The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/lint_runner.py executes system commands including npm, npx, ruff, and mypy using subprocess.run to perform code linting.
Evidence: The script detects project configuration files (e.g., package.json, pyproject.toml) and executes the corresponding linter commands found in the detect_project_type function.
The execution uses structured argument lists and specific working directories, which is a standard safety practice.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by reading and analyzing project files (.ts, .tsx, .py, package.json) to generate reports.
Ingestion points: Project source files are read in scripts/type_coverage.py using pathlib.Path.read_text and project metadata is read in scripts/lint_runner.py via json.loads.
Boundary markers: None identified. The content of these files is processed directly by regex or external linters.
Capability inventory: The skill can execute shell commands via subprocess.run (as listed in scripts/lint_runner.py) and perform file system reads.
Sanitization: scripts/type_coverage.py uses regular expressions to extract type information, and scripts/lint_runner.py relies on the security of the underlying tools (like ESLint or Ruff) to handle file content safely.

lint-and-validate