The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local development tools and scripts in SKILL.md, specifically uv run pytest, uv run ruff, uv run mypy, and bin/ci-local, to verify code behavior during refactoring.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted source code and tests while having command execution capabilities. 1. Ingestion points: Source code and test files as specified in SKILL.md. 2. Boundary markers: None; the instructions do not implement delimiters to isolate untrusted code. 3. Capability inventory: Shell execution of uv run commands and the bin/ci-local script in SKILL.md. 4. Sanitization: No sanitization or content validation is performed on ingested files before command execution.

refactor