Conduit

MCP server for Conduit — connect any AI agent to your data streams.

📖 API Docs · 🌐 Website · 📦 npm · 💻 GitHub

Quick Start

Get an API key from platform.usecondu.it/tokens, then add to your AI editor config:

Claude Code / Claude Desktop

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "conduit": {
      "command": "npx",
      "args": ["-y", "@useconduit/mcp"],
      "env": {
        "CONDUIT_API_KEY": "conduit_sk_..."
      }
    }
  }
}

Cursor

Add to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "conduit": {
      "command": "npx",
      "args": ["-y", "@useconduit/mcp"],
      "env": {
        "CONDUIT_API_KEY": "conduit_sk_..."
      }
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "conduit": {
      "command": "npx",
      "args": ["-y", "@useconduit/mcp"],
      "env": {
        "CONDUIT_API_KEY": "conduit_sk_..."
      }
    }
  }
}

Environment Variables

Variable	Required	Default	Description
CONDUIT_API_KEY	✅	—	Your Conduit API key
CONDUIT_API_URL	—	https://api.usecondu.it	Custom API endpoint

Tools

Tool	Description
conduit_list_streams	List all data streams
conduit_get_schema	Get schema for a stream (columns, types, codecs)
conduit_create_stream	Create a new stream
conduit_ingest	Send events to a stream
conduit_list_events	Query events with pagination & time range filters
conduit_add_forward	Add a forwarding destination (HTTP, MQTT, or WebSocket) with auth options
conduit_stream_stats	Get ingestion statistics (event count, rate, latency)
conduit_analyze_schema	Analyze a JSON payload for optimal schema
conduit_feedback	Submit feedback to the Conduit team
conduit_backfill	Replay historical events to forwarding destinations
conduit_backfill_status	Check backfill job progress or list all jobs
conduit_time_range	Get earliest/latest event timestamps for a stream

Resources

URI	Description
conduit://streams	All streams
conduit://streams/{name}	Stream details + schema
conduit://stats	Platform-wide statistics

API Documentation

Full interactive API docs are available at api.usecondu.it/docs (powered by Scalar).

Key endpoints:

Method	Endpoint	Description
POST	/v1/{tenant}/{stream}	Ingest events (any protocol: HTTP, WebSocket, MQTT)
GET	/api/v1/streams	List streams
GET	/api/v1/streams/{name}	Stream details + schema
GET	/api/v1/streams/{name}/events?from=&to=&limit=	Query events with time range
GET	/api/v1/streams/{name}/stats	Ingestion statistics
GET	/api/v1/streams/{name}/schema-history	Schema evolution history
POST	/api/v1/streams/{name}/forwards	Add forwarding destination
POST	/api/v1/streams/{name}/backfill	Start backfill job
GET	/api/v1/streams/{name}/backfill	List backfill jobs
GET	/api/v1/streams/{name}/time-range	Event time range
POST	/api/v1/tokens	Create API token
GET	/api/v1/account	Account info + usage

Machine-readable API specs:

• OpenAPI 3.1: api.usecondu.it/openapi.json
• LLM-friendly: api.usecondu.it/llms.txt
• LLM full spec: api.usecondu.it/llms-full.txt

What is Conduit?

Conduit is the lightweight data layer between your services. Send any JSON — schemas aren't defined, they emerge. And they evolve as your sources do.

• One endpoint, any protocol — HTTP, WebSocket, MQTT/S on the same path
• AI-powered schema detection — zero configuration, runs locally (your data never leaves)
• Real-time forwarding — HTTP webhooks, MQTT brokers, WebSocket endpoints with full auth options
• Backfill — replay historical events to new destinations
• Built for agents — MCP-native from day one, feedback loop included
• European infrastructure 🇪🇺 — all data hosted and processed in the EU, GDPR-compliant by design

How agents use Conduit

Agent: "I see live sensor data flowing in. I'll pipe it to our datawarehouse and add hot storage."

▸ conduit_list_events(stream: "sensors", limit: 1000)
  ↳ 1,000 events · 18.4 MB raw → 1.1 MB stored (94% compressed)

▸ conduit_add_forward(stream: "sensors", type: "http", url: "https://dwh.acme.io/ingest")
  ↳ Destination added · forwarding to datawarehouse

▸ conduit_add_forward(stream: "sensors", type: "mqtt", broker: "mqtt://redis.acme.io", topic: "hot/sensors")
  ↳ Destination added · 2 destinations active

▸ conduit_backfill(stream: "sensors", from: "2026-02-24", to: "2026-03-02")
  ↳ Backfilling 89,929 events from last 7 days

Learn more at usecondu.it

License

MIT