The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes common shell commands such as find, ls, grep, and wc to analyze the project structure and documentation coverage. It also includes instructions to test code examples found in markdown files when deemed safe, which involves executing code extracted from documentation.
[DATA_EXFILTRATION]: The skill is designed to extract and test external URLs found in documentation files to identify broken links. This involves making network requests to arbitrary external domains discovered during the validation process.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests untrusted content from markdown files across the repository.
Ingestion points: Processes files within docs/, source/README.md, and .claude/ directories.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore embedded commands within the processed documentation.
Capability inventory: The agent has access to bash for command execution and write for file modification.
Sanitization: No evidence of sanitization or validation of the ingested markdown content before processing or testing commands.

docs-validator