langgraph-routing
Pass
Audited by Gen Agent Trust Hub on Feb 15, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (LOW): The semantic routing implementations use untrusted user input to determine the next node in the graph.
- Ingestion points: state['query'] in SKILL.md and state['input'] in references/semantic-routing.md.
- Boundary markers: None identified.
- Capability inventory: The routing logic influences agent flow but does not perform direct side effects.
- Sanitization: No input validation or sanitization is performed on the user query.
Audit Metadata