jj-workflow

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

• Third-party content exposure detected (high risk: 0.80). Flagged because SKILL.md and the referenced guides explicitly instruct the agent to run commands like "jj git fetch", inspect remote refs (e.g., master@origin, teammate/their-feature), and then rebase/bookmark/push based on those fetched commits—i.e., the agent pulls and interprets commits from remote Git remotes (e.g., GitHub) which are user-generated third‑party content and can materially change subsequent tool actions.