make-pr
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from git diffs to generate pull request descriptions.
- Ingestion points: The skill reads output from
git diff <target>...HEADas described inSKILL.mdandREFERENCE.md. - Boundary markers: Instructions require a "clean-room trace" where every generated line must be mapped to a specific code hunk, and it explicitly blocks inferring info from conversation or branch names.
- Capability inventory: The skill is authorized to use
gitcommands (fetch, push) and interact with PR platform APIs (create, update, read). - Sanitization: The skill includes a "Body hygiene" step in Step 4 to strip unauthorized marketing footers or identity trailers before finalizing the PR.
Audit Metadata