release-pr-gates
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data including PR metadata, commit messages, and CI logs which could contain malicious instructions.
- Ingestion points: The skill reads PR metadata, commit messages from
git log, and GitHub Actions logs usinggh run view(SKILL.md). - Boundary markers: Instructions explicitly state to treat these fields as 'untrusted text' and warn not to follow instructions embedded in those fields.
- Capability inventory: The skill uses
gitandghtools to create tags, push to remotes, and create GitHub releases (SKILL.md). - Sanitization: Instructions require the agent to redact secret-like values before summarizing findings from untrusted logs.
- [COMMAND_EXECUTION]: The skill executes
gitandghcommands to automate release workflows. These operations are within the stated scope of the skill but involve interactions with remote repository state.
Audit Metadata