datagrout-mcp

DataGrout MCP tool sub-node for n8n AI Agents — exposes every server tool as a separate agent tool (self-hosted)

Package Information

Released: 2 mo ago
Downloads: 0 weekly / 0 monthly
Latest Version: 0.1.0

Available Nodes

Documentation generating…

DataGrout MCP icon
DataGrout MCP
Give an AI Agent access to every tool on a DataGrout MCP server — each server tool appears as a separate agent tool

Documentation

@datagrout/n8n-nodes-datagrout-mcp

DataGrout MCP — an n8n AI Agent tool sub-node that exposes every tool on
your DataGrout MCP server as a separate agent tool (the same experience as the
built-in Atlassian/Linear MCP entries). No operation selector: just the
credential and a Tools to Include filter.

Self-hosted only. This package ships runtime dependencies
(@langchain/core, zod) and is intentionally not eligible for n8n
verification. For a verified, zero-dependency node, see
@datagrout/n8n-nodes-datagrout.

What the node handles for you

  • Background tasks: slow DataGrout requests are moved to a background task
    server-side; the node waits and returns the finished result — the agent just
    receives the answer.
  • Lean responses (on by default): large result sets return as a short
    preview plus a server-side reference that DataGrout's compute tools accept,
    keeping huge row sets out of your workflow and agent context.
  • Connection reuse: calls share one server session instead of
    re-connecting every time.

Installation (self-hosted)

  1. Install @datagrout/n8n-nodes-datagrout-mcp from Settings → Community Nodes.
  2. Set N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE=true and restart n8n.

Supported n8n versions: 2.29.x (dependencies are pinned to match; other
versions may work but are untested).

Credential — DataGrout API

Same credential as the main package (configure once, use in both):
API Token (from your DataGrout dashboard) + Server ID (UUID) +
Gateway Base URL.

Usage

  1. Add DataGrout MCP to an AI Agent's tool connector.
  2. Pick the credential; choose Tools to Include:
    • All — every tool on the server
    • Selected — only the chosen tools
    • All Except — everything but the chosen tools
  3. Chat — the agent sees each DataGrout tool separately (name, description,
    input schema) and calls them directly.

The filter is the permission boundary: filtered-out tools are never exposed
to or callable by the agent. With large servers (100+ tools), use Selected
to keep the model's context focused.

Click Execute step on the node to preview the tools it will expose.

Resources

License

MIT

Discussion