LangGraph Patterns Expert
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides code patterns for agents that ingest user messages and interact with external tools, which defines an indirect prompt injection attack surface.
- Ingestion points: The
app.invoke(input)andagent.stream(...)methods inSKILL.mdare designed to process external user input. - Boundary markers: The example snippets do not demonstrate the use of explicit delimiters or instructions to isolate untrusted data from system prompts.
- Capability inventory: The documented agents include tools for database queries and web searches (
db_query_tool,search_tool). - Sanitization: No input validation or sanitization logic is present in the architectural examples.
- [EXTERNAL_DOWNLOADS]: The skill references documentation and source code from established repositories.
- Links to official LangGraph documentation hosted on GitHub Pages.
- Links to the official LangGraph source code repository on GitHub.
Audit Metadata