# JustinX

> Real-time data streaming infrastructure for AI agents. Connect any data source
> (MQTT, Webhook, Kafka) and stream it through Redis into WebSockets, with
> server-side TypeScript watchers for alerts, anomaly detection, and aggregation.
> Designed for MCP integration with Claude Code, Cursor, and any MCP-compatible client.

JustinX is a hosted service at https://justinx.ai. It exposes a set of MCP tools that
let AI agents create data connections, inspect live streams, and deploy server-side
processing scripts (watchers) — all without leaving the coding environment.

Architecture: Data Source -> Redis Stream -> WebSocket relay -> Watchers

## Instructions for AI Agents

- Use `create_connection` to connect a data source. Three types are supported:
  - **MQTT**: Connect to any MQTT broker (e.g. HiveMQ, EMQX, Mosquitto)
  - **Webhook**: Creates an HTTP ingest endpoint — external systems POST data to it
  - **Kafka**: Connect to a Kafka cluster via a consumer group
- After creating a connection, use `read_stream` to sample live data before building anything
- The WebSocket URL returned by `create_connection` can be passed to any consumer — a generated app, a script, a browser page — to receive live data directly
- Watchers are server-side TypeScript processes for background logic. Deploy them with `create_watcher`
- Watcher scripts receive `WATCHER_ID`, `WATCHER_CONFIG`, and `WATCHER_CONNECTION_ID` as environment variables
- Always prefer `update_watcher_config` over `delete_watcher` + `create_watcher` when tuning parameters
- Use `get_watcher_logs` to debug crashed or misbehaving watchers
## MCP Tools

- [create_connection](/llms-full.txt): Create a data connection. For MQTT: connects to a broker and subscribes to topics. For Webhook: creates an HTTP ingest endpoint. For Kafka: connects to a Kafka broker via a consumer group and consumes from topics. All stream through Redis into a WebSocket.
- [list_connections](/llms-full.txt): List all active connections and their status.
- [get_connection](/llms-full.txt): Get status of a specific connection (message count, WebSocket clients, stream key).
- [destroy_connection](/llms-full.txt): Destroy a connection — stops source, cleans up Redis stream, kills watchers.
- [read_stream](/llms-full.txt): Read entries from a connection's live stream. Reads backfill + live entries directly from Redis, then returns them. Use this to inspect what data is flowing through a connection.
- [create_watcher](/llms-full.txt): Deploy a watcher script on a connection. The script runs as a child process with access to the Redis stream. It receives WATCHER_ID, WATCHER_CONFIG, and WATCHER_CONNECTION_ID as environment variables.
- [list_watchers](/llms-full.txt): List all watchers for a connection with their status, PID, and restart count.
- [get_watcher](/llms-full.txt): Get details of a specific watcher including its source code, status, config, and restart count.
- [get_watcher_logs](/llms-full.txt): Get stdout/stderr logs from a running or crashed watcher.
- [update_watcher_config](/llms-full.txt): Update a watcher's config. The watcher process will be restarted with the new config.
- [restart_watcher](/llms-full.txt): Restart a stopped or crashed watcher. Sets the watcher status back to running so it will be re-spawned by the reconciliation loop.
- [delete_watcher](/llms-full.txt): Stop and remove a watcher from a connection.

## Getting Started

- [Full Tool Reference with Parameters](/llms-full.txt): Auto-generated detailed reference for all MCP tools
- [Sign Up](https://justinx.ai/signup): Create an account and get an API key
- [Dashboard](https://justinx.ai/dashboard): Manage connections and watchers in the UI

## Optional

- [Pricing](https://justinx.ai/pricing): Plans and pricing
- [Contact](https://justinx.ai/product): Product details and contact form