n8n-agents

Pass

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill documents the creation of agents that ingest untrusted natural language data (e.g., chatInput from Slack or Webhooks in CHAT_AGENT_PATTERNS.md).
  • Ingestion points: Untrusted data enters via Execute Workflow Trigger inputs and Slack Trigger payloads as documented in EXAMPLES.md and CHAT_AGENT_PATTERNS.md.
  • Boundary markers: The skill explicitly instructs users to use specific tool descriptions, modular system prompts (SYSTEM_PROMPT.md), and to plumb sensitive parameters (like user IDs and session keys) deterministically to prevent the LLM from fabricating or influencing them.
  • Capability inventory: The agents described are designed to interact with tools for messaging (Slack, Gmail), database operations (Notion), and external APIs (HTTP Request Tool).
  • Sanitization: The skill mitigates risks by mandating human-in-the-loop review for all destructive tools (HUMAN_REVIEW.md) and requiring structured output parsing with schema validation and auto-fix capabilities to ensure the agent's output is safe for downstream processing (STRUCTURED_OUTPUT.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 29, 2026, 05:43 PM
Security Audit — agent-trust-hub — n8n-agents