The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs extensive command-line operations using git, gh (GitHub CLI), bun, npm, uv, and cargo. This includes cloning vendor repositories, managing branches, and executing a local smoke test by compiling the project into a temporary binary and running it. These actions are within the expected scope of a release automation tool.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it fetches Pull Request titles and summaries from GitHub (gh pr view) to generate changelog entries. An attacker could potentially embed malicious instructions in a PR description. However, this is mitigated by a mandatory human review step (Step 7) where the user must verify the generated changelog before any commits or PRs are created.
Ingestion points: Data enters the context via gh pr view output as described in Step 5 of SKILL.md.
Boundary markers: None identified in the prompt interpolation of the PR data.
Capability inventory: The skill has significant capabilities including file writes, git commits/pushes, and remote GitHub API interactions via gh across multiple steps in SKILL.md.
Sanitization: No explicit sanitization of PR content is performed; the skill relies on the agent to 'rewrite' the entries and the user to review them.

release