The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and follow instructions from external markdown files (sitemaps) located in sitemaps/<site>/ and the user home directory at ~/.opencli/sites/<site>/sitemap/.
Ingestion points: The skill reads SITE.md, pages/*.md, workflows/*.md, and pitfalls.md to guide agent navigation and action choices.
Boundary markers: The instructions include a "Trust Reality Rule" and state that the sitemap is "prior knowledge, not ground truth," which provides a conceptual boundary for the agent but does not programmatically sanitize the file contents.
Capability inventory: The skill uses Bash(opencli:*) for browser automation and adapter execution, alongside Read, Write, and Edit for local file management.
Sanitization: There is no evidence of sanitization or schema validation for the contents of the sitemap files before they are used to influence tool execution.
[COMMAND_EXECUTION]: The skill instructs the agent to dynamically select and execute opencli commands and adapters (e.g., opencli twitter post) based on the "Best path" recommendations found within the untrusted sitemap data.

opencli-browser-sitemap