Loading ToolBench
Agent-to-n8n: AI agent that composes and deploys n8n workflows from natural language. Exposes n8n-a2e capabilities as MCP tools for AI agents to search nodes, compose workflows, deploy to n8n instances, and manage workflow patterns.
Server has 9 tools with explicit schemas and descriptions, but quality is uneven. Naming is verb-forward and clear (search_, get_, compose_, deploy_, list_, generate_, refresh_). Descriptions are present but often generic, most lack actionable context about when to use each tool or what happens on failure. Schemas are well-formed JSON Schema with types and required fields, but output schemas are completely undocumented. Error handling is absent from descriptions. Parameters lack constraints (enums, ranges, patterns). The compose_workflow and deploy_workflow tools are high-risk (WRITE/IRREVERSIBLE) but have no confirmation or dry-run guidance. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels.
Compose an n8n workflow from a structured plan. Returns valid workflow JSON ready for deployment.
Deploy a workflow JSON to the configured n8n instance.
Generate a rich context string with relevant nodes and patterns for a given use case. Useful for composing workflows.
Get the required fields schema for a credential type (e.g. slackApi, openAiApi). Use before creating credentials to know what data is needed.
Get full details of a specific n8n node type including all parameters, credentials, inputs/outputs.
List available n8n instance profiles from n8n-cli. Shows connection details and catalog status for each profile.
No output schemas documented. LLMs cannot plan downstream calls or extract required fields (e.g., what does search_n8n_nodes return? field names? structure?). Violates pattern:tool and pattern:response-shaper.
Irreversible operations (deploy_workflow, refresh_node_catalog) lack confirmation or dry-run guidance. No error recovery hints. Agents can deploy broken workflows without safeguards. Violates pattern:confirmation-request and pattern:recovery-guide.
Parameters lack constraints. 'limit' in search_n8n_nodes has no min/max (unbounded). 'query' fields have no length limits or format guidance. Enums missing where applicable (e.g., credential types). Violates pattern:constrained-input.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
List known workflow patterns/templates that can be used as starting points.
Re-extract all node definitions from the running n8n instance and update the local search index. Use after installing custom nodes and restarting n8n, so new nodes become searchable and usable in compose_workflow.
Search for n8n nodes by natural language description. Returns matching nodes with relevance scores.
Descriptions are generic and lack actionable context. 'Compose an n8n workflow from a structured plan' does not explain when to call it vs generate_workflow_context, what 'structured plan' means, or what happens on validation failure. Violates pattern:tool-description.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels. deploy_workflow is IRREVERSIBLE but unmarked. search_n8n_nodes is READ_ONLY but unmarked. Violates current MCP spec (2026-07-28) tool annotations.