The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses pnpm dlx to execute afdocs and is-agentic, which are auditing tools for AI agent readiness. These tools are fetched from public registries during the workflow.- [COMMAND_EXECUTION]: Includes a bundled Bash script (scripts/check-surfaces.sh) used to verify HTTP headers and content types of the implemented machine-readable files. This verification step ensures that the implemented changes meet the required technical specifications.- [REMOTE_CODE_EXECUTION]: The skill provides instructions to download and run the url-discovery-bench tool from its GitHub repository (mintlify/url-discovery-bench). This tool is used to measure the efficiency of agent navigation through documentation.- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from audit reports and server logs to define the implementation scope.
Ingestion points: Processes external reports from services like afdocs, is-agentic, and isitagentready.com, as well as local access.log files.
Boundary markers: The workflow includes a specific step to map findings to real product surfaces and explicitly skip any findings that do not correspond to the actual product (Step 3).
Capability inventory: The agent is empowered to perform file system modifications and run the scripts/check-surfaces.sh tool for network probing.
Sanitization: The process includes a mandatory verification step (Step 6) that uses automated probing to validate that implementations (like llms.txt and markdown twins) respond correctly to HTTP requests.

agent-ready