The issue_bridge crashed on startup because python-dotenv was not listed in requirements.txt, causing an ImportError when loading the .env file. The bridge was down for ~4 hours until the dependency was added and a clean reinstall performed.
ModuleNotFoundError: No module named 'dotenv'. GitHub issue ingestion halts.
dotenv was a transitive dependency in the dev environment but never pinned in requirements.
python-dotenv>=1.0.0 to requirements.txt. Clean reinstall + bridge restart succeeds.
A developer had python-dotenv installed globally in their venv from an earlier session, so the from dotenv import load_dotenv import worked locally. The CI environment installs from requirements.txt, but since the package wasn't listed there, it was never tested in CI.
The issue was one of 4 regressions in the Sep 21 batch push — all authored Aug 30 but never exercised in the deployment pipeline until the push landed as a group.
| Ingestion downtime | ~4 hours |
|---|---|
| Issues missed | 12 (backlog ingested after fix) |
| Data loss | None — GitHub retains issue events |
| SLA breach | No |
The verify_gate.py had a hardcoded path /Users/brandonbennett/tmp that caused a PermissionError on Ubuntu CI runners. The test was checking for write access to a local dev directory that doesn't exist in CI.
/Users/brandonbennett/tmp not present on ubuntu runner.
tempfile.gettempdir(). CI green on all runners.
A debug path was committed instead of using the stdlib tempfile module. The test passed locally because the macOS dev directory exists, but CI runs on Ubuntu where the path doesn't exist and the parent isn't writable.
| PRs blocked | 3 (including #88 which was later superseded) |
|---|---|
| CI downtime | ~2 hours |
| Data loss | None |
| SLA breach | No |
When a bot (phymora, scribe, student-ci, student-executor, student-whymage, etc.) resolves a qualifying event — production bug, CI red >30min, agentic loop failure, data corruption, auth failure, or anything requiring rollback or manual intervention — they MUST append a new incident entry following the template structure above.
Qualifying events: SEV-1/2/3 incidents, CI red >30 min, agentic loops wasting >5 cycles, data integrity issues, any event requiring rollback or manual intervention.
ID format: INC-YYYY-MMDD-NNN (sequential per day).