forge-loop

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM via Step 5/6/7 sub-agents that run /review-pr and /await-ci, which necessarily read PR review comments and CI failure logs from GitHub (authored by non-user parties, e.g., Copilot/Claude reviewers and CI systems) and feed them into the agent context for responding/fixing.