langchain-perplexity
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interacts with external data by fetching live web search results via Perplexity AI. This introduces a surface for indirect prompt injection if malicious content is retrieved and processed by the agent.
- Ingestion points:
ChatPerplexity.invoke(),PerplexitySearchRetriever.invoke(), andPerplexitySearchResultstool mentioned inSKILL.md. - Boundary markers: None explicitly defined in the code snippets, though the underlying components are standard LangChain tools.
- Capability inventory: The skill retrieves real-time search data and interprets it as chat context or RAG documents.
- Sanitization: The skill includes
strip_think_tagsfor reasoning models, but specific sanitization for external search data is not provided in the examples. - [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the
langchain-perplexitypackage and references a comprehensive list of LangChain ecosystem packages from official registries and repositories. - Evidence:
pip install -U langchain-perplexityinSKILL.mdand the extensive package index inreferences/llms.md.
Audit Metadata