The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute various shell commands, including Git operations for branching and worktrees, GitHub CLI (gh) for pull request management, and several workflow management scripts (feature-state.py, adr-query.py, learning-db.py) located in the user's home directory under ~/.claude/scripts/.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It ingests and processes untrusted repository data (such as ADRs, seeds, and source code) to generate implementation plans and tasks for sub-agents. During the validate phase, it executes project-controlled build and test commands (e.g., npm test, make check, pytest), providing a surface where malicious project configuration files could execute code or influence agent behavior. The instructions in references/validate.md explicitly direct the agent to execute these commands without reasoning about their results or safety, and the skill lacks explicit boundary markers or sanitization for the ingested data.

feature-lifecycle