The Agent Skills Directory

Command Execution: The skill workflow includes the command make build-docs to verify the integrity of the documentation structure after modifications. This involves executing a subprocess via the local shell, which is a standard procedure for project-specific builds.
Indirect Prompt Injection Surface: The skill is designed to ingest and analyze untrusted content from the repository's source files (src/agents/) and examples to identify documentation gaps. This creates a surface where instructions embedded in code comments could potentially influence the agent's behavior during the audit or update process.
Ingestion points: Analyzes implementation details in src/agents/ and examples/ via ripgrep and git show commands.
Capability inventory: The skill has the ability to write to files within the docs/ directory and execute make build-docs.
Boundary markers: The instructions do not define specific delimiters or "ignore" warnings to prevent the agent from following instructions found within the analyzed source code.
Sanitization: No explicit sanitization or filtering is mentioned for content extracted from implementation files before it is used to generate documentation updates.
Trusted Service Integration: The workflow utilizes the $openai-knowledge MCP server to retrieve information from the official OpenAI Developer Documentation. This is an informative step used to reconcile SDK source code with platform features using a known service.

docs-sync