attack-path-stitcher

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime reads outsider-authored free text from findings/finding-NNN/evidence/raw-source.txt (evidence produced by external parties/other users), which is then parsed by tools/chain-merger.py to extract secrets and generate edges that are included in the LLM context/output artifacts.