fix(webui): remediate PR #876 REQUEST_CHANGES for analytics (#651)

Address review blockers and medium/low findings:

- F1: HTML-escape all dynamic analytics fields (html.escape quote=True)
- F2: Gate POST /api/v1/analytics/usage through console_authz
  record_analytics_usage (fail closed for unauthenticated / Phase 1)
- F3: Enforce usage_events retention (max rows + max age)
- F4: Coerce None remote/org/repo to empty strings in load_analytics

Add tests for XSS escaping, unauthorized ingest deny, retention, and
None scope coercion. Document authz + retention in analytics guide.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
This commit is contained in:
2026-07-24 08:11:12 -04:00
co-authored by Claude Opus 4.8
parent 0041542fe2
commit b179610e7f
9 changed files with 279 additions and 55 deletions
+34 -1
View File
@@ -577,6 +577,11 @@ class ControlPlaneDB:
conn.execute("CREATE INDEX IF NOT EXISTS idx_usage_events_role_model ON usage_events(role, model);")
conn.execute("CREATE INDEX IF NOT EXISTS idx_usage_events_stage ON usage_events(stage);")
# #651 retention: cap growth so unauthenticated or high-volume ingest
# cannot DoS the control-plane DB (PR #876 F3). Applied after every write.
USAGE_EVENTS_MAX_ROWS = 10_000
USAGE_EVENTS_RETENTION_DAYS = 90
def record_usage_event(
self,
*,
@@ -649,7 +654,35 @@ class ControlPlaneDB:
ts,
),
)
return cursor.lastrowid
usage_id = cursor.lastrowid
self._enforce_usage_events_retention(conn)
return usage_id
def _enforce_usage_events_retention(self, conn: sqlite3.Connection) -> None:
"""Drop aged and excess usage_events rows (PR #876 F3)."""
# Age-based: ISO-8601 UTC timestamps compare lexicographically.
cutoff = (
datetime.now(timezone.utc)
- timedelta(days=int(self.USAGE_EVENTS_RETENTION_DAYS))
).strftime("%Y-%m-%dT%H:%M:%SZ")
conn.execute(
"DELETE FROM usage_events WHERE created_at < ?",
(cutoff,),
)
# Count-based: keep the newest USAGE_EVENTS_MAX_ROWS by usage_id.
max_rows = int(self.USAGE_EVENTS_MAX_ROWS)
if max_rows > 0:
conn.execute(
"""
DELETE FROM usage_events
WHERE usage_id NOT IN (
SELECT usage_id FROM usage_events
ORDER BY usage_id DESC
LIMIT ?
)
""",
(max_rows,),
)
def query_usage_events(
self,