The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of external software packages from GitHub or the skills.sh registry. It specifically identifies vercel-labs/agent-skills (a trusted organization) as a source, but the npx skills add command can be directed to fetch content from any unverified user or repository.- [COMMAND_EXECUTION]: Instructs the agent to run subprocesses using the npx utility. Specifically, it uses npx skills find for search and npx skills add <package> -g -y for installation. The use of the -y flag is particularly notable as it bypasses interactive confirmation prompts, allowing for silent installation of external code.- [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection via tool output poisoning. The agent is instructed to process and present search results from an external registry. If a malicious package is registered with a deceptive name or description, the agent might ingest and potentially follow instructions embedded in that metadata.
Ingestion points: Output of the npx skills find command processed by the agent.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are provided for the search result processing step.
Capability inventory: The skill possesses the capability to execute shell commands and install software via npx.
Sanitization: There is no evidence of sanitization or validation of the text returned from the skills registry before it is presented to the user or acted upon.

find-skills