1e87354d7e
* feat(home): Getting Started + Overview + Usage modes sections
Three new content cards rendered between the install-hero and the
existing connector tiles on /home. Order: Getting Started → Overview
→ Usage modes → connectors.
- Getting Started — dismissible card with two clickable rows linking
to /setup (install flow) and /setup-advanced (deeper reference).
Subsumes the legacy `.advanced-pointer` row that sat above the news
section. Per-device dismiss via a generic localStorage handler:
`.home-card-close[data-dismiss-key="..."]` inside a <section> wires
itself up at page load — drop in any future dismissible card without
per-card JS.
- Overview — operator-owned HTML body sourced from the new
`instance.overview` yaml field (env override
`AGNES_INSTANCE_OVERVIEW`). HTML in, HTML out via the same `| safe`
filter as news_intro. Empty default hides the section entirely,
keeping the OSS vendor-neutral; operators paste their product
framing / privacy posture into instance.yaml. New helper
`get_instance_overview()` in app/instance_config.py mirrors
`get_instance_logo_svg()`.
- Usage modes — three OSS-shipped tiles (Terminal / VS Code / Claude
Desktop · claude.ai) explaining each surface and linking to the
matching /setup-advanced anchors. Closes the gap for users
wondering "where do I actually run this".
Supporting changes:
- setup_advanced.html gains a new `#claude-app` section between
#vscode and #workspace, anchored by the Usage modes Claude Desktop
tile. Covers the marketplace registration paths and when to prefer
the terminal. Added to the table of contents.
- Three new tests in test_web_home_page.py pin the Getting Started
card markup, the Overview-on-when-yaml-set path, and the
Overview-off-by-default path. All 13 tests in the file pass.
Operator follow-up (separate infra PR — NOT this PR): paste the
Foundry-specific Overview body into instance.yaml's
`instance.overview` field. OSS ships with an empty default.
* fix(home): Overview is operator-owned content — drop dismiss button
Earlier iteration added a close X to the Overview section to match
the Getting Started card's dismiss UX. Wrong call: Overview is
operator-authored reference content (privacy posture, telemetry
policy, project framing) and a per-device localStorage hide means
returning users who want to re-read the policy can't recover it
without clearing storage.
Reverts the close button + the data-dismiss-key on the Overview
section. Test inverted to assert the dismiss key is absent (defends
against a future drive-by adding it back). Getting Started still
dismisses — that's procedural getting-started content users
legitimately stop needing once they've finished setup. Overview is
always reachable; whole section is still opt-in at the operator
level via the empty-yaml default.
* fix(home): Terminal usage-mode tile is informational (no click-through)
The setup hero above /home's Usage modes already walks the user
through the Claude Code CLI install — the Terminal tile click-through
to /setup just round-trips back to content the user already scrolled
past. Switch Terminal to a non-anchor <div> and scope the hover
affordance to a.home-usage-item so VS Code + Claude Desktop tiles
keep their click-through (those legitimately deep-link into
/setup-advanced anchors).
* fix(home): point Usage modes guidance at ~/{workspace}/Projects/ subfolder
The bundled plugin scopes the session-analysis loop and the
central-catalog sync to ~/<workspace>/Projects/, not the workspace
root itself — that convention already appears in the install hero's
Step 4 manual-fallback note ('Don't create ~/<workspace>/Projects/
manually — the bundled plugin offers to set it up after install').
Usage modes' footer guidance now matches: 'create every project
under ~/<workspace>/Projects/'. Also calls out that the
session-analysis loop is scoped to that root so users understand
why working outside the workspace dir is invisible to the platform.
494 lines
24 KiB
Text
494 lines
24 KiB
Text
# AI Data Analyst - Instance Configuration
|
||
# ==========================================
|
||
# This is the main configuration file for your instance.
|
||
# Copy to instance.yaml and fill in your values.
|
||
#
|
||
# SECRET VALUES use ${ENV_VAR} syntax - actual values go in .env file.
|
||
# Non-secret values are set directly here.
|
||
|
||
# --- Config version ---
|
||
# Incremented when the config schema changes. Must match SUPPORTED_CONFIG_VERSIONS
|
||
# in config/loader.py. Currently only version 1 is supported.
|
||
config_version: 1
|
||
|
||
# --- Instance branding ---
|
||
instance:
|
||
name: "AI Data Analyst"
|
||
subtitle: "Your Organization"
|
||
copyright: "Your Organization"
|
||
# brand: "Agnes" # Product-name brand string used in analyst-facing UI copy
|
||
# (/home hero, /setup, /login messages, clipboard setup
|
||
# script). Default "Agnes". Set to e.g. "Foundry AI" to
|
||
# rebrand without forking. Distinct from `name` above:
|
||
# `name` is the deploying organization, `brand` is the
|
||
# product. Env override: AGNES_INSTANCE_BRAND.
|
||
# workspace_dir: "FoundryAI" # Filesystem-safe folder name for the analyst's local
|
||
# workspace (~/<workspace_dir>). When unset, derived from
|
||
# `brand` by stripping non-alphanumerics ("Foundry AI" ->
|
||
# "FoundryAI"). Set explicitly only if you want a folder
|
||
# name that differs from the auto-derivation. Env override:
|
||
# AGNES_WORKSPACE_DIR_NAME.
|
||
# logo_svg: | # Inline <svg> element rendered into the header brand slot.
|
||
# <svg width="120" height="30" viewBox="0 0 100 30" xmlns="http://www.w3.org/2000/svg">
|
||
# <text y="22" font-size="24" fill="#333">Logo</text>
|
||
# </svg>
|
||
# # When set, the SVG replaces the text brand in the header.
|
||
# # `name` above still drives browser <title> text and page
|
||
# # headings — keep it populated. Env override:
|
||
# # AGNES_INSTANCE_LOGO_SVG.
|
||
# overview: | # Operator-authored Overview body rendered in the new
|
||
# <p>Free-form HTML — paragraphs, links, lists.</p>
|
||
# # Overview section on /home (between Getting Started and
|
||
# # Usage modes). Use for product framing, privacy posture,
|
||
# # what-data-flows summary — operator-specific copy stays
|
||
# # out of the OSS this way. HTML in, HTML out (same `| safe`
|
||
# # filter as news_intro). Empty/unset = section hidden.
|
||
# # Env override: AGNES_INSTANCE_OVERVIEW.
|
||
# sync_interval: "1 hour" # Cadence shown in analyst CLAUDE.md (e.g., "1 hour", "30 minutes", "daily")
|
||
# admin_email: "ops@acme.com" # Operator contact shown on /home GWS connector tile as
|
||
# an "Email admin" mailto button (analysts whose operator
|
||
# hasn't pre-provisioned a shared OAuth app can request
|
||
# one without leaving the workspace). Empty/unset hides
|
||
# the button. Env override: AGNES_INSTANCE_ADMIN_EMAIL.
|
||
|
||
# --- Server ---
|
||
server:
|
||
hostname: "" # DNS name (e.g., "data.acme.com")
|
||
host: "" # IP address
|
||
app_dir: "/opt/data-analyst" # Installation directory
|
||
# --- Client setup (shown in "Get Started" on dashboard) ---
|
||
# ssh_alias: "data-analyst" # SSH config Host alias for analysts (default: "data-analyst")
|
||
# ssh_key: "~/.ssh/data_analyst_server" # SSH key path for analysts (default: "~/.ssh/data_analyst_server")
|
||
# project_dir: "data-analyst" # Local project folder name (default: "data-analyst")
|
||
|
||
# --- Admin users ---
|
||
# Manage the server, own data files, get unlimited resource limits.
|
||
# SSH keys are used by server/setup.sh during provisioning.
|
||
admins:
|
||
- username: "admin"
|
||
ssh_public_key: "ssh-ed25519 AAAA..."
|
||
|
||
# --- Deployment ---
|
||
deployment:
|
||
method: "manual" # manual | github_actions
|
||
repo_url: "" # e.g., "git@github.com:acme/ai-data-analyst.git"
|
||
branch: "main"
|
||
|
||
# --- Authentication ---
|
||
# At minimum, set allowed_domain and webapp_secret_key.
|
||
# Email magic link auth works out of the box (no external service needed).
|
||
# Google OAuth is optional - add credentials to enable it.
|
||
auth:
|
||
allowed_domain: "" # Email domain(s) for login, comma-separated (e.g., "acme.com" or "acme.com, partner.org")
|
||
webapp_secret_key: "${WEBAPP_SECRET_KEY}"
|
||
# Optional: Google OAuth (if not set, only email magic link is available)
|
||
google_client_id: "${GOOGLE_CLIENT_ID}"
|
||
google_client_secret: "${GOOGLE_CLIENT_SECRET}"
|
||
|
||
# --- Webapp username shaping ---
|
||
#
|
||
# By default, a user's OS account is derived from their full email:
|
||
# e.psimecek@acme.com -> e_psimecek_acme_com
|
||
#
|
||
# Two options let you control this:
|
||
#
|
||
# username_strip_domain: true
|
||
# Use only the local part of the email (before @).
|
||
# Safe when allowed_domain ensures all users share a single domain.
|
||
# e.psimecek@acme.com -> e_psimecek
|
||
# Keeps usernames short and readable.
|
||
#
|
||
# username_prefix: "myapp_"
|
||
# Prepend a fixed string to every webapp-created account name.
|
||
# Necessary when an external identity system (GCP OS Login, LDAP, SAML)
|
||
# already creates OS accounts in /home/ using the same naming scheme.
|
||
# Without a prefix, the webapp sees those existing OS accounts and refuses
|
||
# to register new analyst accounts ("already in use by a system account").
|
||
# With prefix "myapp_" and strip_domain true:
|
||
# e.psimecek@acme.com -> myapp_e_psimecek
|
||
# Linux enforces a 32-character username limit. Keep the prefix short.
|
||
# Changing or removing either option later will invalidate all existing
|
||
# analyst accounts. Use username_mapping (top-level) to bridge legacy accounts.
|
||
#
|
||
# username_strip_domain: false
|
||
# username_prefix: ""
|
||
# disabled_providers: # Hide auth methods from login page
|
||
# - "email" # Disable email magic link (use when Google OAuth is configured)
|
||
|
||
# --- Theme (optional) ---
|
||
# Customize colors, fonts, and shape to match your brand.
|
||
# All values are optional - defaults provide a clean blue theme.
|
||
# See docs/theme-reference.html for a visual guide.
|
||
theme:
|
||
# primary: "#0073D1" # Main brand color (buttons, links, accents)
|
||
# primary_dark: "#005BA3" # Hover/active state of primary
|
||
# primary_light: "rgba(0, 115, 209, 0.1)" # Light tint backgrounds
|
||
# text_primary: "#1A253C" # Main text color
|
||
# text_secondary: "#6B7280" # Muted/secondary text
|
||
# background: "#F5F7FA" # Page background
|
||
# surface: "#FFFFFF" # Card/panel background
|
||
# border: "#E5E7EB" # Borders and dividers
|
||
# font_primary: "'Inter', system-ui, sans-serif"
|
||
# font_url: "https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&display=swap"
|
||
# radius: "6px" # Border radius (cards, buttons, inputs)
|
||
# success: "#10B77F"
|
||
# warning: "#F59F0A"
|
||
# error: "#EA580C"
|
||
|
||
# --- Data source ---
|
||
data_source:
|
||
type: "keboola" # keboola | bigquery | local
|
||
keboola:
|
||
storage_token: "${KEBOOLA_STORAGE_TOKEN}"
|
||
stack_url: "" # e.g., "https://connection.keboola.com"
|
||
project_id: ""
|
||
bigquery:
|
||
project: "${BIGQUERY_PROJECT}" # GCP project hosting the data (used in FROM clause)
|
||
location: "${BIGQUERY_LOCATION}" # BigQuery location (e.g., "us-central1", "US")
|
||
# Uses ADC (Application Default Credentials) - VM service account on GCP
|
||
# Data can live in a different project -- use fully-qualified table IDs in data_description.md
|
||
# billing_project: "prj-billing" # GCP project to bill BQ jobs to / submit jobs from.
|
||
# # Defaults to `project`. Set when the SA has bigquery.data.* on
|
||
# # the data project but lacks serviceusage.services.use there.
|
||
# # Mismatch -> every BQ call 403 USER_PROJECT_DENIED.
|
||
# # `da diagnose` warns when this falls back to `project`.
|
||
# # Configurable via /admin/server-config UI.
|
||
# max_bytes_per_materialize: 10737418240
|
||
# # Cost guardrail (bytes) for query_mode='materialized' BQ scans.
|
||
# # Dry-run check before running; exceeding -> registration / sync
|
||
# # rejected. Default 10 GiB (10737418240). Set 0 to disable.
|
||
# # null falls through to default. Configurable via /admin/server-config UI.
|
||
# query_timeout_ms: 600000
|
||
# # DuckDB BigQuery extension query timeout (milliseconds).
|
||
# # Applied via `SET bq_query_timeout_ms` after every LOAD bigquery
|
||
# # on every BQ-touching DuckDB session. Extension default is
|
||
# # 90 000 ms = 90 s, which is too tight for analyst queries against
|
||
# # view-backed datasets -- bumped to 600 000 ms = 10 min by default.
|
||
# # Set 0 to fall through to the extension default. Configurable via
|
||
# # /admin/server-config UI.
|
||
# session_pool_size: 4
|
||
# # Number of pre-warmed DuckDB+bigquery-extension sessions kept
|
||
# # in a process-local pool. Each acquire amortizes the
|
||
# # ~0.5 s INSTALL/LOAD/CREATE-SECRET cost across requests; a fresh
|
||
# # build only happens when the pool is empty. Default 4. Set 0
|
||
# # to disable pooling (every acquire builds + closes a fresh
|
||
# # session; matches pre-pool behavior).
|
||
|
||
# --- OpenMetadata catalog (optional) ---
|
||
# Enriches table and column metadata from OpenMetadata REST API.
|
||
# If not configured, app works normally without catalog enrichment.
|
||
# All openmetadata.* fields configurable via /admin/server-config UI.
|
||
# openmetadata:
|
||
# url: "https://your-catalog.example.com"
|
||
# token: "${OPENMETADATA_TOKEN}" # JWT bearer token
|
||
# cache_ttl_seconds: 3600 # Cache TTL in seconds
|
||
# verify_ssl: true # set to false ONLY for internal
|
||
# # CAs / self-signed certs; defaults
|
||
# # to true. Setting false ships the
|
||
# # JWT over an unverified channel.
|
||
|
||
# --- Email delivery (optional, for magic link auth) ---
|
||
# Without SMTP, magic links are shown directly in browser (development mode).
|
||
# For production, configure any SMTP relay (Gmail, Mailgun, SendGrid SMTP, etc.)
|
||
email:
|
||
from_address: "noreply@example.com"
|
||
from_name: "AI Data Analyst"
|
||
smtp_host: "${SMTP_HOST}" # e.g., "smtp.gmail.com"
|
||
smtp_port: 587 # 587 for STARTTLS, 465 for SSL
|
||
smtp_user: "${SMTP_USER}"
|
||
smtp_password: "${SMTP_PASSWORD}"
|
||
|
||
# --- Desktop app (optional) ---
|
||
# All desktop.* fields configurable via /admin/server-config UI (rarely changed once set).
|
||
desktop:
|
||
jwt_issuer: "data-analyst"
|
||
jwt_secret: "${DESKTOP_JWT_SECRET}"
|
||
url_scheme: "data-analyst"
|
||
|
||
# --- Telegram notifications (optional) ---
|
||
telegram:
|
||
bot_token: "${TELEGRAM_BOT_TOKEN}"
|
||
bot_username: ""
|
||
domain_suffix: ""
|
||
|
||
# --- Jira integration (optional) ---
|
||
jira:
|
||
domain: ""
|
||
email: ""
|
||
api_token: "${JIRA_API_TOKEN}"
|
||
webhook_secret: "${JIRA_WEBHOOK_SECRET}"
|
||
sla_email: ""
|
||
sla_api_token: "${JIRA_SLA_API_TOKEN}"
|
||
cloud_id: ""
|
||
|
||
# --- Corporate Memory AI (optional) ---
|
||
# Extracts shared knowledge from team members' CLAUDE.local.md files.
|
||
# Provider: "anthropic" (direct API) or "openai_compat" (LiteLLM, OpenRouter, Azure, etc.)
|
||
ai:
|
||
provider: "anthropic" # or "openai_compat"
|
||
api_key: "${ANTHROPIC_API_KEY}" # or "${LLM_API_KEY}" for proxy
|
||
# base_url: "https://litellm.example.com" # Required for provider='openai_compat' (LiteLLM,
|
||
# OpenRouter, vLLM). Ignored when provider='anthropic'.
|
||
# Configurable via /admin/server-config UI.
|
||
model: "claude-haiku-4-5-20251001" # any model available on your provider
|
||
# --- Structured output quality control ---
|
||
# AI models can return JSON in three ways, each with different reliability:
|
||
#
|
||
# Layer 1 - "json_schema" (best):
|
||
# The provider enforces an exact schema. Every field, type, and structure
|
||
# is guaranteed. Available on: Anthropic, OpenAI, Claude via LiteLLM.
|
||
#
|
||
# Layer 2 - "json_object" (good):
|
||
# The provider guarantees valid JSON, but does not enforce a specific schema.
|
||
# Fields may be missing or have wrong types. Available on most providers.
|
||
#
|
||
# Layer 3 - "prompt" (acceptable):
|
||
# The AI is asked to respond in JSON via instructions in the prompt.
|
||
# No technical enforcement -- the model may still return invalid JSON.
|
||
# Works everywhere, but least reliable.
|
||
#
|
||
# "strict" = only Layer 1. Fail if provider doesn't support json_schema.
|
||
# Use when data quality is non-negotiable.
|
||
# "json" = Layer 1, fall back to Layer 2. No prompt-based fallback.
|
||
# Good balance of quality and compatibility.
|
||
# "auto" = All three layers as progressive fallback. Maximum compatibility.
|
||
# Use when you'd rather get imperfect data than no data.
|
||
structured_output: "auto"
|
||
|
||
# Legacy format (still supported, equivalent to provider: "anthropic"):
|
||
# ai:
|
||
# anthropic_api_key: "${ANTHROPIC_API_KEY}"
|
||
|
||
# Examples:
|
||
# --- LiteLLM proxy ---
|
||
# ai:
|
||
# provider: "openai_compat"
|
||
# base_url: "https://litellm.example.com"
|
||
# api_key: "${LLM_API_KEY}"
|
||
# model: "claude-haiku-4-5-20251001"
|
||
# structured_output: "strict"
|
||
#
|
||
# --- OpenRouter ---
|
||
# ai:
|
||
# provider: "openai_compat"
|
||
# base_url: "https://openrouter.ai/api/v1"
|
||
# api_key: "${OPENROUTER_API_KEY}"
|
||
# model: "anthropic/claude-3-haiku"
|
||
# structured_output: "auto"
|
||
|
||
# --- Flea-market upload guardrails (optional) ---
|
||
# Controls the pre-publish check pipeline for skill/agent/plugin uploads
|
||
# to /store. See docs/STORE_GUARDRAILS.md for the full check catalogue.
|
||
#
|
||
# guardrails:
|
||
# # Master kill-switch. When false, inline manifest/security/quality
|
||
# # checks still run (they're free) but the LLM step is skipped and new
|
||
# # uploads are auto-approved. Useful for local dev without an LLM key.
|
||
# enabled: true
|
||
#
|
||
# # Anthropic model tier for the LLM security review.
|
||
# # haiku — ~$0.001/review, default, good enough for routine uploads
|
||
# # sonnet — ~$0.015/review, deeper reasoning, fewer false negatives
|
||
# # opus — ~$0.075/review, only for high-stakes deployments
|
||
# # You can also pin a concrete model ID (e.g. "claude-haiku-4-5-20251001").
|
||
# review_model: "haiku"
|
||
#
|
||
# # Per-submitter daily cap on inline-blocked uploads. Bounds disk +
|
||
# # admin-queue spam. Set to 0 to disable. Default 50.
|
||
# blocked_quota_per_day: 50
|
||
#
|
||
# # How many days to keep blocked bundle bytes on disk before the
|
||
# # daily TTL job purges them. Submission row + sha256 + size always
|
||
# # survive — only the bundle bytes go. Set to 0 to retain forever
|
||
# # (rely on admin Delete). Default 30.
|
||
# blocked_bundle_ttl_days: 30
|
||
|
||
# --- Corporate Memory governance (optional) ---
|
||
# Controls how AI-extracted knowledge is reviewed and distributed.
|
||
# If not present, system operates in legacy mode (democratic wiki, no admin review).
|
||
#
|
||
# The corporate_memory.* schema is editable via /admin/server-config UI; you can
|
||
# also continue to manage it via this YAML file. The UI surfaces every leaf with
|
||
# a hint, so use it to discover the schema if this comment block has aged.
|
||
#
|
||
# corporate_memory:
|
||
# # How knowledge reaches users:
|
||
# # "mandatory_only" — admin controls everything, no user voting
|
||
# # "admin_curated" — admin controls, users vote as feedback signal
|
||
# # "hybrid" — mandatory from admin + optional from user voting (default)
|
||
# distribution_mode: "hybrid"
|
||
#
|
||
# # How new AI-extracted items enter the system:
|
||
# # "review_queue" — nothing published without admin approval (default)
|
||
# # "auto_publish" — items go live immediately, admin intervenes retroactively
|
||
# # "threshold" — high-confidence auto-publish, low-confidence to review queue
|
||
# approval_mode: "review_queue"
|
||
#
|
||
# # Default review period for approved/mandatory items (months)
|
||
# review_period_months: 6
|
||
#
|
||
# # Notify km_admins about new pending items
|
||
# notify_on_new_items: true
|
||
#
|
||
# # --- V1 Context Engineering ---
|
||
#
|
||
# sources:
|
||
# claude_local_md:
|
||
# enabled: true
|
||
# confidence_base: 0.50
|
||
# session_transcripts:
|
||
# enabled: true
|
||
# confidence_base: 0.60
|
||
# max_turns_per_session: 100
|
||
# detection_types:
|
||
# - correction
|
||
# - confirmation
|
||
# - unprompted_definition
|
||
#
|
||
# extraction:
|
||
# model: "claude-haiku-4-5-20251001"
|
||
# sensitivity_check: true
|
||
# contradiction_check: true
|
||
#
|
||
# confidence:
|
||
# # Base score per extraction source. Key format: "source_type" or "source_type.detection_type"
|
||
# base:
|
||
# user_verification.correction: 0.90
|
||
# user_verification.unprompted_definition: 0.90
|
||
# user_verification.confirmation: 0.60
|
||
# admin_mandate: 1.00
|
||
# claude_local_md: 0.50
|
||
# session_transcript: 0.50
|
||
# # Per-key modifier step sizes applied to base when optional signals are present.
|
||
# modifiers:
|
||
# user_verification.correction:
|
||
# additional_verifiers: 0.05 # per extra unique verifier
|
||
# user_verification.unprompted_definition:
|
||
# additional_verifiers: 0.05
|
||
# user_verification.confirmation:
|
||
# admin_confirmed: 0.20
|
||
# session_transcript:
|
||
# user_confirmed_in_session: 0.20
|
||
# # Confidence decay applied to items as they age.
|
||
# decay:
|
||
# mode: exponential # linear | exponential
|
||
# half_life_months: 12 # used when mode=exponential
|
||
# decay_rate_monthly: 0.02 # used when mode=linear
|
||
# floor:
|
||
# admin_mandate: 0.50 # admin policies don't silently decay to zero
|
||
# user_verification: 0.40 # user-verified facts never fall below 0.40
|
||
# default: 0.0
|
||
#
|
||
# contradiction_detection:
|
||
# enabled: true
|
||
# max_candidates: 10
|
||
#
|
||
# entity_resolution:
|
||
# enabled: true
|
||
# entities:
|
||
# metrics: ["churn", "MRR", "ARR", "NPS", "CAC", "LTV"]
|
||
# products: ["Platform", "API", "Dashboard"]
|
||
#
|
||
# domain_owners:
|
||
# finance: ["cfo@company.com"]
|
||
# engineering: ["cto@company.com"]
|
||
# product: ["pm@company.com"]
|
||
#
|
||
# domains:
|
||
# - finance
|
||
# - engineering
|
||
# - product
|
||
# - data
|
||
# - operations
|
||
# - infrastructure
|
||
|
||
# --- User groups for audience targeting (optional) ---
|
||
# Used with Corporate Memory governance to target mandatory knowledge to specific groups.
|
||
#
|
||
# groups:
|
||
# finance:
|
||
# label: "Finance & Analytics"
|
||
# members: ["analyst1@company.com", "analyst2@company.com"]
|
||
# engineering:
|
||
# label: "Engineering"
|
||
# members: ["dev1@company.com", "dev2@company.com"]
|
||
|
||
# --- User display and permissions ---
|
||
# Corporate Memory avatars + optional km_admin flag for governance.
|
||
# users:
|
||
# admin@company.com:
|
||
# display_name: "Admin User"
|
||
# km_admin: true # Corporate Memory admin (approve/mandate knowledge)
|
||
# analyst@company.com:
|
||
# display_name: "Analyst User"
|
||
users: {}
|
||
|
||
# --- Username mapping (webapp email -> server username, only if different) ---
|
||
username_mapping: {}
|
||
|
||
# --- Optional datasets (sync settings UI) ---
|
||
datasets: {}
|
||
|
||
# --- Data catalog ---
|
||
catalog:
|
||
categories: {}
|
||
order: []
|
||
|
||
# --- Data profiler (optional) ---
|
||
# profiler:
|
||
# sample_size: 500000 # If table > this, sample this many rows; otherwise use all
|
||
# max_categorical_distinct: 50 # Treat as categorical if unique <= this
|
||
# top_values_limit: 10 # Top values per categorical column
|
||
# histogram_bins: 15 # Bins in histogram visualizations
|
||
# sample_rows_limit: 5 # Sample rows to show in UI "Sample" tab
|
||
# alert_high_missing_pct: 30.0 # Alert threshold for high missing %
|
||
# alert_missing_pct: 5.0 # Alert threshold for missing %
|
||
# alert_imbalance_pct: 60.0 # Alert threshold for imbalance %
|
||
# alert_high_cardinality: 50 # Alert threshold for high cardinality columns
|
||
|
||
# --- Remote query (optional) ---
|
||
# Settings for remote BigQuery queries via `python -m src.remote_query`.
|
||
# Used when tables have query_mode: "remote" in data_description.md.
|
||
# remote_query:
|
||
# timeout_seconds: 300 # BQ + DuckDB query timeout
|
||
# max_result_rows: 100000 # Max rows in final output
|
||
# max_bq_registration_rows: 500000 # Max rows per --register-bq sub-query
|
||
# default_format: "table" # Default output format
|
||
# output_dir: "/tmp/remote_query" # Directory for Parquet/CSV exports
|
||
|
||
# --- v2 API knobs (optional) ---
|
||
# Controls for the /api/v2/{catalog,schema,sample,scan,scan/estimate} endpoints.
|
||
# All values are optional — the defaults shown below are applied if keys are absent.
|
||
#
|
||
# api:
|
||
# # --- Scan / fetch limits ---
|
||
# scan:
|
||
# max_limit: 10000000 # Hard row cap per /api/v2/scan request (default: 10 M)
|
||
# max_result_bytes: 2147483648 # Hard byte cap on Arrow stream response: 2 GB (default)
|
||
# # If exceeded, partial result returned with X-Agnes-Truncated header.
|
||
# max_concurrent_per_user: 5 # In-flight /api/v2/scan requests allowed per user (default: 5)
|
||
# # Note: quota is process-local; N replicas → effective N× cap.
|
||
# max_daily_bytes_per_user: 53687091200 # Per-user daily byte quota: 50 GB (default). Resets at UTC midnight.
|
||
# bq_cost_per_tb_usd: 5.00 # Cost rate shown in /api/v2/scan/estimate response (default: $5/TB)
|
||
# request_timeout_seconds: 300 # Server-side timeout for a single scan request (default: 300 s)
|
||
# # --- Discovery cache TTLs ---
|
||
# catalog_cache_ttl_seconds: 300 # /api/v2/catalog response cache lifetime (default: 5 min)
|
||
# schema_cache_ttl_seconds: 3600 # /api/v2/schema/{table_id} cache lifetime (default: 1 h)
|
||
# sample_cache_ttl_seconds: 3600 # /api/v2/sample/{table_id} cache lifetime (default: 1 h)
|
||
# # Admins can force-refresh via POST /api/v2/sample/{id}?refresh=true
|
||
|
||
# --- Materialize concurrency safety (optional) ---
|
||
# Concurrency safety net for the materialize path (BQ + Keboola). When
|
||
# two materialize attempts race for the same table_id, the second one
|
||
# raises MaterializeInFlightError and skips. The lock is held in a
|
||
# .parquet.lock sibling file; if a holder process is hard-killed before
|
||
# kernel-level flock release, the next attempt reclaims the lock once
|
||
# the file's mtime is older than this TTL.
|
||
#
|
||
# Default 86400 (24h) is generous on purpose — anything shorter risks
|
||
# a long-running COPY being interrupted by its own scheduler successor.
|
||
# Lower it only if you know your materialize never exceeds the new
|
||
# value AND your host has a habit of hard-killing processes.
|
||
# Min 60 (1 minute), max 604800 (7 days). Configurable via /admin/server-config UI.
|
||
materialize:
|
||
lock_ttl_seconds: 86400
|