The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs shell operations including project initialization, package installation, and plugin life-cycle management using the "paseo" CLI and NPM. It incorporates a critical safety control: the agent must verify the global "pluginsEnabled" setting and obtain explicit user consent before activation, presenting a clear security warning about the risks of running unsandboxed code.\n- [EXTERNAL_DOWNLOADS]: Fetches guidance and reference materials from the author's official domain (paseo.sh). These references are utilized to maintain compatibility with updated API contracts and are treated as trusted vendor resources.\n- [PROMPT_INJECTION]: The skill exhibits both direct and indirect injection surfaces:\n * Direct Injection: User input from "$ARGUMENTS" is placed directly into the instruction flow without delimiters.\n * Indirect Injection: The agent is instructed to fetch and follow rules from an external index (llms.txt) which guides its subsequent actions.\n * Ingestion points: SKILL.md (via the external documentation links) and variable user input.\n * Boundary markers: Absent; the instructions explicitly direct the agent to override its internal knowledge with the content of the fetched remote files.\n * Capability inventory: The skill possesses capabilities for file system modification, command execution, and network access (via npm).\n * Sanitization: No sanitization is performed on the data fetched from the documentation index or provided by the user.

paseo-plugin