The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions for the agent to install external software using the npx skills add command. This enables the installation and execution of code from the agent skills ecosystem, including potentially untrusted third-party repositories.
[COMMAND_EXECUTION]: The documentation in SKILL.md suggests using the -y flag during the installation process. This bypasses interactive confirmation prompts, facilitating the automated installation of software without a human-in-the-loop review of the package source.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and interpret untrusted data from an external search index via the npx skills find command. Malicious metadata within the skills ecosystem (such as skill names or descriptions) could be crafted to influence agent behavior or mislead the agent into recommending unsafe packages.
Ingestion points: Output from the npx skills find command, which is processed by the agent in SKILL.md.
Boundary markers: No explicit delimiters are specified to isolate external search data from the agent's internal instructions.
Capability inventory: The skill uses the npx skills CLI to install and execute code as described in SKILL.md.
Sanitization: The skill includes verification heuristics for the agent, such as prioritizing skills with high install counts (>1K), established source reputations (referencing official vendors), and GitHub star counts.

find-skills