A pipe module that integrates N8N workflows into an agent system, forwarding messages to N8N and returning responses
Single tool 'pipe' has critical definition gaps. Tool name lacks action verb (should be 'forward_to_n8n' or 'invoke_n8n_workflow'). Description is present but generic (48 chars, below 50-200 char baseline for LLM optimization). Input schema visible but parameters lack proper type definitions and descriptions, 'body' is dict with nested 'messages' array, but no enum constraints, format specs, or validation guidance. Output schema undocumented. No error handling guidance for LLM recovery. Tool performs a WRITE operation (calls external API) but lacks idempotent/destructive annotations. Credentials (n8n_bearer_token) exposed in Valves config, not server-side injected.
Processes incoming messages by forwarding them to an N8N workflow and returning the workflow response
Tool name 'pipe' lacks action verb. Should be 'invoke_n8n_workflow', 'forward_to_n8n', or 'send_to_n8n' to signal intent to LLM.
Input schema parameters lack descriptions. 'body', '__user__', '__event_emitter__', '__event_call__' have no guidance on what values are valid or expected. LLM cannot infer parameter meaning.
Output schema undocumented. Tool returns dict but no schema specifies fields, types, or structure. LLM cannot plan downstream operations or extract required data.
Credentials (n8n_bearer_token) exposed in Valves config. Should use server-side secret injection via environment variables. Tokens in config risk leaking into logs and traces.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 22 | 2026-07-28+ | v2 |
No error handling guidance. Tool catches exceptions and returns {'error': str(e)}, but LLM receives no recovery hints (retry? ask user? fatal?). Error messages lack actionable next steps.