Skills
Navigate
skills/jeremylongshore/tons-of-skills-marketplace/langchain-langgraph-streaming/Gen Agent Trust Hub

langchain-langgraph-streaming

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of API and WebSocket endpoints that ingest untrusted user data for processing by an LLM graph.\n
  • Ingestion points: The q query parameter in SKILL.md (Step 3) and references/astream-events-filtering.md, and the user_message JSON field in the WebSocket handler in SKILL.md (Step 5) and references/websocket-and-reconnect.md.\n
  • Boundary markers: The provided code snippets wrap user input in HumanMessage(user_input) before passing it to the LangGraph astream method.\n
  • Capability inventory: The skill is intended for use in environments with Read, Write, Edit, and Bash(python:*) tool access, which determines the scope of potential impact from a successful injection attack.\n
  • Sanitization: The implementation templates do not include input validation, sanitization, or safety filtering for the user-provided text.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:45 AM
Security Audit — agent-trust-hub — langchain-langgraph-streaming