The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dynamically identifies and executes commands found in project configuration files such as justfile, Makefile, and package.json. This behavior is intended to allow the skill to run project-specific pre-merge checks (e.g., tests, linting, visual regressions) without hardcoding every possible command.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from repository files (justfile, Makefile, package.json) and git history to determine execution steps. While there is an inherent risk in executing commands derived from project files, the skill is designed to run within the user's controlled repository environment. \n
Ingestion points: Project configuration files (justfile, Makefile, package.json) and git log output (SKILL.md).\n
Boundary markers: None explicitly implemented to separate instructions from repo data.\n
Capability inventory: Shell command execution via Bash (limited to git, gh, just, make, npm, pnpm, bun, yarn), file reading via Read, Grep, and Glob, and PR management via gh CLI.\n
Sanitization: The skill instructs the subagent to 'Return the exact commands to run, do not invent any', but does not implement programmatic sanitization of the discovered commands.

ship