Install Conduit MCP

Pick your client, copy the command, done.

claude mcp add -e "CONDUIT_API_KEY=${CONDUIT_API_KEY}" conduit-0c97 -- npx -y @useconduit/mcp

Required:CONDUIT_API_KEY+ 1 optional

Environment Variables

Set these before running Conduit MCP.

VariableDescriptionRequired

CONDUIT_API_KEYYour Conduit API keyYes

CONDUIT_API_URLCustom API endpointNo

Once configured, Conduit MCP gives your AI agent access to:

conduit_list_streamsList all data streams

conduit_get_schemaGet schema for a stream (columns, types, codecs)

conduit_create_streamCreate a new stream

conduit_ingestSend events to a stream

conduit_list_eventsQuery events with pagination & time range filters

conduit_add_forwardAdd a forwarding destination (HTTP, MQTT, or WebSocket) with auth options

conduit_stream_statsGet ingestion statistics (event count, rate, latency)

conduit_analyze_schemaAnalyze a JSON payload for optimal schema

conduit_feedbackSubmit feedback to the Conduit team

conduit_backfillReplay historical events to forwarding destinations

conduit_backfill_statusCheck backfill job progress or list all jobs

conduit_time_rangeGet earliest/latest event timestamps for a stream

After setup, try these prompts with your AI agent:

→List all available data streams in my Conduit account.

→Analyze the schema for the 'sensors' stream and show me the latest 10 events.

→Set up a new HTTP forward for the 'logs' stream to my data warehouse endpoint.

→Check the ingestion statistics for the 'user-activity' stream.

→Replay historical events from the last 24 hours for the 'transactions' stream.

Prerequisites & system requirements

An MCP-compatible client (Claude Code, Cursor, Windsurf, Claude Desktop, or Codex)
Node.js 18+ with npm/npx installed
CONDUIT_API_KEY — Your Conduit API key

Save the docs, env vars, and workflow around Conduit MCP in Conare so Claude Code, Codex, and Cursor remember it next time.