release-skills
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses standard version control and repository management tools (
gitandgh) to analyze history, create commits, and manage tags. These operations are restricted to the local repository context and are consistent with the tool's primary purpose of automating release workflows. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from the repository, including commit messages, PR author details, and local configuration files (
.releaserc.yml,package.json). While this external content influences versioning logic and changelog text generation, the risk of malicious instructions being obeyed is mitigated by a mandatory user confirmation step before final release actions (commit/push) are taken. - Ingestion points: Commit messages, git log output, GitHub PR metadata, and project configuration files.
- Boundary markers: Not explicitly mentioned for LLM processing, but output is presented to the user for review.
- Capability inventory: Git repository management (
git add,git commit,git tag,git push), GitHub CLI operations (gh pr view,gh repo view), and file system writes (version and changelog files). - Sanitization: Relies on user confirmation in Step 8 to validate the generated release content and version bump decision.
Audit Metadata