The Agent Skills Directory

[COMMAND_EXECUTION]: The skill incorporates a local Python script, scripts/verify_doc_only_diff.py, designed to help the agent verify that its code modifications are limited to non-executable segments like docstrings. The script uses safe subprocess.run calls to interact with the local git binary. Additionally, the skill instructs the agent to utilize standard project runners like uv for testing and linting.
[INDIRECT_PROMPT_INJECTION]: As a coding assistant, the skill is designed to ingest and process untrusted data from the user's repository, including source code, design documents, and project rules. This capability represents a standard indirect prompt injection surface. Ingestion points: Project instructions (SKILL.md), nearby implementation files, tests, and design documents. Boundary markers: No explicit instructions to ignore embedded adversarial directives in processed files. Capability inventory: File system writes, git subprocess execution via provided script, and project tool execution. Sanitization: No specific sanitization or filtering of external content described.
[EXTERNAL_DOWNLOADS]: The skill references official guidance and repositories from Pydantic on GitHub for advanced modeling tasks. These references target a well-known and trusted organization.

coding-standards