The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands through Bash blocks and example scripts to perform release operations, including gh release, npm publish, git checkout, and docker build.
[EXTERNAL_DOWNLOADS]: The skill frequently invokes npx claude-flow, which triggers a download and execution of the vendor's orchestration package from the npm registry.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions designed to influence the agent's behavior during the release process.
Ingestion points: Untrusted data enters the agent context via gh pr list and gh api repos/:owner/:repo/compare in SKILL.md, which capture commit messages and PR titles.
Boundary markers: The instructions do not specify the use of delimiters or provide explicit "ignore embedded instructions" warnings when passing these variables (e.g., $CHANGELOG, $PRS) to AI-driven tools.
Capability inventory: The skill has extensive capabilities, including writing to the file system (Write, TodoWrite), executing shell commands (Bash), and performing authenticated network requests (gh commands, npm publish).
Sanitization: The skill does not demonstrate any sanitization, filtering, or validation of the fetched commit messages or PR titles before they are processed by the AI agents.

qe-github-release-management