github-release
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of local CLI tools to manage the release workflow. This includes git for versioning, the GitHub CLI (gh) for remote release management, and gitleaks for static analysis of secrets. These commands are well-defined and appropriate for the skill's context as a repository management utility.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests git commit history to automatically generate release notes, which presents an attack surface for indirect prompt injection.\n
- Ingestion points: The skill executes
git login SKILL.md and references/release-workflow.md to retrieve commit messages.\n - Boundary markers: There are no specific delimiters or instructions to treat commit messages as non-instructional data when formatting them into the release description.\n
- Capability inventory: The skill possesses the capability to execute shell commands (git, gh, npm) which could potentially be targeted if the agent misinterprets commit content as instructions.\n
- Sanitization: Commit messages are used directly in the
gh release createcommand without sanitization, which is standard behavior for release tools but noted as a potential surface.
Audit Metadata