The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest data from browser sessions and existing sitemap files to create task execution graphs. This creates a surface where malicious content from a website could be recorded into a sitemap and later influence the agent's actions when the sitemap is used.
Ingestion points: Reads content from browser sessions via opencli browser and existing sitemap files in sitemaps/<site>/ and ~/.opencli/sites/<site>/sitemap/.
Boundary markers: Utilizes structured Markdown and YAML blocks to define fields, but does not provide explicit instructions to sanitize or escape data harvested from web pages before including it in sitemap actions.
Capability inventory: The skill has the ability to write and edit local files and execute opencli browser commands, which could be misused if the agent follows instructions embedded in harvested data.
Sanitization: No explicit validation or filtering logic is defined for the content extracted from websites during the authoring loop.
[COMMAND_EXECUTION]: The skill uses the opencli suite of tools through Bash(opencli:*) to perform browser automation, analyze site states, and manage navigation data. These commands are necessary for the skill's intended purpose of mapping website structures for AI agents.

opencli-sitemap-author