The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the uv tool to execute its internal scripts/invoke.py wrapper script and an external manager engine.- [DYNAMIC_EXECUTION]: The scripts/invoke.py script dynamically searches for a vendor-specific engine script (kntnt.py) across dozens of hidden directories associated with common AI agent platforms (such as .claude/skills, .cursor/skills, and .config/goose/skills) and executes it using a computed path.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted text from local files or remote URLs. While the instructions include explicit boundary guidance (e.g., treating code blocks as quoted material and strictly limiting changes to mechanical errors), the processing of external content into the model's context remains an inherent surface for indirect prompt injection.- [EXTERNAL_DOWNLOADS]: The skill allows the user to provide a URL as a source for proofreading, resulting in the retrieval of content from remote servers.

proofread