changelog
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands including
git log,gh pr list, andgh release create. These commands use arguments derived from user-supplied git references and PR numbers, as well as repository data. - Evidence: Phase 1 uses
git logandgh pr listto gather data. Phase 5 usesgh release create <tag> --notes "<changelog content>"to publish the results. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted commit messages and pull request descriptions from the project's git history to generate human-friendly release notes.
- Ingestion points: Commit messages, PR titles, and PR descriptions gathered in Phase 1, Step 1.
- Boundary markers: The instructions lack explicit delimiters or warnings to ignore instructions embedded within the commit messages during the polishing phase.
- Capability inventory: The skill has the ability to write to the local file system (
CHANGELOG.md) and perform network operations via theghCLI (gh release create). - Sanitization: There is no evidence of sanitization or escaping of the gathered git data before it is processed by the LLM or passed to the GitHub Release command.
Audit Metadata