The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes ls in the shell to enumerate markdown files within the skills/claude-code-reference/reference/ directory.
[EXTERNAL_DOWNLOADS]: It extracts URLs from documentation headers and uses the WebFetch tool via a sub-agent to retrieve content from external websites.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes untrusted data from the internet to generate file summaries. * Ingestion points: External documentation fetched from URLs defined in local files. * Boundary markers: The delegation prompt to the reference-researcher agent does not use delimiters or instructions to disregard instructions embedded in the fetched content. * Capability inventory: The skill can list files, fetch web data, and rewrite local markdown files. * Sanitization: No sanitization or safety filtering is applied to the content retrieved from the web before it is passed to the LLM for processing.

update-reference