npm stats
  • Search
  • About
  • Repo
  • Sponsor
  • more
    • Search
    • About
    • Repo
    • Sponsor

Made by Antonio Ramirez

teemux

1.9.1

@GitHub Actions

npmHomeRepoSnykSocket
Downloads:151
$ npm install teemux
DailyWeeklyMonthlyYearly

teemux

teemux

Aggregate logs from multiple processes in a single view

View in browser or terminal • Filter with patterns • Stream to AI agents via MCP

npm version npm downloads license

teemux demo

View a demo of Teemux in action


Demo

npx teemux -- curl -N https://teemux.com/random-logs

Motivation

  • Needed a simple way to browse logs aggregated across multiple processes.
  • Needed a simple way to give agents a unified view of all the logs

Install

npm install -g teemux

Usage

teemux --name api -- node api.js
teemux --name worker -- node worker.js
teemux -- redis-server  # name defaults to "redis-server"

The first process starts a local server on port 8336. Others connect automatically.

Options

OptionAliasDefaultDescription
--name-ncommand nameIdentifier for this process in logs
--port-p8336Port for the log aggregation server
--buffer-b10000Number of log lines to keep in server buffer
--force-leader-ffalseForce this process to become the leader, replacing any existing leader

All options can also be set via environment variables with TEEMUX_ prefix:

TEEMUX_PORT=9000 teemux -- node app.js

Viewing Logs

Browser

Open http://127.0.0.1:8336/ to view aggregated logs with:

  • Color-coded process names
  • Auto-scroll (sticks to bottom like a terminal)
  • Scroll up to pause, scroll back down to resume

Terminal / curl

curl http://127.0.0.1:8336/

Plain text stream of all logs.

MCP Server for AI Agents

teemux includes a built-in Model Context Protocol (MCP) server that enables AI agents to programmatically access logs in your development environment. This makes it easy for coding assistants like Claude, Cursor, or other AI tools to inspect application logs, search for errors, and understand what's happening in your running processes.

The MCP server runs on the same port as the HTTP server at /mcp and provides these tools:

ToolDescriptionParameters
get_logsGet recent logs from bufferlimit?, include?, exclude?
search_logsSearch logs with patternslimit?, include?, exclude?
clear_logsClear the log buffernone
get_process_namesList all process names that have loggednone

Configuring your AI agent

Add teemux as an MCP server in your AI tool's configuration. For Claude Code, add to your MCP settings:

{
  "mcpServers": {
    "teemux": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://127.0.0.1:8336/mcp"
      ]
    }
  }
}

Once configured, your AI agent can:

  • Inspect logs when debugging issues ("What errors are in the logs?")
  • Search for specific events ("Find all database connection errors")
  • Monitor processes ("What processes are currently running?")
  • Clear logs to start fresh when testing

Example MCP usage

# Initialize session
curl -X POST http://127.0.0.1:8336/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","method":"initialize","params":{"protocolVersion":"2025-03-26","capabilities":{},"clientInfo":{"name":"test","version":"1.0"}},"id":1}'

# List available tools (use session ID from response above)
curl -X POST http://127.0.0.1:8336/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "mcp-session-id: <session-id>" \
  -d '{"jsonrpc":"2.0","method":"tools/list","params":{},"id":2}'

# Get recent logs
curl -X POST http://127.0.0.1:8336/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "mcp-session-id: <session-id>" \
  -d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"get_logs","arguments":{"limit":50}},"id":3}'

Filtering Logs

Use query parameters to filter logs:

ParameterLogicDescription
includeORShow lines matching any of the patterns
excludeORHide lines matching any of the patterns

Patterns support * as a wildcard (matches any characters):

# Show only logs from the api process
curl "http://127.0.0.1:8336/?include=api"

# Show only error logs (using wildcard)
curl "http://127.0.0.1:8336/?include=*error*"

# Show logs from api OR worker
curl "http://127.0.0.1:8336/?include=api,worker"

# Hide healthcheck and ping logs
curl "http://127.0.0.1:8336/?exclude=health*,ping"

# Show GET requests to /api endpoints
curl "http://127.0.0.1:8336/?include=*GET*/api*"

# Show api logs but exclude verbose debug output
curl "http://127.0.0.1:8336/?include=api&exclude=DEBUG,TRACE"

# In browser
open "http://127.0.0.1:8336/?include=api&exclude=health*"

Filters apply to both buffered logs and new incoming logs in real-time.

Output Example

Terminal (where teemux runs):

● started (pid 12345)
Server listening on :3000
Processing jobs...
GET /health 200
○ exited (code 0)

Browser / curl (aggregated with prefixes):

[api] ● started (pid 12345)
[api] Server listening on :3000
[worker] Processing jobs...
[api] GET /health 200
[worker] ○ exited (code 0)

FAQ

What's the origin of the name?

The name combines tee (the Unix command that duplicates output) and mux (multiplexer) – it multiplexes multiple log streams into one.

How does teemux work?

teemux uses automatic leader discovery to coordinate log aggregation across multiple processes:

  1. Leader Discovery: When the first teemux process starts, it attempts to bind to the configured port (default 8336). If successful, it becomes the leader and starts the log aggregation server.

  2. Client Registration: When subsequent teemux processes start, they detect the port is already in use, verify a server is responding, and automatically become clients that forward their logs to the leader.

  3. Leader Election: If the leader process exits, clients detect this through periodic health checks (every 2 seconds). When a client detects the leader is gone, it attempts to become the new leader. Random jitter prevents multiple clients from racing to claim leadership simultaneously.

This design requires no configuration – just run multiple teemux commands and they automatically coordinate.

Docker output appears corrupted with strange spacing

When running Docker with the -t flag, output may appear corrupted:

Initializing database...
                        The files belonging to this database system...

Cause: The -t flag allocates a pseudo-TTY, which adds terminal control sequences (cursor positioning, colors, etc.) to the output. These sequences are meant for interactive terminal use, not for piping.

Solution: Remove the -t flag when running through teemux:

# ❌ Don't use -t
teemux --name db -- docker run --rm -it my-database

# ✅ Use -i only (or neither flag)
teemux --name db -- docker run --rm -i my-database
teemux --name db -- docker run --rm my-database

The flags:

  • -i = keep stdin open (for interactive input) ✅
  • -t = allocate pseudo-TTY (adds terminal formatting) ❌

Developing

# Install dependencies
npm install

# Build and watch for changes
npm run dev

# In another terminal, run with fake logs
node dist/teemux.js -- node scripts/fake-logs.js