_groups, which DELETEs all rows with
source='google_sync' for the user then INSERTs zero — silently
wiping every Workspace-synced membership on a hiccup.
Callback now skips replace_google_sync_groups when group_names is
empty and logs "preserving existing memberships". Trade-off: a user
whose Workspace groups were genuinely cleared keeps stale memberships
until the next non-empty sync. Admin-added rows (source='admin') were
already protected by source-scope and are unaffected. The previous
guard against this exact regression was test_callback_empty_groups_
does_not_overwrite_existing in tests/test_auth_providers.py — that
test class has been skipped since v12 (asserts users.groups JSON,
needs rewrite for user_group_members).
Follow-up to the RBAC v13 + marketplace work in the parent commit. Addresses
deferred Devin findings, gemini-flagged blockers, and adds three guard rails.
== Schema v14 — FK constraints on user_group_members + resource_grants ==
Adds DuckDB foreign-key constraints so cascade deletes can no longer leave
orphaned member / grant rows pointing at a deleted group_id (which were
relying on application-level cascades up to v13). Migration is RENAME →
CREATE-with-FK → INSERT → DROP, wrapped in BEGIN TRANSACTION so a partial
failure rolls back without leaving the DB at a half-applied schema.
== AGNES_ENABLE_TABLE_GRANTS feature flag (default off) ==
ResourceType.TABLE was shipped in the parent commit as listing-only — admins
can record grants but runtime enforcement still flows through legacy
dataset_permissions. To avoid the misleading-UX surface area, the chip is
hidden from /admin/access and POST /api/admin/grants returns 422 with the
env-var name in detail until the operator opts in. Existing TABLE rows in
resource_grants stay listable + deletable so cleanup is never blocked.
Helpers: is_resource_type_enabled(rt), enabled_resource_types().
== Break-glass admin CLI ==
`da admin break-glass <user>` adds the user to the Admin user_group with
source='system_seed' regardless of RBAC state. Bypasses authentication —
relies on filesystem access to ${DATA_DIR}/state/system.duckdb implying
host-level trust. Recovery path when the operator has locked themselves
out of /admin/access.
== Devin round-2 fixes (deferred on b4ec4c4) ==
- src/repositories/user_groups.py — narrow update() guard from blocking any
mutation on system groups to blocking name change only. Description edits
now pass through. Endpoint pre-check stays as defense-in-depth. Prior
behavior surfaced as a misleading 409 'Cannot rename a system group' on
description-only PATCH.
- app/api/access.py:delete_group — wrap cascade DELETEs + repo.delete in
BEGIN TRANSACTION / COMMIT / ROLLBACK. Prevents orphan rows if any
DELETE fails after the user_groups row is gone.
- app/marketplace_server/{packager,router}.py — split compute_etag_for_user()
from build_zip(); router resolves etag first and 304-shorts before any
file read or ZIP_DEFLATED. In-process cachetools.TTLCache (default 120s,
env-tunable via AGNES_MARKETPLACE_ETAG_TTL, set 0 to disable).
invalidate_etag_cache() called by sync to force re-hash on content drift.
== Tests ==
- TestTableGrantsFeatureFlag (4 cases) — endpoint exclude/include, grant
rejection/acceptance under the flag.
- test_v12_to_v13_finalize_rollback_on_failure — destructive: monkeypatches
_seed_system_groups to raise mid-transaction, asserts schema_version stays
at 12, legacy tables intact, new tables empty (rollback fired). Then
restores the real function and asserts the retry succeeds.
- test_update_system_group_description_allowed,
test_update_system_group_same_name_no_op — repo-level coverage of the
narrowed guard.
This squashes 13 commits from ma/staging plus a small docstring translation
into a single coherent unit. Three workstreams.
== RBAC v13 redesign ==
- Drops core.viewer/analyst/km_admin/admin hierarchy and the
internal_roles / group_mappings / user_role_grants / plugin_access tables.
- Replaced by user_group_members + resource_grants. Atomic v12→v13 backfill
wrapped in BEGIN/COMMIT; ROLLBACK leaves schema_version at 12 for retry.
- Two authorization primitives in app.auth.access:
require_admin — Admin-group god-mode
require_resource_access(rt, "{path}") — entity-scoped grants
Single DB lookup per request; no session cache; no implies BFS.
- /admin/access UI (single page) replaces /admin/role-mapping +
/admin/plugin-access. CLI `da admin group/grant *` replaces
`da admin role/mapping/grant-role/revoke-role/effective-roles`.
- ResourceType.TABLE listing-only — admins can record table grants,
runtime enforcement still flows through legacy dataset_permissions
(migration plan in docs/TODO-rbac-data-enforcement.md).
== Claude Code marketplace ==
- Aggregated /marketplace.zip + /marketplace.git/* (PAT-gated,
RBAC-filtered, content-addressed cache via dulwich).
- Admin god-mode dropped on the marketplace surface — admins curate
their own view via grants like everyone else.
- Bare-repo cache materializes per RBAC-filtered ETag; stale entries
not pruned in this iteration (disclaimed in git_backend.py docstring).
== #81#83#44 security/ops hardening ==
- #81 Group A — orchestrator ATTACH allow-listing (extension/url/alias).
- #81 Group B — Keboola extractor 3-state exit codes:
0 success / 1 total fail / 2 PARTIAL fail
Sync API logs PARTIAL FAILURE alert on exit 2. Operators with binary
alerting must teach it the new partial signal.
- #81 Group C — schema v10 view_ownership; rejects silent overwrite
of a prior connector's view name on collision.
- #81 Group D — extractor-side identifier validation.
- #83 — Jira webhook fail-closed when JIRA_WEBHOOK_SECRET unset
+ path-traversal fix.
- #44 — entire /api/scripts/* surface is admin-only (planted-script +
sandbox-bypass risk closed).
== Web UI polish + deploy fix ==
- /admin/access: live grant-count badges (no stale snapshot revert),
shared-header CSS link added to /catalog and /admin/{tables,permissions},
per-resource-type colored stripes.
- docker-compose.host-mount.yml: bind,rbind so dual-disk hosts don't
silently shadow sub-mounts and write state to the wrong disk.
== OSS vendor-neutralization (waves 1+2) ==
- scripts/grpn/ → scripts/ops/. Customer-specific identifiers
(project IDs, internal hostnames, dev/prod VM IPs, brand names)
replaced with placeholders across code, docs, Terraform, Caddyfile,
OAuth probe, and planning docs. Downstream infra repos that copied
scripts/grpn/agnes-tls-rotate.sh or agnes-auto-upgrade.sh must
update the path.
== Translation ==
- src/repositories/user_groups.py::ensure_system docstring translated
from Czech to English for codebase consistency.
Co-authored-by: Mina Rustamyan <mina@keboola.com>
Closes the C1 findings from issue #81 plus the round-3/4 follow-ups
on the read-only query path.
Both _attach_remote_extensions (rebuild path) and
_reattach_remote_extensions (query path) now apply the same hard
allowlists for extensions and token-env names, single-quote-escape
the URL, and split built-in vs community install. The CHANGELOG bullet
documents the full scope including the table_schema → table_catalog
fix that made the rebuild path a silent no-op for every connector.
New module src/orchestrator_security.py centralises the policy. Tests
in tests/test_orchestrator_remote_attach_security.py — 28/28 pass.
Refs #81.
* fix(security): gate Script-API /run on admin role (#44)
The AST + string-blocklist sandbox in `_execute_script` is defense-in-depth,
not a primary trust boundary. It does not block `vars()`, `type()`, or
`__class__.__bases__` introspection chains, and the string blocklist is
trivially evadable via concatenation/dunder encoding. Treat the role gate
as the actual barrier: only admin can run scripts.
- `POST /api/scripts/run` and `POST /api/scripts/{id}/run` now require admin.
- `POST /api/scripts/deploy` stays analyst-accessible (storing != executing).
- Existing /run tests retargeted to admin_token; added regression tests
asserting analyst → 403 on both endpoints.
- CHANGELOG: BREAKING (security) bullet under Unreleased/Changed.
Closes#44.
* fix(security): admin-gate /deploy + harden sandbox blocklist (review #92)
Reviewer of PR #92 flagged three MUST-FIXes that #44 wasn't fully closed:
1. /api/scripts/deploy still accepted analyst → planted-script attack
path (analyst plants malicious source, waits for admin to /run).
Now: /deploy also requires admin; the entire Script API is admin-only.
2. The "Minimum (same-day)" blocklist mitigations from issue #44 weren't
applied. Added the introspection-chain dunders that the issue PoC
pivots through: __subclasses__, __globals__, __class__, __base__,
__bases__, __mro__, __dict__, __code__, __builtins__. Plus `vars`
in BLOCKED_FUNCTIONS. Deliberately NOT adding __init__ /
__getattribute__ (substring match would flag every legit `def __init__`)
nor `type`/`dir` (frequent in legitimate admin scripts). Documented
the trade-off inline.
3. Tests didn't cover the actual PoC payload nor non-analyst non-admin
roles. Added test_run_pwn_payload_blocked parametrized over the issue's
own PoC + two equivalent variants (lambda+__globals__, __mro__
traversal); these stay green only as long as the dunder list does.
test_*_requires_admin tests now parametrize over (analyst, viewer,
km_admin) so all three non-admin core roles are pinned at 403.
Conftest extension: seeded_app now exposes viewer_token and
km_admin_token as siblings to admin_token / analyst_token.
CHANGELOG bullet updated to reflect /deploy gate change and new
internal regression tests. 35/35 scripts tests pass locally.
Refs review of #92.
* fix(tests): test_security TestScriptSandbox needs admin token after #44 hardening
CI failure on PR #92 caught a missed test file. tests/test_security.py
seeded only an analyst user and used the analyst token to drive sandbox
tests. After the #44 admin-gate (deploy + run both admin-only), every
sandbox test got 403 from the role gate before the AST/string check
could run, so 'blocks os.system' / 'blocks eval' / etc. all failed.
Fix: extend the fixture to also seed an admin user and return the admin
token. Sandbox tests now reach the sandbox layer; access-control tests
further down in the module continue to use the analyst that was kept
around. 41/41 test_security.py tests pass locally.
* fix(security): #92 round-3 — gate GET /api/scripts on admin role
Devin Review caught: GET /api/scripts (app/api/scripts.py:44-51) was
left on Depends(get_current_user) when the rest of the API moved to
admin-only. ScriptRepository.list_all() does SELECT * FROM script_registry
which returns ALL columns including 'source' (the full script body).
So any authenticated user (viewer / analyst / km_admin) could read
admin-deployed scripts — leak of code that may contain credentials,
business logic, or admin-only operational details.
CHANGELOG already says 'The entire Script API is now admin-only',
which was true for /deploy, /run, /{id}/run, DELETE — just not for
GET. Now consistent: every Script endpoint requires admin.
Tests:
- New parametrized test_list_scripts_requires_admin over (analyst,
viewer, km_admin) tokens — all assert 403.
- Updated test_list_scripts_empty in both test_scripts_api.py and
test_api_scripts.py to use admin_token.
79 tests pass.
Refs Devin Review of #92.
* fix: cleanup unused imports, stale docstrings, and incomplete CHANGELOG
- Remove unused imports: Path, List, get_current_user (ruff F401)
- Trim docstrings to describe current behavior, not change history
- CHANGELOG now lists GET /api/scripts among admin-gated endpoints
- Remove diff-commenting inline comments from tests
Co-Authored-By: zdenek.srotyr <zdenek.srotyr@keboola.com>
* fix: merge duplicate Changed sections into one per CLAUDE.md convention
Co-Authored-By: zdenek.srotyr <zdenek.srotyr@keboola.com>
---------
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(security): close Jira webhook fail-open + path traversal (#83)
Two related vulnerabilities:
1. Fail-open signature check: when JIRA_WEBHOOK_SECRET was unset,
_verify_signature returned True and any unauthenticated POST to
/webhooks/jira would run the full ingest pipeline. Now fail-closed —
the handler short-circuits with 503 (operator-misconfiguration signal,
distinct from 401 wrong-signature) when the secret is missing.
2. Path traversal via attacker-controlled issue_key: webhook payloads
carry issue.key, which flowed unsanitized into save_issue (issues_dir /
"{issue_key}.json"), download_attachment (attachments_dir / issue_key),
and incremental_transform (raw_dir / "issues" / "{issue_key}.json"). A
crafted webhook with issue.key="../../etc/passwd" could write outside
the Jira data dir.
Defense-in-depth: new connectors/jira/validation.py exposes
is_valid_issue_key (whitelist regex ^[A-Z][A-Z0-9_]{0,31}-\d{1,12}$) and
safe_join_under (Path.resolve() containment check). Both are enforced at
the webhook entry point AND at every filesystem boundary in the connector.
Tests:
- New tests/test_jira_validation.py — unit tests for both helpers
(parametrized invalid keys, traversal/symlink/absolute-path cases).
- Webhook tests: test_unconfigured_secret_returns_503,
test_path_traversal_in_issue_key_rejected (parametrized over 10 bad keys),
test_valid_issue_key_accepted.
CHANGELOG: two CRITICAL Fixed bullets under Unreleased.
Closes#83.
* fix(security): close remaining #83 review findings — webhookEvent traversal, _handle_deletion guard, regex tightening
Reviewer of PR #93 flagged four MUST-FIXes:
1. _log_webhook_event used the attacker-controlled `webhookEvent` field
as a filename component without sanitization. Payload with
`webhookEvent: "../../tmp/pwn"` could escape WEBHOOK_LOG_DIR. Now:
- non-`[A-Za-z0-9_-]` runs are replaced with `_` (dot excluded so
`..` cannot survive sanitization as a directory component)
- length capped at 64 chars
- final path routed through safe_join_under
New regression test `test_webhook_event_path_traversal_sanitized`.
2. _handle_deletion (connectors/jira/service.py:530) and
process_webhook_event (line 487) still used raw issue_key in path
builds. Even though the webhook handler validates upstream, the
"defense-in-depth at every filesystem boundary" claim required these
too. Both now run is_valid_issue_key and safe_join_under guards.
3. Regex `^[A-Z][A-Z0-9_]{0,31}-\d{1,12}$` permitted underscores in
project keys. Atlassian's project-key validator does not — `A_B-1`
is rejected by Jira itself. Tightened to `[A-Z0-9]` and updated
tests: `ABC_DEF-1` is now invalid, added Cyrillic А-1 (lookalike),
CRLF, and oversize cases to the bad-key parametrization.
4. Existing test test_deletion_of_nonexistent_issue_returns_true used
`PROJ-NOEXIST` which is not a real Jira key shape. Updated to
`PROJ-99999`. The test still exercises the same intent (deletion of
issue with no local file is idempotent).
73/73 jira tests pass locally (test_jira_webhooks + test_jira_validation
+ test_jira_service + test_jira_service_full + test_jira_incremental).
CHANGELOG updated to document the regex tightening and the new
webhookEvent sanitization.
Refs review of #93.
* fix(tests): test_journey_jira tests assumed fail-open before #83 fix
CI failure on PR #93 caught two journey tests that pinned the OLD
fail-open contract:
- test_webhook_with_no_secret_configured_accepted asserted 200 when
JIRA_WEBHOOK_SECRET was unset. After the #83 fix that's a 503
(operator misconfig). Renamed to _refused and flipped the assertion.
- test_webhook_empty_payload_rejected didn't set the secret, so the
503 short-circuit fired before the empty-payload 400 could. Set
JIRA_WEBHOOK_SECRET in the patched Config so the test exercises the
intended path.
56/56 jira journey + webhook + validation tests now pass.
* fix(security): #93 round-3 — webhook fallback format + save_issue early validation
Devin Review caught two real findings:
1. Webhook handler regression: the round-2 fix extracted issue_key only
from event_data['issue']['key'], but process_webhook_event has long
supported a fallback 'issue_key' top-level field for certain Jira
event formats (e.g. delete events historically). The handler now
blocks those events with 400 before they reach the service layer.
Fix: mirror process_webhook_event's fallback in the handler — try
issue.key first, fall through to event_data.get('issue_key') when
empty. is_valid_issue_key still validates whichever source provided
the key.
2. save_issue defense-in-depth was incomplete: is_valid_issue_key ran
AFTER fetch_remote_links and fetch_sla_fields had already used the
unvalidated issue_key in HTTP URL construction
({base_url}/issue/{issue_key}/remotelink etc.). A future internal
caller invoking save_issue directly with attacker-controlled input
could trigger outbound requests with a malicious path component
(limited SSRF / URL-path manipulation against the Jira API server).
Fix: move the is_valid_issue_key check to immediately after the
null guard, before any HTTP request or filesystem op. Webhook layer
still validates upstream, this is the second layer.
66 jira tests pass.
Refs Devin Review of #93.
* fix(changelog): #93 round-4 — add BREAKING marker to fail-closed bullet
Devin Review caught: the JIRA_WEBHOOK_SECRET fail-closed change is a
behavior change for operators (response code 503 vs old 200) that
existing alerting may treat differently. Per CLAUDE.md changelog
discipline rule, operators grep for **BREAKING** before bumping the
pin. Added the marker + a short note on what action operators need
to take (set the env var if they haven't).
Refs Devin Review of #93.
* fix: #93 round-5 — null-issue crash + comment drift
Devin Review caught two findings on the round-4 commit:
1. Pre-existing crash on null issue field: a webhook payload with
{"issue": null} (rather than omitting the key) caused
event_data.get("issue", {}) to return None, then issue.get("key")
raised AttributeError → unhandled 500. Pre-existing but reachable.
Fix: 'event_data.get("issue") or {}' normalises None to {}, then
the existing fallback / validation path returns 400 cleanly.
New regression test test_null_issue_field_does_not_crash.
2. Inline comment drift: the comment at line 77 documented the allowed
character class as [A-Za-z0-9._-] (with dot) but the regex at line 27
excludes dot deliberately (so '..' cannot survive sanitization).
Fixed the comment to match.
52 jira tests pass.
Refs Devin Review of #93 round 5.
* fix: #93 round-6 — process_webhook_event also normalises null issue field
Devin Review caught: the webhook handler at app/api/jira_webhooks.py
correctly handles {"issue": null} via 'event_data.get("issue") or {}',
but process_webhook_event at connectors/jira/service.py:509 still
used the bare 'event_data.get("issue", {})' which returns None on
explicit null. Internal callers (anything that invokes
process_webhook_event without going through the HTTP handler) would
hit the same AttributeError the round-5 fix closed at the handler
layer. Same one-line fix.
32 jira tests pass.
Refs Devin Review of #93 round 5.
* fix: #93 round-7 — issue-key regex uses [0-9] not \d
Devin Review caught: Python 3's \d matches any Unicode decimal digit
(Arabic-Indic ٣, Bengali ৩, Devanagari ३, …). A key like TEST-٣ would
pass the regex even though it's not a valid Jira input. Tightened to
[0-9] (ASCII only).
Added three Unicode-digit cases to the bad-key parametrization in
test_jira_validation.py to lock in the contract.
Refs Devin Review of #93 round 6.
* fix: #93 round-8 — use \\Z anchor not $ in issue-key regex
Devin Review caught: Python's $ anchor matches before a trailing \\n,
so re.match('…$', 'TEST-1\\n') returns a match. is_valid_issue_key
returned True for CRLF-injected keys. \\Z is hard end-of-string and
closes that bypass.
Manual verification:
is_valid_issue_key('TEST-1\\n') → False (was True before fix)
is_valid_issue_key('TEST-1\\r\\n') → False
is_valid_issue_key('TEST-1') → True
Refs Devin Review of #93 round 7.
* docs: #93 round-9 — CHANGELOG regex matches implementation
Cuts 0.11.5 with all the [Unreleased] bullets that landed on top of PR #73
between commit a899877 (the original "v0.11.4" tag in the chain) and the
final merge commit on main. No new public-API surface; the user-visible
payoff is that v8→v9-migrated installations work end-to-end (login flows,
GET /api/users, admin nav, the new role-management REST API and its
last-admin protection) and `make local-dev` startup is finally quiet.
Bullets covered (full text in CHANGELOG.md [0.11.5]):
- _hydrate_legacy_role re-resolves from grants on every request — fixes
privilege-retention after grant revoke via the role-management API.
- Dev-bypass + OAuth callback now pass user_id to resolve_internal_roles
so direct grants land in the session cache (not the DB-fallback path).
- GET /api/users hydrates user dicts before Pydantic validation
(HTTP 500 on every migrated install) + same fix for update/delete
paths so last-admin protection triggers on migrated admins.
- Scheduler stopped spamming POST /auth/token 401 — the auto-fetch
fallback was always broken; SCHEDULER_API_TOKEN is now the only path.
- POST /auth/token / Google OAuth / password / email-magic-link all
hydrate user["role"] before issuing the JWT (Pydantic 500 + wrong
token payload). New TestAuthLoginFlowsPostMigration regression class.
- docs/RBAC.md no longer documents the non-existent implies= keyword
on register_internal_role.
- _seed_core_roles now actually runs on every connect (the docstring
was lying — only ran during fresh install + v8→v9). New
TestSeedCoreRolesSafetyNet regression class.
This commit also adds:
- AuthlibDeprecationWarning suppression at app/main.py top — upstream-
internal forward-compat note from authlib._joserfc_helpers, not
actionable on our side. Filter is targeted by class (with a
message-based fallback) so other DeprecationWarnings remain visible.
- pyproject.toml version: 0.11.4 → 0.11.5.
- CHANGELOG.md: [Unreleased] → [0.11.5] — 2026-04-27, new empty
[Unreleased] skeleton appended for the next PR to land on.
Tag v0.11.5 follows; keboola-deploy-v0.11.5 tag triggers the
keboola-deploy.yml workflow for agnes-dev.keboola.com.
* feat(auth): v9 schema — unified role management foundation (WIP)
Tasks 1-5, 10 of the role-management-complete plan. Foundation only,
follow-up commits add REST API, CLI, UI, and tests.
Schema v9:
- user_role_grants table: direct user → internal_role mapping
(complementary to group_mappings). Drives PAT/headless auth and
persists across sessions. Source field tracks 'direct' vs auto-seed.
- internal_roles.implies (JSON): transitive role hierarchy. core.admin
implies core.km_admin → core.analyst → core.viewer. Resolver does BFS
expand at lookup time.
- internal_roles.is_core (BOOL): distinguishes seeded core.* hierarchy
from module-registered roles. UI renders them differently.
- v8→v9 migration: ADD COLUMN, CREATE TABLE, _seed_core_roles +
_backfill_users_role_to_grants, then NULL legacy users.role values.
DuckDB FK constraint blocks DROP COLUMN — sloupec zůstává jako
deprecated artifact (UserRepository ignoruje), fyzický drop deferred.
Resolver:
- Regex extended to allow dotted namespace (core.admin,
context_engineering.admin), max 64 chars total.
- expand_implies(role_keys, conn): BFS over implies JSON column.
- resolve_internal_roles signature gains optional user_id parameter;
unions group-mapping resolution with user_role_grants direct grants
before implies expansion.
require_internal_role:
- Two-path resolution: session cache (OAuth) → DB grants (PAT/headless
fallback). PAT clients now legitimately satisfy gates without the
OAuth round-trip, fixing the v8 limitation where every PAT-callable
admin endpoint needed require_role(Role.ADMIN) instead of
require_internal_role(...).
Backward-compat:
- require_role(Role.X) and require_admin become thin wrappers over
require_internal_role(f"core.{role}"). Implies hierarchy preserves the
legacy "at least this level" semantics automatically — no per-level
comparison code needed.
- src/rbac.py helpers (is_admin, has_role, get_user_role,
set_user_role, can_access_table, get_accessible_tables) all read from
the resolver via _get_internal_role_keys.
- UserRepository.create() and update() now mirror role changes into
user_role_grants via _grant_core_role helper. Preserves API while
making the new table the source of truth.
- UserRepository.delete() pre-deletes user_role_grants rows
(FK cascade — DuckDB doesn't auto-cascade).
- count_admins() reads user_role_grants ⨝ internal_roles instead of the
now-NULL users.role column.
First consumer:
- app/api/admin.py module-level docstring documents the v9 pattern for
future module authors. Existing require_role(Role.ADMIN) callsites
flow through the wrapper; no behavior change for OAuth callers, and
PAT callers gain access via direct grants.
Tests: full suite green (1396 passed, 6 skipped). Existing tests
exercise the new pathway transparently because UserRepository.create
auto-grants. New test_pat_caller_with_direct_grant_passes pins the
PAT-aware contract.
Schema: v9 (was v8). pyproject.toml + CHANGELOG bump deferred to the
final PR-prep commit.
* feat(auth): role management complete — REST API + CLI + UI + docs (v0.11.4)
Sjednocuje legacy users.role enum s v8 internal-roles foundation pod jeden
model s implies hierarchií, dodává admin UI + REST API + CLI pro správu
group mappings i přímých user grants, a dělá require_internal_role
PAT-aware tak, aby admin endpointy fungovaly uniformly napříč OAuth
i headless callery.
REST API (app/api/role_management.py, +496 LOC):
- 8 endpointů pod /api/admin: internal-roles list, group-mappings CRUD,
users/{id}/role-grants CRUD, users/{id}/effective-roles debug.
- Všechny gated require_internal_role("core.admin"). Audit-log na každé
mutaci (role_mapping.created/deleted, role_grant.created/deleted).
- Last-admin protection: refuse to delete the final core.admin grant
(mirrors users.py:count_admins protection).
- Nový UserRoleGrantsRepository v src/repositories/user_role_grants.py.
CLI (cli/commands/admin.py extension, +258 LOC):
- da admin role list / show <key>
- da admin mapping list / create <group-id> <role-key> / delete <id>
- da admin grant-role <email> <role-key>
- da admin revoke-role <email> <role-key>
- da admin effective-roles <email>
- Všechno přes typer + PAT auth, --json flag, response-shape tolerantní.
UI (admin_role_mapping.html + admin_user_detail.html + nav + user list):
- Nová stránka /admin/role-mapping: internal_roles read-only table +
group_mappings table with create/delete forms.
- Nová stránka /admin/users/{id}: core role single-select + capabilities
multi-checkbox + effective-roles debug (direct + group + expanded).
- Existing user list dostává "Detail" link na novou stránku.
- Nav link na /admin/role-mapping.
Tests: +85 nových testů přes 4 nové soubory:
- test_schema_v9_migration.py (8) — fresh install + v8→v9 backfill +
legacy column NULL semantics + unknown-role fallback + invariants.
- test_api_role_management.py (33) — všech 8 endpointů, happy + error
paths, audit-log assertions, last-admin protection.
- test_cli_admin_role.py (25 + 1 conditional) — typer subcommands,
text + json output, PAT integration smoke.
- test_admin_role_mapping_ui.py (9) + test_admin_user_capabilities_ui.py (10)
— page rendering, auth gating, form contracts, JS hooks.
Full suite: 1482 passed, 6 skipped (was 1396 → +86, žádné regrese).
Docs:
- docs/internal-roles.md kompletní rewrite — odstranil "no UI yet",
přidal hierarchy diagram, dual-path resolution, dotted-namespace
convention, admin workflow přes UI/CLI/REST, refresh semantics
for group mappings vs direct grants, migration notes.
- CLAUDE.md schema v8 → v9.
- CHANGELOG.md [0.11.4] s BREAKING marker pro users.role NULL
semantics + complete Added/Changed/Removed/Internal sekce.
- pyproject.toml: 0.11.3 → 0.11.4.
Sequencing: po mergi tohoto PR Pabu rebasuje pabu/local-dev (PR #72)
na main, jeho schema migrations se posouvají z v9/v10/v11 na v10/v11/v12.
Implementation breakdown:
- Sequential (já): foundation tasks — schema v9, resolver, PAT-aware
require_internal_role, backward-compat wrappers, rbac refactor,
UserRepository auto-grant.
- Parallel sub-agents (3 worktrees, ~10 min): REST API, CLI, UI.
- Sequential (já): integrace, docs/CHANGELOG/version, schema tests,
fullsuite verification.
* fix(auth): address Devin review on PR #73 — three regressions
Three concrete bugs caught in Devin's PR review, all fixed in this commit.
1. **users.role hydration on read** (the big one):
v8→v9 migration NULLs users.role for every existing user, but a long
tail of read sites still inspect user["role"] directly:
- app/web/templates/_app_header.html:15 — admin nav gate
- app/web/templates/_app_header.html:36-37 — role badge in dropdown
- app/web/router.py:319-321 — UserInfo.is_admin/is_analyst/is_privileged
- app/web/router.py:489 — corporate memory is_km_admin
- app/api/catalog.py:54 — admin "see all tables" bypass
- app/api/sync.py:215 — admin "see all sync states" bypass
Without a fix, every existing admin loses the entire admin nav (and
API admin bypasses) immediately after upgrade — a serious regression.
Fix: new helper _hydrate_legacy_role() in app/auth/dependencies.py
maps the highest-level core.* grant back into user["role"] as the
legacy enum string. Called from get_current_user() on both auth paths
(LOCAL_DEV_MODE + JWT/PAT). Idempotent — skips when role is already
populated. Net effect: every pre-v9 callsite keeps working transparently
for both OAuth and PAT callers, with one extra DB round-trip per
authenticated request (same cost as the existing PAT-aware
require_internal_role fallback).
3 regression tests in tests/test_schema_v9_migration.py:
- test_hydration_recovers_role_from_user_role_grants
- test_hydration_returns_highest_grant (multi-grant → highest wins)
- test_hydration_falls_back_to_viewer_when_no_grants (safe fallback)
2. **CLI effective-roles TypeError**:
API returns direct/group as List[Dict] (RoleGrantResponse-shaped),
but the CLI did ', '.join(direct) which raises TypeError on dicts.
Tests masked it because mocks used bare string lists. Replaced
raw .join() with a _names() helper that extracts role_key from
each item, falling back to str() for legacy mock shapes.
3. **UI template field-name mismatch**:
admin_user_detail.html JS reads data.groups but the API serializes
the field as group (singular, per EffectiveRolesResponse pydantic).
Currently benign because the API always returns group:[], but the
field would silently disappear once the group-derived view is wired
up. Added data.group as the primary lookup, kept the legacy aliases
for shape-drift tolerance.
Full suite: 1485 passed (was 1482, +3 hydration tests), 6 skipped, no
regressions.
* fix(auth): Devin review #2 + UX self-service + RBAC docs rename
Three threads landed in one commit because they share the same
auth/role surface and CHANGELOG entry.
Devin review #73 second round (2 actionable findings):
- _hydrate_legacy_role no longer short-circuits on truthy users.role.
The role-management endpoints (POST/DELETE /api/admin/users/{id}/
role-grants + the changeCoreRole UI flow) only mutate
user_role_grants — they don't update the legacy column. The early
return trusted that stale value, so a user downgraded via the new
REST/UI kept role="admin" in their dict on subsequent requests,
which fooled _is_admin_user_dict (src/rbac.py) and the catalog/sync
admin-bypass short-circuits into retaining elevated table access
even though require_internal_role correctly denied the API gates.
Always re-resolves now, making user_role_grants the single source
of truth on every authenticated request. Cost: one DB round-trip
per request — same as the existing PAT-aware fallback. Pinned by
test_hydration_ignores_stale_legacy_role_after_grant_revoke.
- Dev-bypass (app/auth/dependencies.py) and OAuth callback
(app/auth/providers/google.py) now pass user_id to
resolve_internal_roles so direct grants land in
session["internal_roles"] alongside group-mapped roles. Pre-fix,
every admin-gated request fell through to the per-request DB
fallback inside require_internal_role and the dev-bypass log line
read "resolved 0 internal role(s)" for an obviously-admin user.
test_session_internal_roles_populated updated to assert union.
User-visible UX (also addresses local-test feedback):
- HTTP 500 on /admin/users post-v8→v9 migration — UserResponse.role
is required str, but legacy users.role was NULL-ed by the
migration. _to_response in app/api/users.py now routes every dict
through _hydrate_legacy_role; same fix lifts the silent no-op of
last-admin protection in update_user/delete_user (the role-equality
short-circuits would skip the count_admins guard for migrated
admins). Three regression tests under TestAPIUsersPostMigration.
- /profile is now a real self-service detail page for *every*
signed-in user (not just admins). Three new server-side sections:
Effective roles (resolver output as chip cloud), Direct grants
(rows in user_role_grants with source label), Roles via groups
(which Cloud Identity / dev group grants which role for the
current user). Non-admins finally see *why* a feature is or isn't
accessible. Admins additionally see a deep-link to
/admin/users/{id} for editing their own grants.
- /admin/role-mapping group-id picker. New "Known groups" panel
above the create form: clickable chips for the calling admin's
own session.google_groups (tagged "your group") merged with
external_group_ids already used in existing mappings (tagged
"already mapped"). Click a chip → fills the form. Empty-state
copy points operators at LOCAL_DEV_GROUPS / Google sign-in
instead of leaving them to guess Cloud Identity opaque IDs from
memory.
Operational fixes:
- Scheduler log-noise: every cron tick produced a
POST /auth/token 401 because the auto-fetch fallback called the
endpoint with just an email (no password) and silently fell
through. Removed the broken path entirely. Operators set
SCHEDULER_API_TOKEN (long-lived PAT) in production; in
LOCAL_DEV_MODE the dev-bypass auto-authenticates the un-tokenized
request, so jobs continue to work.
Docs:
- docs/internal-roles.md → docs/RBAC.md (git mv preserves history).
Standard industry term, more discoverable for engineers grepping
for RBAC in a new repo. Restructured: Quickstart-by-role
(operator / end-user / module author), step-by-step
Module-author workflow with code examples (register key, gate
endpoint, declare implies, write contract test), naming pitfalls,
refresh semantics. CLAUDE.md gets a new
"Extensibility → RBAC" section pointing contributors at the doc
before they add gated endpoints. Cross-refs in app/api/admin.py
+ tests/test_role_resolver.py updated.
Tests: 293 in the auth/role/scheduler/UI test set passed, 0 regressions.
* fix(auth): Devin review #3 — login flows + RBAC docs
Two new findings on commit 7d1c048, both real and addressed.
Finding 1 (BUG, HTTP 500): every auth login flow loaded users via
UserRepository.get_by_email and passed user["role"] straight to
create_access_token, Pydantic response models, and _set_login_cookie
without going through _hydrate_legacy_role. Post-v9 the legacy column
is NULL for migrated users, and TokenResponse.role is a required str —
so POST /auth/token raised ValidationError → HTTP 500 for any v8-admin
trying to log in via password. Same root cause produced non-crashing
but semantically wrong JWTs (role: null) from Google OAuth, password
web flows, and email magic-link verification.
Fix: hydrate inline in every login flow before reading user["role"]:
- app/auth/router.py — POST /auth/token (the crash site)
- app/auth/providers/google.py — OAuth callback (was just stale JWT)
- app/auth/providers/password.py — 5 flows: JSON login, web login,
JSON setup, web reset confirm, web setup confirm
- app/auth/providers/email.py — centralized in _consume_token,
covers both /verify endpoints
New regression class TestAuthLoginFlowsPostMigration pins both the
no-crash and the correct-role contracts for all four legacy levels
(viewer/analyst/km_admin/admin) on POST /auth/token.
Finding 2 (DOCS): docs/RBAC.md showed register_internal_role() being
called with implies=[...], but the function signature is (key, *,
display_name, description, owner_module). A module author copying the
example would TypeError at import time. The implies field on
internal_roles IS honored at runtime by expand_implies, but the
registry-side write path (register_internal_role + InternalRoleSpec +
sync_registered_roles_to_db) doesn't exist yet — implies is currently
seeded only for the core.* hierarchy via _seed_core_roles in src/db.py.
Rewrote the Implies hierarchy and Module-author workflow sections to
document what's actually supported in 0.11.4 and what a future change
would need to add. The "for cross-module hierarchies, register each
level + grant both" pattern works today.
Tests: 322 in the auth/role/scheduler/UI/password test set passed,
0 regressions.
* fix(db): _seed_core_roles actually runs on every connect (Devin review #4)
Devin flagged that the docstring on `_seed_core_roles` promised per-connect
execution as a safety net for accidental DELETEs and in-code seed changes,
but the only call sites lived inside `if current < SCHEMA_VERSION:` — so
once a DB was on v9 the function never ran again, and the docstring lied.
Picked option (b) from the review (actually call it on every startup) over
option (a) (fix the docstring) because the safety net is genuinely useful:
- recovery from accidental admin DELETE on internal_roles,
- in-code _CORE_ROLES_SEED tweaks (display_name/description/implies)
ship without a manual SQL deploy,
- fresh installs and migrations stop needing their own seed call sites.
Tail call gated by `get_schema_version(conn) <= SCHEMA_VERSION` so the
future-version-is-noop rollback contract still holds — a v9 binary won't
touch a DB that's been upgraded past v9.
Test coverage: new TestSeedCoreRolesSafetyNet class (3 tests) pins the
three contracts — deleted row re-seeds, mutated display_name re-syncs
from in-code seed, applied_at on schema_version doesn't churn on
already-current DBs. Existing TestMigrationSafety::test_future_version_is_noop
still passes (verified against the gating logic).
* feat(auth): internal roles + external→internal group mapping (foundation)
Two-layer authorization model: external Cloud Identity groups (org-managed)
get mapped onto internal Agnes-defined capabilities (app-managed) via an
admin-curated many-to-many table. Per-request permission checks read off
the session — no DB hit. Refresh requires re-login.
Schema v8 — new tables:
- internal_roles (id, key UNIQUE, display_name, description, owner_module, …)
— app-defined capabilities like 'context_admin'. Modules self-register at
import; the startup hook syncs the registry into this table (idempotent).
- group_mappings (id, external_group_id, internal_role_id FK, …)
— admin-managed bindings, UNIQUE(external_group_id, internal_role_id).
app/auth/role_resolver.py — new module:
- register_internal_role(key, display_name, description, owner_module)
Module-author entry point. lower_snake_case key, immutable, validated.
Same key + same fields = no-op (re-import safe); same key + different
fields = ValueError so two modules can't silently overwrite each other.
- sync_registered_roles_to_db(conn) — startup reconciliation. Inserts new
keys, updates drifted metadata, never deletes (preserves mappings).
- resolve_internal_roles(external_groups, conn) — joins group_mappings.
Sorted, deduplicated role-key list. Plugged into google_callback +
dev-bypass branch in get_current_user.
- require_internal_role('key') — FastAPI dependency factory; reads
session.internal_roles; 403 with explicit message when missing.
Resolution runs at sign-in only (Google callback + LOCAL_DEV_GROUPS change
in dev-bypass) — same semantics as session.google_groups. No admin UI yet;
mappings created via repository directly until follow-up PR ships UI.
21 new tests in tests/test_role_resolver.py: register/list, idempotency,
collision detection, key-format validation; sync insert/update/no-delete;
resolve empty/single/many-to-many/malformed-input; e2e via
LOCAL_DEV_GROUPS — gated endpoint allowed/denied + direct session-cookie
inspection. Full sweep: 178/178 passed across auth + db + repo tests.
(Two pre-existing test_catalog_export.py failures verified unrelated.)
* fix(auth): polish review feedback — first-request dev populate + PAT doc
Two follow-ups from a code-reviewer pass on the foundation commit before
opening the PR:
- Dev-bypass populates session["internal_roles"] on the first request
after sign-in, not just when external groups change. The previous
guard only resolved when groups_changed=True, which left a hole for
the LOCAL_DEV_GROUPS=`""` (explicit empty) flow: target=[],
current=None, neither write branch fires, internal_roles stays
unset, and require_internal_role then 403s with no roles to check
against. The OAuth callback writes session["internal_roles"]
unconditionally on sign-in (even []); dev-bypass now matches that
semantics. Adds a single-pass populate gated on the key being
absent from the session, so subsequent same-state requests still
no-op (cheap session lookup, no resolver call).
- Document that internal roles are session-scoped and PAT/headless
clients will get 403 from any require_internal_role(...) endpoint.
Same constraint already applies to session.google_groups (PAT JWTs
deliberately don't snapshot group memberships — they could change
after issuance with no way to re-sign), but the doc didn't surface
this — an operator pointing a CLI at a role-gated endpoint would
see 403 with no clue why. New "PAT and headless requests" section
spells out the constraint, the rationale, and the three escape
valves (use users.role for the gate; route through OAuth; wait for
the planned `da admin grant-role` CLI helper).
54 auth tests still pass locally (21 role-resolver + 33 existing
auth-provider).
* release(0.11.3): cut release for the internal-roles foundation
Bumps pyproject.toml 0.11.2 → 0.11.3 and renames CHANGELOG's
[Unreleased] section to [0.11.3] — 2026-04-26 (with a fresh
empty [Unreleased] skeleton appended). Adds the matching
[0.11.3] link reference at the bottom of CHANGELOG so the
section heading renders as a hyperlink to the GitHub release
page once the tag lands.
The bullet itself is unchanged content; the rephrasing of
"dev-bypass when external groups change" → "dev-bypass —
populates on first request and whenever external groups
change, mirroring the OAuth callback's always-write
semantics" reflects the polish committed in d590579, plus
the appended PAT/headless caveat pointing at the doc
section that landed in the same polish pass.
* fix(auth): address review feedback from Pavel — PAT-specific 403, audit logs, hardening
Round-2 polish over the internal-roles foundation, addressing Pavel's review
on PR #71. No behavior change for the happy path; tightens the safety rails
and makes the failure modes self-explanatory.
User-visible:
- require_internal_role now distinguishes "no session" (Bearer/PAT caller)
from "signed in but missing role" and surfaces a PAT-specific 403 detail
in the first case ("This endpoint needs an interactive (OAuth) session
— Bearer/PAT tokens do not carry session-resolved roles by design").
- docs/internal-roles.md documents deactivate+reactivate as the supported
"force re-resolve now" lever for users that can't be made to log out.
Internal hardening:
- INFO-level audit log on every successful resolve (OAuth callback +
dev-bypass) so a wrong-role complaint is debuggable from the log alone.
- Startup warning when SESSION_SECRET is shorter than 32 chars, matching
the existing JWT_SECRET_KEY gate — both HMAC surfaces sign trust-laden
state (session.internal_roles, session.google_groups, JWTs).
- _clear_registry_for_tests() now refuses to run unless TESTING=1 so a
stray import path in production can't drop the registered capabilities.
Tests:
- 4 new tests in tests/test_role_resolver.py covering: stale-session
contract after a mid-session mapping revoke (pin the documented
limitation), PAT 403 detail wording, OAuth pipeline data flow from
external groups to internal_roles, and the dev-bypass empty-list
fallback when the resolver raises.
CHANGELOG.md updated under [0.11.3] (### Changed + ### Internal).
CLAUDE.md schema doc bumped from v7 to v8.
---------
Co-authored-by: Claude <noreply@anthropic.com>
* feat(auth): mock session.google_groups in LOCAL_DEV_MODE via LOCAL_DEV_GROUPS
LOCAL_DEV_MODE auto-logged-in the dev user but left session.google_groups
empty, so group-aware UI/code paths can't be exercised on localhost without
a real Google OAuth round-trip. New LOCAL_DEV_GROUPS env var (JSON array
matching the production {id, name} shape) populates the session on every
dev-bypass request — same structure the OAuth callback writes, so mock and
prod stay in lockstep. Compare-then-write avoids spurious Set-Cookie noise
on PAT/CLI requests; malformed input falls back to [] with a WARNING so
the dev mock never breaks the dev flow.
* refactor(auth): fail-fast LOCAL_DEV_GROUPS at startup + cache + no-mutate
Three small follow-ups on the same dev-mock vector before merge:
- Validate LOCAL_DEV_GROUPS at app startup and report the parsed group IDs
in the LOCAL_DEV_MODE banner. A malformed value now warns loudly at boot
instead of silently logging on the first authenticated request, where
it's easy to miss.
- Cache the parsed result single-slot, keyed by the raw env-string. Avoids
re-parsing JSON on every authenticated request without test-isolation
surprises — when the env value changes, the key changes and the cache
transparently rebuilds.
- Stop mutating the parsed-input dicts (item.setdefault → spread-merge)
so the cached list stays a fresh value on every rebuild.
- Replace the try/except guard around request.session with hasattr —
SessionMiddleware is always registered, the silent except was paranoid.
Tests grow by a direct session-cookie inspection (decoupled from the
profile template) and three startup-banner log assertions.
* fix(auth): drop fragile session-decoder test + actually skip empty-target write
Two follow-ups on the LOCAL_DEV_GROUPS feature before merge:
- Drop test_session_holds_mocked_groups_directly. It manually decoded the
signed session cookie via TimestampSigner + base64, hardcoding both the
Starlette session-cookie format and the 14-day max_age. Starlette has
changed its session encoding before (URLSafeTimedSerializer pre-0.20)
and would do so again silently — the test would fail with a cryptic
BadSignature, not a clear "mock is broken" signal. The remaining
test_dev_user_sees_mocked_groups_on_profile already covers the same
observable signal (mocked groups in /profile body) without coupling to
Starlette internals.
- Actually skip the session write when target_groups is empty. The previous
comment claimed compare-then-write avoided spurious Set-Cookie noise on
PAT/CLI requests, but on those requests session.get("google_groups") is
None and target is [], so None != [] always evaluates True and the write
fired anyway, marking the session dirty and re-issuing Set-Cookie on
every request. Adding `target_groups and ...` to the guard makes the
comment honest: empty mock now genuinely no-ops, stable browser sessions
still skip via value-equality, and the only remaining write is the one
that actually changes state.
33 auth tests still pass locally.
* fix(auth): match production's always-write semantics for stale dev groups
Devin code-review finding on PR #70: my earlier `target_groups and ...`
short-circuit silently diverged from the production OAuth callback. In
app/auth/providers/google.py:189-194 the callback always writes
session.google_groups on each login — including [] on failure or empty
token — so the session always reflects authoritative current state. The
mock should match.
Failure mode the previous guard left open: a developer sets
LOCAL_DEV_GROUPS=[{...}] for a session, the groups land in the signed
cookie, then the developer unsets the env var and reloads. target → [],
session.get → [{...}], `if target_groups and ...` is False, no write,
stale groups stay in the browser session indefinitely. Mock now lies
about state until logout.
Fix splits the guard:
- target_groups truthy + value-changed → write the new mock (existing path)
- target_groups falsy + non-empty stored → write [] to clear stale state
- otherwise no-op (target [] + stored None/[]: no transition to record)
PAT/CLI requests with no prior session still take the no-op path
(target=[], session.get → None which is falsy), so the original goal of
suppressing spurious Set-Cookie noise on token traffic is preserved.
Tests already cover the populated and unset paths; the new clear-stale
branch is correct by construction (production has the same shape) and
the rare manual reset workflow.
* release(0.11.2): default mocked groups in make local-dev + docs/local-development.md
Cuts 0.11.2 around the LOCAL_DEV_GROUPS work plus a small dev-experience
follow-up: every `make local-dev` now boots with two sensible default
mocked groups (Local Dev Engineers + Local Dev Admins on example.com),
so /profile and group-aware code paths render something realistic
without the operator having to discover and set LOCAL_DEV_GROUPS.
Layered so the default lives in the workflow, not the contract:
- scripts/run-local-dev.sh seeds LOCAL_DEV_GROUPS via shell ":="
syntax — only sets the var when the operator hasn't already.
Override: LOCAL_DEV_GROUPS='[...]' make local-dev. Disable:
LOCAL_DEV_GROUPS= make local-dev.
- docker-compose.local-dev.yml swaps the commented JSON example for
a bare `- LOCAL_DEV_GROUPS` passthrough — the value comes from the
shell, the compose file just propagates it. Operators running
`docker compose up` directly without the wrapper script get an
empty mock (correct: they didn't opt into the make-driven defaults).
- Makefile help line mentions the mocked groups so the behavior is
visible without grepping.
New docs/local-development.md consolidates dev-onboarding instructions
that were previously scattered across docker-compose.local-dev.yml
inline comments, docs/auth-groups.md "Local-dev mock" section, the
Makefile help text, and CLAUDE.md "First-Time Setup". Single page now
covers TL;DR, what LOCAL_DEV_MODE actually bypasses, group mocking
controls + verification, what is *not* mocked (Cloud Identity, real
OAuth, admin Workspace permissions), and the safety rails that keep
the dev shortcuts off production.
Version bump 0.11.1 → 0.11.2 in pyproject.toml, CHANGELOG cuts
[Unreleased] → [0.11.2] — 2026-04-26 with a fresh empty [Unreleased]
skeleton.
* fix(local-dev): default LOCAL_DEV_GROUPS truncated by shell parameter expansion
Reported by an operator running `make local-dev` against the freshly
released 0.11.2 — the LOCAL_DEV_MODE banner showed:
LOCAL_DEV_GROUPS is not valid JSON, ignoring:
Expecting ',' delimiter: line 1 column 70 (char 69)
LOCAL_DEV_GROUPS is set but produced no valid groups —
check the WARNING above for the parse error.
Cause: the default value lived inside `${LOCAL_DEV_GROUPS:=…}` parameter
expansion. Bash matches `}` to close the expansion at the *first* `}`
encountered in the body, regardless of context — even one inside a
nested JSON object literal. The two-element JSON array was therefore
truncated to the first group's closing brace, leaving an unparseable
fragment:
[{"id":"local-dev-engineers@example.com","name":"Local Dev Engineers"
There is no escaping syntax for `}` inside parameter expansion (the
backslash escapes I had only escaped the quotes — `}` reaches bash
literally). Fix: hold the default in a single-quoted variable and
reference it through `${LOCAL_DEV_GROUPS:-$DEFAULT_LOCAL_DEV_GROUPS}`.
The variable's value is opaque to the expansion — no `}` matching
inside it — so the JSON survives intact. Verified with `python -m json`:
parsed OK: 2 groups: ['local-dev-engineers@example.com',
'local-dev-admins@example.com']
Operators on a running 0.11.2 stack: `make local-dev-down && make
local-dev` to pick up the corrected default.
* fix(local-dev): respect LOCAL_DEV_GROUPS= disable path + add 0.11.2 changelog link
Two follow-ups from a Devin code-review pass on PR #70:
- run-local-dev.sh: switch ${LOCAL_DEV_GROUPS:-$DEFAULT} to
${LOCAL_DEV_GROUPS-$DEFAULT} (no leading colon). The :- form
substitutes the default when the variable is unset OR set-but-empty,
silently overwriting the documented disable knob. Three places
promise this works — docs/local-development.md, the CHANGELOG entry,
and the script's own comment — so the bug was an operator-facing
lie, not just an implementation detail. The bare - form only
substitutes on unset, so `LOCAL_DEV_GROUPS= make local-dev` now
reaches the Python parser as "" and short-circuits to []. Verified
with both empty and unset shells.
- CHANGELOG.md: add the [0.11.2] link reference at the bottom.
Keep-a-Changelog convention is to mirror every version heading
with a release-tag link in the footer; the 0.11.2 heading was
missing its counterpart, breaking the Markdown link rendering on
GitHub.
---------
Co-authored-by: Claude <noreply@anthropic.com>
* feat(auth): display Google Workspace groups on /profile
- Request cloud-identity.groups.readonly scope in Google OAuth
- Fetch groups via Cloud Identity API after callback; tolerate 4xx
(non-Workspace tenants) and network errors — never break login
- Store result in Starlette session as google_groups
- Replace /profile redirect with a real profile page rendering
account details (email, name, role) and the group list; show a
friendly empty state when no groups are available
- Tests: helper parsing + 403 + exception paths; profile page
smoke test; updated the old redirect test
* test: remove stale /profile redirect tests
Cherry-pick of Zdeněk's 4f7e4cd ("display Google Workspace groups on
/profile") replaces the /profile redirect with a real profile page —
but only updated one of three tests that expected the old behaviour.
These two tests in test_admin_tokens_ui.py and test_pat.py were left
asserting `/profile → 302 /tokens`, which now returns
`/profile → 302 /login?next=%2Fprofile` for unauth users (the standard
auth guard) or `/profile → 200 HTML` for authenticated users.
Removed both rather than patched — coverage for the new behaviour
already exists in tests/test_auth_providers.py (added by the same
commit). The /tokens render assertions in the deleted test_pat.py case
are redundant with test_admin_tokens_ui.py's own /tokens UI tests.
* fix(auth): Google groups search query needs parent + labels predicates
Cloud Identity Groups Search API returns 400 INVALID_ARGUMENT when the
CEL query lacks the required `parent == 'customers/<id>'` predicate AND
a `'<label>' in labels` membership predicate. Zdeněk's original 4f7e4cd
query had only `member_key_id == '<email>'` — every fetch silently
returned [] and the /profile groups list was always empty.
Fix: build the query with all three required pieces:
parent == 'customers/my_customer' (alias = caller's own Workspace
org; no need to look up customer ID)
member_key_id == '<email>' (filter to this user's memberships)
'cloudidentity.googleapis.com/groups.discussion_forum' in labels
(Workspace mailing-list groups —
the common case; security-group
coverage is a follow-up)
Also: log the full error body (not truncated to 200 chars) and the
query string so the next time Google rejects something we can diagnose
in one log line instead of a re-deploy.
Caught when first agnes-dev login completed normally (HTTP 302) but app
log showed `Google groups fetch returned 400 for petr@keboola.com:
{"error":{"code":400,"message":"Request contains an invalid argument."}}`
on the same VM (kids-ai-data-analysis / agnes-dev.keboola.com).
Reference: https://cloud.google.com/identity/docs/reference/rest/v1/groups/search
* feat(web): add Profile link to user dropdown menu
The /profile page (Zdeněk's 4f7e4cd cherry-pick) renders a real profile
view including Google Workspace groups, but had no entry point in the
UI — users could only reach it by typing the URL manually. Add a
"Profile" menu item between the user header (email + role) and
"My tokens" so the page is discoverable.
Side effect: cleaned up the leftover `or _path.startswith('/profile')`
condition on the "My tokens" active class, which dated from the old
/profile → /tokens redirect (removed in c789617). Now each menu item
owns its own active state.
* fix: profile-link tests + .env quoting for CADDY_TLS
Two issues caught by Keboola's first agnes-dev deploy + agnes-auto-upgrade
cron run:
1. tests/test_web_ui.py — two negative assertions ("href=/profile" NOT in
body) date from when /profile was a redirect-only stub. Now /profile
is a real page (groups display) AND has a dropdown menu link, so the
negative assertions flip to positive. Same for ">Profile<" text in
the non-admin nav test.
2. startup-script.sh.tpl — CADDY_TLS line must be QUOTED in .env, because
agnes-auto-upgrade.sh sources .env via `set -a; . .env; set +a` and
bash treats `KEY=value with spaces` as `KEY=value` followed by `with`
and `spaces` exec attempts. Symptom: cron log spam
`/opt/agnes/.env: line 14: petr@keboola.com: command not found`,
the cron exits non-zero, and no auto-upgrade ever happens. Caddy
itself reads the value fine because docker-compose env_file=.env
parses key=value properly without shell-evaluating the rest.
Fix: emit `CADDY_TLS="tls <email>"` instead of `CADDY_TLS=tls <email>`.
Both the cron source and docker-compose env_file accept the quoted
form; cron stops failing.
* fix(auth): use searchTransitiveGroups + security label for non-admin user
Three bugs in the original cherry-pick + my prior fix attempt, all caught
by a stdlib probe script (scripts/debug/probe_google_groups.py) run
locally with a Playground-issued OAuth token:
1. Wrong endpoint. `groups:search` is the admin "find groups in org"
endpoint and 400s for non-admin users regardless of query. Switched
to `groups/-/memberships:searchTransitiveGroups` which is the
user-perspective "what groups am I in" endpoint.
2. Wrong label. Querying with `cloudidentity.googleapis.com/groups.discussion_forum`
returns 403 "Insufficient permissions to retrieve memberships" even
on the new endpoint — Workspace policy denies non-admin reads of
discussion-forum groups. Switching to `groups.security` returns 200
with the actual membership list. Empirically every Workspace group
at Keboola carries BOTH labels, so the security filter sees the full
set anyway. Confirmed with the probe script.
3. Wrong response shape. `searchTransitiveGroups` returns
{"memberships": [...]}, not {"groups": [...]}. Parser updated
accordingly.
Also adds scripts/debug/probe_google_groups.py — stdlib-only standalone
probe that hits 6 candidate endpoints with a user OAuth token. Saved a
deploy cycle (~10 min) per query iteration; future API-syntax debugging
should start there.
Verified end-to-end: petr@keboola.com login on agnes-dev returns 5
groups (LIC-1PASSWORD, ROLE_ATLASSIAN_*, etc.) via the probe; once
deployed, the same will populate session["google_groups"] and render
on /profile.
* test(auth): update Google groups parser fixture to match searchTransitiveGroups shape
Mock payload was `{"groups": [...]}` (the shape `groups:search` returns).
After switching to `groups/-/memberships:searchTransitiveGroups` in the
prior commit, the actual response is `{"memberships": [...]}` and the
parser iterates that key. Test now mirrors the real shape.
The per-item structure (groupKey.id + displayName) is unchanged, so the
expected output dict stays the same: [{"id": "...", "name": "..."}].
* docs(auth): add docs/auth-groups.md — Google Workspace groups runbook
Captures the non-obvious bits: the GCP-side setup checklist (Cloud
Identity API + scope on consent screen + Internal user type), the
`security` vs `discussion_forum` label trap (the latter 403s for
non-admins, the former 200s — one of those is a 4-iteration debug
session and shouldn't have to be repeated), where groups are stored
(session, not DB) and how to refresh (re-login), plus how to use the
probe script for future API-syntax issues.
Deliberately stops short of explaining "what is Cloud Identity" or
"what is OAuth scope" — those belong in Google's own docs, not ours.
* docs(claude): document release workflows + module versioning + recreate trick
New "Release & deploy workflows" section in CLAUDE.md covers what didn't
exist anywhere in the repo before:
- Distinction between release.yml (auto-build per push) vs the new
keboola-deploy.yml (tag-triggered, explicit deploy only) — plus when
to use which (per-developer convenience vs shared dev VM safety)
- Module versioning (infra-vX.Y.Z) and the bump-after-merge dance
- The lifecycle.ignore_changes [metadata_startup_script] gotcha and how
to force a recreate via workflow_dispatch's recreate_targets input
All generic — no customer hostnames, project IDs, IPs. Customer-specific
deploy steps belong in the consuming infra repo's README.
Also: cross-reference docs/auth-groups.md from the Authentication
section so future Claude sessions find the Workspace-groups runbook
without grepping.
---------
Co-authored-by: ZdenekSrotyr <zdenek.srotyr@keboola.com>
