The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the subprocess module to interact with the system environment.
Multiple scripts (e.g., validate_resolution.py, check_foundation.py, check_production.py) execute pytest to run tests on the codebase.
holes_to_beads.py interacts with the bd (Beads) CLI tool to create and update issues.
discover_holes.py and generate_report.py execute git and radon commands to analyze repository state and code complexity.
[EXTERNAL_DOWNLOADS]: The documentation (README.md and SKILL.md) instructs the user to install the beads tool via go install github.com/steveyegge/beads/cmd/bd@latest. This is a remote download and installation of an external binary from a third-party repository not on the trusted vendor list.
[DYNAMIC_EXECUTION]: The validate_resolution.py script dynamically identifies and executes Python test files based on the hole_id provided at runtime. Specifically, it uses Path("tests/refactor").glob(f"test_{self.hole_id.lower()}_*.py") to select files which are then passed to pytest. This could lead to the execution of arbitrary test code generated during the refactoring process.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze an entire codebase to identify "holes" for refactoring.
Ingestion points: The discover_holes.py script parses and analyzes the codebase at the project root.
Boundary markers: There are no explicit delimiters or instructions in the generated REFACTOR_IR.md to ignore malicious instructions embedded in the analyzed code.
Capability inventory: The skill possesses capabilities for file writing (REFACTOR_IR.md), git branch manipulation, and arbitrary code execution via pytest.
Sanitization: The holes_to_beads.py script performs minimal sanitization on data extracted from the REFACTOR_IR.md (e.g., slicing strings) before passing them as arguments to the bd command line.

typed-holes-refactor