Explore the workspace
Browse folders and workflows, create a folder, then discover available node types and their configuration schemas.
Your assistant imagines. nodulz executes.
With the nodulz MCP server, your assistant can do more than describe a workflow. It can explore your workspace, build the graph, validate it, start a run and read the results, according to the permissions you grant.
Your AI assistantMCPYour nodulz workspace
From the first node to the result of a run.
The MCP server covers the full working cycle: discover, build, check, run and understand what happened.
Build and refine
Create a draft, retrieve its definition, update its graph and manage its status from a compatible assistant.
Check before acting
Check the graph structure and branch rules before putting the workflow into action.
Run it for real
Start a workflow manually with optional input data. Your assistant can wait for completion or retrieve the run later, if the API key allows execution.
Read every step
Get the status, output, logs and errors of each node. Browse the workflow’s recent runs too.
One request. One workflow run. A verifiable answer.
Ask your assistant to start an automation. MCP passes the call to nodulz, with input data when needed. The assistant can wait for the result, then inspect the run to explain what succeeded or failed.
Running a workflow is a real action on your instance. It requires an API key with permission to run workflows.
A conversation that actsMCP
- 1
You ask
Run the synchronization workflow for this customer and tell me whether it succeeded.
MCP-compatible assistant
- 2
nodulz runs
The workflow starts with the supplied data. The assistant can wait until the run ends.
workflow_run · input · wait
- 3
The assistant checks
It retrieves status, outputs and any errors to give you a precise answer.
workflow_run_get
Illustrative scenario: content and results depend on your workflow.
A result with the details that matter.
A simple “done” is not enough. MCP lets your assistant look into the workflow’s nodes to understand the run.
The assistant can inspect a specific run or list the latest runs of a workflow. Long values are shortened in the MCP response.
StatusSee whether the run or a node succeeded, failed or is still in progress.
OutputsSee the data produced at each step.
Logs and errorsFind the blocked step and useful diagnostic details.
HistoryBrowse the workflow’s latest runs.
Access stays in your hands.
Create an API key in nodulz with the permissions you need. MCP tools enforce those rights: a design-only key cannot run a workflow without additional permission.
One protocol for your agents.
The nodulz server exposes its tools through MCP. A compatible agent can access workflows, build and run them, and inspect the results without an integration tied to a particular vendor.
Setup depends on the transports and authentication methods each client supports. Access still follows your nodulz API key permissions.
All your compatible clientsMCP
MCP
nodulz
You choose the agent.
Connect your MCP-compatible tool.
The MCP server is available over HTTP through your nodulz instance API. Connect your compatible client, or an MCP Client Tool node in a nodulz workflow. Add the URL and your API key to its configuration.
Connection endpoint
https://your-instance/api/mcpReplace the domain with your instance domain. Authenticate with x-api-key or Authorization: Bearer.
Ready to put ideas in motion?
Explore the visual orchestration behind workflows built with your assistant.