diff-explain

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

• Third-party content exposure detected (high risk: 0.85). The required workflow ingests a PR diff via gh pr diff <url-or-number> or a pasted diff/branch diff, and that diff content is authored by outsiders (e.g., other contributors’ PR changes) which becomes readable text in the agent’s LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches pull-request diffs at runtime via the GitHub CLI (e.g. gh pr diff <url-or-number> which pulls content from URLs like https://github.com///pull/), and that fetched external diff content is injected into the agent's prompt/context to drive its explanation — a direct runtime external dependency controlling agent output.