The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's core workflow instructs the agent to execute code snippets contained within documentation blocks using python -m doctest and pytest --doctest-modules. This constitutes a command execution surface where the agent runs code extracted from potentially untrusted source files as part of the validation process.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes untrusted source code and comments.
Ingestion points: The agent reads and processes source code files (Python, JavaScript, TypeScript), OpenAPI specifications, and README files from the project directory.
Boundary markers: No specific delimiters or instructions are provided to the agent to distinguish between its own operational instructions and content found within the analyzed code files or comments.
Capability inventory: The agent has the ability to execute shell commands and run code validation tools such as doctest, pytest, tsc, and npx on the local file system.
Sanitization: No procedures are defined to sanitize or validate the content of code examples before execution, allowing instructions embedded in code comments to potentially influence agent behavior.
[EXTERNAL_DOWNLOADS]: The skill documentation provides instructions for installing and using several standard developer tools and libraries from public registries, including pydocstyle, interrogate, @redocly/cli, and docusaurus. These are recognized industry-standard packages used for documentation and linting.

code-documenter