explain-pr
Pass
Audited by Gen Agent Trust Hub on Nov 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from Pull Request descriptions, titles, commit messages, and linked issues, which could potentially contain malicious instructions.
- Ingestion points: Untrusted data enters the agent's context through tools like
gh pr viewand by reading PR-related metadata as described inSKILL.md. - Capability inventory: The agent has the ability to read local files, execute Git/GitHub CLI commands, and build/run code in a temporary environment.
- Boundary markers: The instructions explicitly tell the agent to "Treat the PR description as a claim to check, not as fact," which provides a level of instruction-following defense.
- Sanitization: No specific sanitization or filtering of external PR content is mentioned.
- [DYNAMIC_EXECUTION]: The skill includes instructions to build and execute code from the Pull Request's SHAs to capture screenshots of UI changes.
- Evidence: The skill states it can "run the app at the merge-base and head SHAs... Capture only after the user agrees, since it builds and runs the PR's code" (found in
SKILL.md). - [COMMAND_EXECUTION]: The skill performs various shell-level operations using
gitandgh(GitHub CLI) to resolve PRs, fetch branches, and view diffs. - Evidence: References to
gh pr view,gh pr diff,git show, andgit fetchare present throughoutSKILL.md.
Audit Metadata