The Agent Skills Directory

[NO_CODE]: The skill is entirely composed of Markdown templates and guidance for the agent. It does not include any Python, JavaScript, or other executable scripts.
[EXTERNAL_DOWNLOADS]: The skill recommends using badges from well-known and trusted services like GitHub, npm, PyPI, and Shields.io to provide metadata in project documentation.
[REMOTE_CODE_EXECUTION]: One documentation template for CLI tools includes a placeholder for a curl | bash installation command. This is a common pattern for documenting software installation and is presented as a template for the user, not as code to be executed by the agent.
[PROMPT_INJECTION]: The skill processes project manifest files (e.g., package.json, composer.json) and git logs to generate documentation.
Ingestion points: Data is read from local project configuration files and command-line outputs (git logs) to populate templates.
Boundary markers: The skill does not define specific delimiters for untrusted data, as the data is used for static documentation generation.
Capability inventory: The skill facilitates directory scaffolding and file writing through the agent's file system tools.
Sanitization: Ingested project data is used directly in generated markdown; however, since the skill only generates documentation and does not execute the data, the risk is negligible.

project-docs