Build, check and fix n8n workflows from Claude, Cursor or ChatGPT. Built on 1,100+ real n8n node definitions and curated templates, with a validator that repairs common mistakes (wrong node types, missing settings) and returns JSON ready to import into n8n. Sign in with a free gydmation.com account; nothing to install.
Scored across 23 tools
Most tools target a distinct stage of the n8n workflow lifecycle (search catalog, generate, validate, deploy, run, inspect), and descriptions explicitly cross-reference each other to draw boundaries. The only soft overlaps are the validate_workflow / autofix_workflow / self_heal_workflow ladder and save_workflow vs update_workflow_partial, but the descriptions clearly state when to use each (static check vs AI repair vs real-run iteration; new copy vs in-place edit).
Every name is snake_case verb_noun (get_node, search_nodes, save_workflow, update_workflow_partial, set_workflow_active), with the resource as the trailing noun so related tools sort together. Even the noun-centric helpers (get_best_practices, get_ai_workflow_patterns) follow the get_ prefix convention.
23 tools is on the heavy side of the 3-15 sweet spot, but the domain genuinely spans catalog lookup, authoring, validation, repair, deployment, and execution inspection. Each tool maps to a real step, though consolidating the validate/autofix pair would trim the surface without losing capability.
The surface covers the full lifecycle: node/template discovery, generation, validation and repair, deploy and incremental edit, activation, deletion, triggering, and execution diagnosis, plus status and credential listing. Minor gaps remain around credential setup guidance, execution retry/bulk inspection, and workflow version history, but agents can work around these via get_workflow and get_execution.
Like validate_workflow, but when issues remain it asks an AI model to
repair them (up to 2 rounds) and re-validates. Nothing is deployed.
Returns the repaired workflowJSON, ready to paste into n8n.
Args:
workflow_json: The workflow JSON to repair.
intent: What the workflow is meant to do -- helps fixes stay faithful.
Inputs: workflow_json (required), intent.
Permanently delete a workflow from your n8n. Cannot be undone. Needs "Allow AI assistants to delete or activate workflows" in Gydmation Settings. Always confirm with the user first.
Inputs: workflow_id (required).
Deploy one of Gydmation's verified templates to your n8n, unmodified
(use get_template + save_workflow to adapt it first).
Args:
template_id: From search_templates.
template_url: Older name for template_id.
activate: Activate right away (needs the delete/activate permission).
Inputs: activate, template_id, template_url.
Explain an n8n workflow JSON in plain terms: its trigger, the steps in
order, the apps it uses, which credentials to set up in n8n, any
disconnected nodes, and how to import it (paste into the n8n canvas).
Use it before handing a workflow to the user.
Args:
workflow_json: The complete n8n workflow JSON.
Inputs: workflow_json (required).
Generate a complete, validated n8n workflow from a natural-language
description, grounded in Gydmation's real node schemas, with automatic
fixes applied where possible. Nothing is deployed -- use save_workflow.
Needs an AI model: your assistant's own model if it supports MCP
sampling (few do yet), otherwise the account's Gydmation Pro plan, or a
GROQ_API_KEY set on this install.
Args:
prompt: What the workflow should do, in plain English.
tools: Optional app/integration names to use (e.g. ["slack", "postgres"]).
Inputs: prompt (required), tools.
Architectures for AI workflows in n8n with how to wire them and common
pitfalls. Patterns: ai_agent_with_tools, rag_knowledge_base,
classify_and_route, summarize_and_notify, human_in_the_loop.
Inputs: pattern.
Practical rules for building reliable n8n workflows. Topics:
workflow_building, credentials, expressions_and_data, error_handling,
webhooks, scheduling, testing. Empty topic returns all of them.
Inputs: topic.
What really happened in one execution: overall status, a diagnosis of
any errors (from n8n's real error data, never guessed), and per node:
status, item count, duration, error and a short output sample.
Inputs: execution_id (required).
The full verified definition of one n8n node: exact type and
typeVersion, default parameters, credential type and usage notes, plus
a ready-to-paste node skeleton. Accepts a catalog name ("slack"), a
full type ("n8n-nodes-base.slack") or a short type ("nodes-base.slack").
Args:
node: Node name or type.
Inputs: node (required).
What this connection can do right now: which Gydmation account it acts
for, whether that account is on Pro, whether an n8n instance is
connected, and whether deleting/activating workflows is allowed. Call
this first in a new conversation.
Inputs: no named parameters.
The complete workflow JSON of one template (from search_templates), to
read, adapt and then save with save_workflow.
Args:
template_id: The template_id from search_templates.
template_url: Older name for template_id (kept so AI apps that cached
the previous tool list, like ChatGPT, keep working).
Inputs: template_id, template_url.
The full JSON of one workflow in your n8n.
Inputs: workflow_id (required).
Credentials in your n8n -- id, name and type only, never secret values.
Reference one in a node as {"<type>": {"id": "...", "name": "..."}}.
Inputs: no named parameters.
Recent executions (id, status, start/stop) -- all, or for one workflow.
Inputs: limit, workflow_id.
List workflows in your n8n (id, name, active).
Inputs: limit.
Deploy a workflow to your n8n. Without workflow_id it creates a new
workflow. With workflow_id the edited version is deployed as a NEW
workflow named "... (updated)" and the original is kept so it can be
compared (for in-place edits use update_workflow_partial).
Args:
workflow_json: The complete workflow JSON.
workflow_id: Existing workflow to base an updated copy on (optional).
activate: Activate right away (needs the delete/activate permission).
Inputs: workflow_json (required), activate, workflow_id.
Search Gydmation's n8n node catalog by keyword (name, n8n type string,
usage notes): every official n8n node plus maintained community nodes
(those carry community_package -- the user installs it in n8n first).
Use before writing any node; then get_node for the full definition.
Args:
query: e.g. "slack", "webhook", "schedule", "postgres".
limit: Maximum matches.
Inputs: query (required), limit.
Meaning-based search over Gydmation's 1,950+ verified workflow templates.
Check here before generating from scratch.
Args:
prompt: What you want to automate.
app: Only templates using these apps, comma-separated, e.g. "Gmail, Slack".
trigger: Only templates started by: webhook, schedule, manual, form,
chat, email, app_event, sub_workflow or error.
complexity: beginner, intermediate or advanced.
industry: e.g. "E-commerce", "Marketing", "Support".
limit: Maximum matches (1-25).
Each match has a template_id to pass to get_template or deploy_template.
Inputs: prompt (required), app, limit, trigger, +2 more.
Runs the workflow for real on your n8n, reads back the actual result,
and either proposes a fix for review (mode "suggest", the default) or
fixes and re-tests up to max_iterations times (mode "auto"). Running
can have real side effects (messages, emails) -- ask the user first.
Args:
workflow_id: The deployed workflow's id.
workflow_json: Its current JSON (starting point for fixes).
original_prompt: What it is meant to do, if known.
mode: "suggest" or "auto".
max_iterations: Fix/re-test attempts (hard max 5).
Inputs: workflow_id (required), workflow_json (required), mode, max_iterations, +1 more.
Turn a workflow on (it starts reacting to its trigger for real) or off. Needs "Allow AI assistants to delete or activate workflows" in Gydmation Settings. Ask the user first.
Inputs: active (required), workflow_id (required).
Run a test execution of a saved workflow (Manual/Webhook-style
triggers). It runs for real, so it may send real messages -- ask the
user first. Then read the result with get_execution. For
Schedule/Chat/Form triggers prefer self_heal_workflow.
Inputs: workflow_id (required).
Edit an existing n8n workflow with small operations instead of
resending the whole JSON. Operations (applied in order):
{"op":"add_node","node":{name,type,typeVersion,parameters,credentials?,position?}}
{"op":"remove_node","name":"..."} (also removes its connections)
{"op":"update_node","name":"...","parameters":{...merged...},"set":{"typeVersion":2,"disabled":false,...}}
{"op":"rename_node","name":"old","new_name":"new"} (connections follow)
{"op":"add_connection","from":"A","to":"B","output_index":0,"input_index":0,"type":"main"}
{"op":"remove_connection","from":"A","to":"B","type":"main"}
{"op":"set_name","name":"..."}
{"op":"set_settings","settings":{"timezone":"Asia/Kolkata","errorWorkflow":"..."}}
The result is validated first. If n8n accepts in-place updates the
workflow is updated directly; otherwise it is saved as "... (updated)".
Args:
workflow_id: The workflow to edit.
operations: List of operations as above.
dry_run: Apply and validate only; return the result without saving.
Inputs: operations (required), workflow_id (required), dry_run.
Check an n8n workflow JSON against Gydmation's real node schemas:
connections (no orphaned nodes), credential structure, node-type
casing and AI-node requirements (an Agent needs a language model).
Applies safe deterministic fixes without any AI call. Returns the
checked (and repaired) workflowJSON, ready to paste into n8n.
Args:
workflow_json: The complete n8n workflow JSON.
Inputs: workflow_json (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Static n8n workflow validator: workflow errors and repair suggestions. No execution.
Free public n8n, AI support and software acceptance templates; scoped private delivery workflows.
Know when your n8n workflows, URLs, and AI apps break, before your customers do.
Security scanner for n8n workflows + live MCP Trust-Check. 18 rules, OWASP mapped. Paid x402 API.
Add one secure layer between your agents and this server.
