retrospective-codify

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The workflow ingests “feedback source” inputs such as PR review comments and bug reports/user corrections (steps 12-14) at runtime to extract lessons and append/update them in the project feedback ledger (steps 19-25), so an outsider can supply free text via those channels that the LLM will read.