chore(cli-rename): replace stale da verbs in active code paths

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.

app/api/admin.py

@@ -2463,7 +2463,7 @@ async def unregister_table(
     For materialized rows, also removes the canonical parquet at
     `${DATA_DIR}/extracts/<source_type>/data/<id>.parquet` and clears
     the matching `sync_state` row. Without these two cleanups, the
-    manifest endpoint kept advertising the dropped table to `da sync`
+    manifest endpoint kept advertising the dropped table to `agnes pull`
     (sync_state-driven) and the orchestrator's next rebuild could
     resurrect a master view from the leftover parquet (E2E sub-agent
     finding 2026-05-01).
@@ -2515,7 +2515,7 @@ async def unregister_table(
     # Clear sync_state for any source/mode (a row that was synced at any
     # point — local/materialized — has a sync_state entry that the manifest
     # serves regardless of registry state). Pre-fix, the manifest still
-    # advertised the dropped table to `da sync` because sync_state was
+    # advertised the dropped table to `agnes pull` because sync_state was
     # never cleaned up, and analysts kept getting it through the manifest.
     try:
         conn.execute("DELETE FROM sync_state WHERE table_id = ?", [name])
@@ -2523,7 +2523,7 @@ async def unregister_table(
     except Exception as e:
         logger.warning(
             "Failed to clear sync_state for unregistered table %s: %s — "
-            "manifest may still advertise the dropped row to da sync",
+            "manifest may still advertise the dropped row to agnes pull",
             table_id, e,
         )
 

app/api/claude_md.py

@@ -119,7 +119,7 @@ async def get_welcome(
 ):
     """Return the rendered CLAUDE.md for the authenticated analyst.
 
-    The CLI calls this endpoint during ``da analyst setup`` to write
+    The CLI calls this endpoint during ``agnes init`` to write
     ``<workspace>/CLAUDE.md``. The content is RBAC-filtered per the
     calling user.
 

app/api/sync.py

@@ -59,7 +59,7 @@ def _run_materialized_pass(conn: duckdb.DuckDBPyConnection, bq) -> dict:
     that are due, dispatching by ``source_type`` to the correct connector's
     materialize_query. Honors per-table `sync_schedule` via `is_table_due()`,
     computes the file hash inline, and updates `sync_state` so the manifest
-    can serve the row to `da sync` without re-hashing on every request.
+    can serve the row to `agnes pull` without re-hashing on every request.
 
     BigQuery rows go through BqAccess + bigquery_query() (jobs API),
     optionally cost-guarded by ``max_bytes_per_materialize``.

app/web/templates/admin_tables.html

@@ -1197,7 +1197,7 @@
                             <textarea class="form-textarea" id="kbSourceQuery" rows="8"></textarea>
                             <div class="form-hint">SELECT against <code>kbc."bucket"."table"</code>.
                                 Result is materialized to parquet and distributed via
-                                <code>da sync</code>.</div>
+                                <code>agnes pull</code>.</div>
                         </div>
 
                         <div class="form-group">
@@ -2404,7 +2404,7 @@
             if (_editOriginalQueryMode === 'materialized' && newMode === 'remote') {
                 msg = '⚠ Switching Synced locally → Live from BigQuery will drop the materialized parquet on the next sync. Analysts who already pulled this table will start getting live BQ queries instead of a local copy; the sync schedule becomes ignored.';
             } else {
-                msg = '⚠ Switching Live from BigQuery → Synced locally: the next scheduled sync runs a SELECT and writes a parquet. Analysts will start reading the local copy on their next `da sync`. Remember to set a sync schedule.';
+                msg = '⚠ Switching Live from BigQuery → Synced locally: the next scheduled sync runs a SELECT and writes a parquet. Analysts will start reading the local copy on their next `agnes pull`. Remember to set a sync schedule.';
             }
             warn.textContent = msg;
             warn.style.display = '';

app/web/templates/admin_workspace_prompt.html

@@ -234,7 +234,7 @@
       <p class="welcome-desc">
         <strong>Default:</strong> a rich markdown briefing about Agnes commands, registered tables
         (RBAC-filtered for the calling analyst), available metrics, and marketplace plugins.
-        Written to <code>CLAUDE.md</code> in the analyst workspace at <code>da analyst setup</code> time.
+        Written to <code>CLAUDE.md</code> in the analyst workspace at <code>agnes init</code> time.
         Use <code>--no-claude-md</code> to skip writing it.
       </p>
       <p class="welcome-desc">

cli/commands/catalog.py

@@ -61,7 +61,7 @@ def catalog(
 
 
 def _list_metrics(as_json: bool, category: Optional[str] = None) -> None:
-    """List metric definitions from the server (lifted from `da metrics list`)."""
+    """List metric definitions from the server."""
     params = {}
     if category:
         params["category"] = category
@@ -99,7 +99,7 @@ def _list_metrics(as_json: bool, category: Optional[str] = None) -> None:
 
 
 def _show_one_metric(metric_id: str, as_json: bool) -> None:
-    """Show details for a single metric (lifted from `da metrics show`)."""
+    """Show details for a single metric."""
     resp = api_get(f"/api/metrics/{metric_id}")
     if resp.status_code == 404:
         typer.echo(f"Metric not found: {metric_id}", err=True)

cli/commands/disk_info.py

@@ -34,7 +34,7 @@ def disk_info(
     used = sum(p.stat().st_size for p in snap_dir.rglob("*") if p.is_file()) if snap_dir.exists() else 0
     count = len(list(snap_dir.glob("*.parquet"))) if snap_dir.exists() else 0
     # Fall back to the user's local dir for free-space stats when the
-    # snapshot dir doesn't exist yet (first run, before any `da fetch`).
+    # snapshot dir doesn't exist yet (first run, before any `agnes snapshot create`).
     # `0` would misleadingly suggest the disk is full.
     free = shutil.disk_usage(snap_dir if snap_dir.exists() else _local_dir()).free
     quota_gb = int(os.environ.get("AGNES_SNAPSHOT_QUOTA_GB", "10"))

cli/commands/pull.py

@@ -101,7 +101,7 @@ def pull(
         # Claude Code's stdout stays clean. Errors still flow to stderr so
         # the user sees them in their terminal even when the hook redirects
         # `2>/dev/null` (the hook explicitly forwards stderr too in the
-        # canonical `da analyst setup` template).
+        # canonical `agnes init` template).
         if result.errors:
             for e in result.errors:
                 typer.echo(f"warn: {e}", err=True)

cli/commands/status.py

@@ -1,8 +1,7 @@
 """`agnes status` — workspace status: initialized? data fresh? hooks active?
 
-Replaces the old `da analyst status` command. The previous server-health
-content (which used to live here) has moved to `agnes diagnose system`
-under the existing `agnes diagnose` group (Task 13).
+Server-health checks live under `agnes diagnose system` (see the
+`agnes diagnose` group).
 """
 
 from __future__ import annotations

cli/error_render.py

@@ -3,7 +3,7 @@
 Three CLI paths surface BigQuery / guardrail / RBAC typed errors today:
 - ``agnes query --remote`` (POST /api/query)
 - ``agnes query --register-bq`` (RemoteQueryEngine wrapping BqAccessError)
-- ``da fetch`` / ``agnes schema`` etc. (cli.v2_client wrappers around v2 endpoints)
+- ``agnes snapshot create`` / ``agnes schema`` etc. (cli.v2_client wrappers around v2 endpoints)
 
 All three previously flattened the structured ``detail`` JSON to a
 truncated single-line string, hiding the operator-facing hint that

cli/skills/corporate-memory.md

@@ -6,7 +6,7 @@ and makes it available to everyone through a curated catalog.
 
 ## How It Works
 1. Analysts write insights in their CLAUDE.local.md
-2. `da sync --upload-only` pushes content to server
+2. `agnes push` uploads sessions and CLAUDE.local.md to the server
 3. Server processes with LLM (Haiku) to extract knowledge items
 4. Items go through governance (pending → approved/mandatory)
 5. Approved items are distributed as Claude rules

cli/skills/troubleshoot.md

@@ -18,9 +18,9 @@ agnes diagnose --json
 3. Re-generate token: `agnes login --email your@email.com`
 
 ### DuckDB errors locally
-1. Re-sync: `da sync` (rebuilds views)
+1. Re-sync: `agnes pull` (rebuilds views)
 2. Check disk space: `du -sh user/duckdb/`
-3. Delete and re-create: `rm user/duckdb/analytics.duckdb && da sync`
+3. Delete and re-create: `rm user/duckdb/analytics.duckdb && agnes pull`
 
 ### Server unresponsive
 1. `docker compose ps` — check container status

cli/snapshot_meta.py

@@ -80,7 +80,7 @@ def delete_snapshot(snap_dir: Path, name: str) -> bool:
 def snapshot_lock(snap_dir: Path):
     """Exclusive flock on snap_dir/.lock — serializes snapshot installs.
 
-    Concurrent `da fetch` invocations queue here.
+    Concurrent `agnes snapshot create` invocations queue here.
     """
     if _fcntl is None:
         raise RuntimeError(

src/db.py

@@ -956,7 +956,7 @@ _V16_TO_V17_MIGRATIONS = [
 # v19 -> v20: source_query column backs query_mode='materialized' for BigQuery.
 # Admin-registered SQL stored verbatim; scheduler runs it through the DuckDB BQ
 # extension (via BqAccess) and writes the result to
-# /data/extracts/bigquery/data/<id>.parquet so the existing manifest + da sync
+# /data/extracts/bigquery/data/<id>.parquet so the existing manifest + agnes pull
 # flow distributes it to analysts. NULL on existing rows.
 _V19_TO_V20_MIGRATIONS = [
     "ALTER TABLE table_registry ADD COLUMN IF NOT EXISTS source_query TEXT",

src/repositories/sync_state.py

@@ -92,7 +92,7 @@ class SyncStateRepository:
         doesn't claim a sync happened. Existing rows keep their last
         successful `last_sync` / `rows` / `hash` fields — only `status` and
         `error` flip — so analysts who already pulled the prior good
-        parquet via `da sync` keep serving from it while the operator fixes
+        parquet via `agnes pull` keep serving from it while the operator fixes
         the source.
         """
         self.conn.execute(