paw-tools-release
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes Git and GitHub CLI (
gh) commands for release management tasks including tagging, committing, and pushing. These actions are legitimate and necessary for the tool's primary purpose. Evidence includes calls togit tag,git push, andgh release createinreferences/04-release.md. - [INDIRECT_PROMPT_INJECTION]: The skill processes Git commit messages to generate changelogs. While commit messages are external data that could contain malicious instructions, the risk is mitigated by the skill's specific focus on formatting release notes. The agent uses this data to update the project's changelog file and GitHub release description.
- Ingestion points: Commit messages extracted by
git logand processed inscripts/parse_commits.py. - Boundary markers: Not present.
- Capability inventory: File system writes (version files,
CHANGELOG.md), Git commands (push), and GitHub CLI actions (release create). - Sanitization: Standard regex categorization of commit types is used, but no specific prompt injection filtering is applied to the content.
Audit Metadata