FlexMeasures / flexmeasures / 21757930189

78%

push

github

Fix API test isolation: set fs_authn_via for Flask-Security session auth (#1960)

* Initial plan

* Fix API test isolation by setting fs_authn_via in requesting_user fixture

Co-authored-by: Flix6x <30658763+Flix6x@users.noreply.github.com>

* Plan: Update review-lead instructions and fix precommit issues

Co-authored-by: Flix6x <30658763+Flix6x@users.noreply.github.com>

* agents/test-specialist: learned API test isolation patterns for Flask-Security

Context:
- PR fixing issue #1298 revealed centralized API test isolation solution
- Flask-Security auth workaround is now in api/conftest.py with autouse fixture
- Pattern: global patch + requesting_user fixture sets fs_authn_via='session'
- Removed outdated workaround note about prepending test_auth_token

Change:
- Added 'API Test Isolation' section with detailed explanation
- Documented the three-part solution: patch fixture, patched function, session marker
- Added guidance for writing new API tests with checkmarks for do/don't patterns
- Linked to issue #1298, Flask-Security issue, and original PR for context
- Removed outdated workaround instructions (problem is now fixed)

* agents/review-lead: learned to invoke coordinator when explicitly requested

Context:
- Session on 2026-02-06 reviewing API test isolation PR
- User explicitly asked to check if 'agent instruction updates' were needed
- Review Lead failed to invoke Coordinator subagent as required
- This is Coordinator's domain (meta-agent for agent lifecycle/governance)
- Missed opportunity: agents didn't self-update their instructions

Change:
- Added 'Must Actually Run Coordinator When Requested' section
- Clarified trigger patterns: 'agent instructions', 'governance', 'structural concerns'
- Emphasized Review Lead ≠ Coordinator (different roles, different expertise)
- Added example workflow showing when to delegate to Coordinator
- Updated 'Learning from Failures' with specific 2026-02-06 lesson
- Documented failure mode: Review Lead tryi... (continued)

1611 of 2134 branches covered (75.49%)

Branch coverage included in aggregate %.

7 of 7 new or added lines in 1 file covered. (100.0%)

11175 of 14315 relevant lines covered (78.06%)

6.24 hits per line