Agent Post-Mortem
Self-improving agent loop. Runs automated root-cause analysis on verification agent failures (hallucinations, missed bugs, missed vulnerabilities, false positives). Classifies against a failure taxonomy, generates executable skill patches, and stages version-controlled changes for human approval. Closes the feedback loop between agent failure and agent improvement.
Expected outcome
Self-improving agent loop. Runs automated root-cause analysis on verification agent failures (hallucinations, missed bugs, missed vulnerabilities, false positives). Classifies against a failure taxonomy, generates executable skill patches, and stages version-controlled changes for human approval. Closes the feedback loop between agent failure and agent improvement.
Typical deliverable
An evidence-backed findings report with severity and recommended action.
01
Inspect context
Loads project authority and checks prerequisites.
02
Run the procedure
Executes defined steps and honors safety gates.
03
Return evidence
Produces a reviewable result instead of an unsupported claim.
Licensed implementation
The licensed module adds trigger phrases, CLI aliases, dependency and coordination maps, and its step-by-step governed procedure.
No per-skill purchases: a plan unlocks every skill in its tier, this one included.