release-skills
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill automates repository management by executing local shell commands for Git (e.g., git log, git tag, git commit) and the GitHub CLI (gh release, gh pr view). These operations are standard for the tool's purpose and are performed within the local project context.
- [DYNAMIC_EXECUTION]: The skill supports project-specific customization through a
.releaserc.ymlconfiguration file. This allows for the execution of user-defined hooks (prepare_artifactandpublish_artifact) as shell commands. The skill mitigates potential shell injection risks by writing release notes to a UTF-8 encoded temporary file and passing the file path to these hooks instead of inlining the text directly into command strings. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill retrieves external, untrusted data from git commit messages and GitHub Pull Request metadata (via
gh pr view). - Boundary markers: This data is used to generate natural language descriptions in Markdown changelogs, but it is not directly evaluated as code by the agent.
- Capability inventory: The skill has access to Git, GitHub CLI, and the ability to execute configured shell hooks.
- Sanitization: The skill avoids direct interpolation of multiline changelog content into shell commands by using temporary files, which reduces the surface area for shell-based injection attacks.
Audit Metadata