The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local system commands to manage the release process.
Evidence: Uses git status, git describe, git log, git add, git commit, git tag, and git push to manipulate the local and remote repository state.
Evidence: Utilizes the GitHub CLI tool (gh) with commands like gh run list and gh run watch to monitor CI/CD pipelines.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the processing of untrusted external data.
Ingestion points: The skill reads untrusted commit messages using git log <previous-tag>..HEAD --oneline in SKILL.md.
Boundary markers: Absent. There are no instructions or delimiters provided to prevent the agent from following instructions embedded within the commit messages it parses.
Capability inventory: The skill has the capability to write to the filesystem (package.json, changelog.mdx), commit changes, and push code to remote origins via git push (documented in SKILL.md).
Sanitization: Absent. Commit messages are categorized and written directly into documentation files without escaping or validation.

release-version