skill-suggest
Warn
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill performs network requests to https://skills.sh/api/search to retrieve a list of available agent skills based on the detected project environment.
- [REMOTE_CODE_EXECUTION]: The skill facilitates the installation and subsequent execution of external agent skills. The identity of the installed code is determined by external API responses, posing a supply-chain risk if the registry or the recommended sources are compromised.
- [COMMAND_EXECUTION]: The skill constructs and runs shell commands using npx skills add to modify the local agent configuration. This includes using the -y flag to bypass CLI-level confirmations for the installation process.
- [INDIRECT_PROMPT_INJECTION]: The skill parses untrusted data from local manifest files to build search queries and categorize technologies. Ingestion points: Project manifest files and repository configurations (SKILL.md Step 1b). Boundary markers: No specific delimiters or safety instructions are provided to the agent when processing the contents of these manifest files. Capability inventory: File system scanning, network requests, and shell command execution (npx). Sanitization: The skill implements a reputational filtering system where only skills with high install counts (>= 10,000) are selected by default.
Audit Metadata