The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions for the agent to fetch and install external components from the inference.sh registry and other specific repositories such as belt-sh/cli using the npx skills tool.
[REMOTE_CODE_EXECUTION]: Through the npx skills add command, the skill enables the dynamic integration and execution of remote code packages at runtime to expand the agent's functionality.
[COMMAND_EXECUTION]: The skill's configuration explicitly allows the use of shell commands via the Bash tool to perform discovery and management tasks, such as npx skills search, npx skills list, and npx skills update.
[INDIRECT_PROMPT_INJECTION]: The skill features an attack surface where metadata (such as descriptions and names) retrieved from the external skill registry is processed by the agent. If the registry contains malicious content, it could potentially influence the agent's subsequent actions.
Ingestion points: Data enters the agent's context when performing npx skills search or npx skills list against the remote registry.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings for registry data.
Capability inventory: The agent has shell access via Bash(npx skills *), allowing it to modify its own execution environment by adding or removing skills.
Sanitization: No explicit sanitization or filtering is described for the text output returned from the skill registry.

related-skill