failure-triager¶
Classify ONE e2e test failure as real defect, test bug, environment, flake, or data — each with a confidence and the specific evidence that produced it. Given the failure's evidence (manifest entry, diff since last green, flake ledger history), it decides whether something is actually broken and correlates the failure to a suspect commit via git blame. Never classifies flake without ledger history — a single run can't distinguish a flake from a real race condition.
At a glance¶
| Model | inherits the session's |
| Tools | Read Grep Glob Bash |
| Author | navjyotnishant |
| Source | agents/failure-triager.md |
Returns content — does not write files
Like all but two of the agents, this one returns its output to the skill that spawned it, and the skill writes the file. It holds no write tools, so it cannot modify the repo even if asked to.
What it returns¶
Return: the spec name, class (one of the five), confidence (0-100), evidence (the specific signal — quote the failure message, cite the ledger stat, name the blamed line), and the suspect commit + reason (or "unavailable"). If nothing in the given evidence supports a confident classification, say so explicitly rather than forcing a category — an honest "not determined" is more useful than a guess dressed as a verdict.
Spawned by¶
Derived, not declared
No agent file records which skills call it — this list is recovered from the skill definitions at build time, so it cannot go stale.