The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes project-defined quality gates, including tests, linters, type-checkers, and build commands as part of the TDD implementation workflow.
[COMMAND_EXECUTION]: A bundled Python utility script (cook.pyz) is used to resolve artifact paths; this script executes git commands via subprocess for environment discovery (e.g., retrieving the project root and origin URL).
[PROMPT_INJECTION]: The skill processes untrusted natural language requirements from external 'spec' files and pasted requirements, creating an indirect prompt injection surface (Category 8).
Ingestion points: SKILL.md specifies reading specs via path or slug and processing pasted text or issues.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the ingested content.
Capability inventory: The skill can execute shell commands for testing/building and write to the filesystem via the cheez-write skill.
Sanitization: No sanitization or validation of the ingested natural language instructions is mentioned.
[EXTERNAL_DOWNLOADS]: The skill references external development tools like gh (GitHub CLI), delta, just, and mergiraf. gh is a well-known service tool, and others are common developer utilities.

cook