commit-archaeologist

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

• Third-party content exposure detected (high risk: 0.85). The required workflow runs scripts/archaeologist.py, which ingests outsider-authored free text from local git metadata—specifically commit subjects/bodies (git show in read_commit) and blame author info (git blame in blame_authors)—and includes those strings in the JSON/human report fed back into the agent’s LLM context.