The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and extract information from local codebase components such as code, comments, and configuration files. This creates a surface for indirect prompt injection where malicious instructions embedded in comments could influence the agent's behavior.
Ingestion points: The skill ingests codebase content via the APIDocExtractor, DocCoverage, and other analysis scripts that read local files.
Boundary markers: No explicit delimiters are used in the prompt templates to separate user-provided code from the agent's core instructions.
Capability inventory: The skill has access to the local filesystem and can generate shell commands for documentation builds and CI/CD pipelines.
Sanitization: The skill contains instructions to avoid exposing secrets but does not include programmatic sanitization for embedded prompt instructions.
[EXTERNAL_DOWNLOADS]: The skill's resources reference external libraries and tools from well-known and trusted sources.
Fetches swagger-ui-dist from jsdelivr, a well-known CDN service.
References official GitHub Actions from the actions organization for environment setup.
Specifies the installation of standard documentation tools such as @redocly/cli and sphinx.
[COMMAND_EXECUTION]: The implementation playbook includes templates for shell commands to automate documentation workflows.
Includes standard package management commands like pip install and npm install.
Includes build commands for documentation generation like sphinx-build and redocly build-docs.

documentation-generation-doc-generate