perplexity-search
Pass
Audited by Gen Agent Trust Hub on Mar 31, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes the LiteLLM library to interface with the OpenRouter API. These are well-known and legitimate services used for AI model integration.
- [CREDENTIALS_UNSAFE]: The skill requires an OpenRouter API key. It correctly instructs users to manage this sensitive information using environment variables or a .env file. The provided setup script (scripts/setup_env.py) automates this process using standard file-writing practices and does not exfiltrate the key.
- [COMMAND_EXECUTION]: The main search script (scripts/perplexity_search.py) uses a command-line interface to accept user queries. These queries are passed to the LiteLLM completion function as text input and are not executed as shell commands or system-level code.
- [DATA_EXFILTRATION]: Network operations are restricted to communication with the OpenRouter API (openrouter.ai) for the purpose of retrieving search results. No unauthorized data transmission to third-party or unknown domains was detected.
- [SAFE]: Analysis of the skill's logic, scripts, and documentation confirms that its behavior aligns with its stated purpose of providing grounded search results with citations. There are no signs of obfuscation, privilege escalation, or persistence mechanisms.
Audit Metadata