perplexity-search
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of 'litellm', a reputable and widely used open-source library for interacting with multiple LLM providers via standard package registries.
- [COMMAND_EXECUTION]: Includes Python scripts for configuring the local environment and executing searches. These scripts perform routine operations such as managing .env files and invoking CLI commands for legitimate tool usage.
- [DATA_EXFILTRATION]: Performs network requests to OpenRouter's official API to retrieve search results. The skill manages credentials securely via environment variables and does not access or transmit sensitive local files.
- [INDIRECT_PROMPT_INJECTION]: The skill's function involves retrieving external web content. Evidence chain: 1. Ingestion point: query argument in scripts/perplexity_search.py. 2. Boundary markers: uses standard message roles for user content. 3. Capability inventory: includes network access for search results. 4. Sanitization: utilizes API-level structuring for data handling. This presents a common surface for untrusted data ingestion inherent to search tools, but lacks high-privilege execution or persistence capabilities.
Audit Metadata