33GOD Pipeline MCP Hub
A multi-domain MCP server that exposes a small, stable surface of three tools (list_domains, list_domain_tools, call_domain_tool) to gate access to various tool domains like Plane and BloodBank, preventing agent schema overload.
README
33GOD Pipeline MCP Hub
A multi-domain MCP server that fronts many tool domains (Plane, BloodBank, ticket-lifecycle, …) behind one small, stable surface — so an agent isn't flooded with hundreds of tool schemas / millions of tokens.
Served at https://mcp.delo.sh/mcp (header-PAT auth).
How it works — discovery + dispatcher gating
The hub exposes only three tools to clients:
| Tool | Purpose |
|---|---|
list_domains() |
List available domains (name, description, tool_count). Start here. |
list_domain_tools(domain) |
Load + list one domain's tools, each with its input_schema. This is the "load a domain's toolset" step — schemas enter context only now. |
call_domain_tool(domain, tool, arguments) |
Invoke a specific domain tool; arguments must match its input_schema. |
Underlying domain tools are held in an internal registry and are never advertised
directly, so the per-request tool-schema cost stays tiny and constant. This works on
any MCP client regardless of notifications/tools/list_changed support.
Trade-off: domain tools are invoked through
call_domain_tool(...)rather than as native function calls. The hub validates arguments against the real tool schema and returns clear errors. (A future "native passthrough" mode could expose real prefixed tools for clients that already defer schemas, like Claude Code.)
Domains
- plane — reuses
plane-mcp-server's tools in-process (work items, projects, cycles, epics, etc.). Auth + Plane API calls use the caller's PAT from the request context. - bloodbank (Phase 1) — publish/trace events on the BloodBank NATS bus.
- lifecycle (Phase 2) — opinionated ticket state machine (pull → triage → … → done).
Auth
HTTP mode uses the same header scheme as plane-mcp-server:
Authorization: Bearer <PLANE_PAT>x-workspace-slug: <workspace-slug>
Run locally
# Install the hub + the plane domain dependency (from the local checkout)
uv pip install -e .
uv pip install -e ../../plane-mcp-server
# stdio (no auth; Plane tools use env fallback)
PLANE_API_KEY=... PLANE_WORKSPACE_SLUG=... python -m mcp_hub stdio
# http (header-PAT auth) on :8998 at /mcp
PLANE_INTERNAL_BASE_URL=http://localhost:8000 python -m mcp_hub http
Deploy (DeLoNET)
Docker stack at ~/docker/stacks/ai/pipeline-mcp-hub/. It builds this repo and pulls the
plane-mcp-server source via a named build context, joins the proxy network, and is
routed by Traefik at Host(mcp.delo.sh). *.delo.sh already wildcards through the
Cloudflare tunnel → Traefik.
cd ~/docker/stacks/ai/pipeline-mcp-hub
docker compose up -d --build
Requires Docker Compose ≥ v2.17 (for build.additional_contexts).
Key env
| Var | Default | Purpose |
|---|---|---|
HUB_PORT |
8998 |
Listen port inside the container |
PLANE_INTERNAL_BASE_URL |
— | Server-to-server Plane API base; set to http://plane-backend:8000 (SDK appends /api/v1) |
PLANE_BASE_URL |
https://api.plane.so |
Fallback Plane API base |
Phase 0 needs no secrets — the Plane PAT arrives per-request in the client's headers.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.