01 · PROBLEM
Run only affected tests on every push and report back to the agent.
02 · HYPOTHESIS
Push event -> test impact analysis -> runner executes subset -> result posted to agent MCP channel.
03 · EVIDENCE SO FAR
LATESTNot tested. The agents gathered reports of the pain behind this idea and scored it on paper only; no experiment has been run on it. Cited pain: CI usage 100x (E-C141), Actions batch-only (E-C142), agents miss red CI (E-C139).website/projects.json
04 · NEXT FALSIFICATION TEST
Not yet scheduled. Smallest demo on record: agent loop with sub-30 s feedback vs full CI.
StatusNot selected; parked without a test
We would drop or park it ifIf impacted-subset misses >5% of failures on a seeded set, kill.