The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill directs the agent to use npx -y clawhub, which downloads and executes the 'clawhub' CLI tool directly from the npm registry during the session.
[EXTERNAL_DOWNLOADS]: The tool is designed to search for and download 'skills' from the external OpenClaw registry, introducing third-party code and markdown into the agent's filesystem and context.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to run shell commands that modify the local environment by writing files to the {skills_folder_path} directory.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection. \n
Ingestion points: Data returned from the clawhub search command and the content of skills installed from the external registry are processed by the agent. \n
Boundary markers: There are no boundary markers or instructions to ignore embedded commands provided for the external data. \n
Capability inventory: The agent has the capability to execute shell commands (npx) and modify its own available tools through skill installation. \n
Sanitization: No mechanisms for validating or sanitizing the content fetched from the OpenClaw registry are defined in the skill.

openclaw