The Agent Skills Directory

[COMMAND_EXECUTION]: The skill extensively uses the Bash() function to execute shell commands for git operations, build processes, and publishing artifacts (e.g., gh release create, npm publish, docker push). These are expected behaviors for a release management tool.\n- [EXTERNAL_DOWNLOADS]: The skill leverages npx to download and execute the claude-flow package and its swarm coordination modules at runtime (e.g., lines 35, 252, 532). These downloads target the official NPM registry.\n- [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8) due to its data processing workflow.\n
Ingestion points: It reads untrusted data such as commit messages and Pull Request titles/bodies using gh api and gh pr list (found in SKILL.md at lines 144, 237, and 393).\n
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore or isolate commands potentially embedded within these ingested strings.\n
Capability inventory: The agent has high-privilege capabilities including shell access (Bash) and file system writes (Write).\n
Sanitization: No sanitization or validation of the commit history or PR metadata is performed before the content is passed to the AI agents for changelog generation and task orchestration.

github-release-management