Skip to main content

Control your AI coding agents from your phone when you're away from your desk

Project description

Tether Agent

Control your AI coding agents from your phone when you're away from your desk.

You start a coding agent, walk away for lunch, and come back to find it stuck waiting for input for an hour. Tether fixes that. Get notified when your agent needs you, respond from anywhere.

Features

  • Local-first — Runs on your machine, your data stays yours
  • Multi-agent — Supports Claude and Codex, more to come
  • Web UI — Monitor sessions from your phone or desktop
  • External Agent API — WebSocket and REST API for connecting any AI agent
  • Messaging Platform Integrations — Telegram, Slack, and Discord bridges
  • MCP Server — Model Context Protocol server for Claude Desktop and other MCP clients
  • No API keys required — Uses Claude / Codex local OAuth by default

Installation

pip install tether-ai

Optional Platform Integrations

Install platform-specific dependencies as needed:

# Telegram bridge
pip install tether-ai[telegram]

# Slack bridge
pip install tether-ai[slack]

# Discord bridge
pip install tether-ai[discord]

# All bridges
pip install tether-ai[telegram,slack,discord]

# Development tools
pip install tether-ai[dev]

Quick Start

Run the Agent Server

# Start the server
tether start

Then open http://localhost:8787 in your browser.

Use with MCP (Claude Desktop)

Add Tether as an MCP server in your Claude Desktop config:

{
  "mcpServers": {
    "tether": {
      "command": "tether-mcp",
      "env": {
        "TETHER_API_URL": "http://localhost:8787"
      }
    }
  }
}

Configuration

Set environment variables to configure:

Variable Description Default
TETHER_AGENT_HOST Host to bind to 0.0.0.0
TETHER_AGENT_PORT Port to listen on 8787
TETHER_AGENT_TOKEN Auth token (optional; if set, API/UI/MCP require bearer auth)
TETHER_AGENT_DEV_MODE Enable dev mode (no token required) 0
TETHER_DEFAULT_AGENT_ADAPTER AI adapter to use claude_auto
TETHER_AGENT_DATA_DIR Data storage directory ./data

AI Adapters

Adapter Description
claude_auto Auto-detect (prefer OAuth, fallback to API key)
claude_subprocess Claude via Agent SDK in subprocess (CLI OAuth)
claude_api Claude via API key (set ANTHROPIC_API_KEY)
codex_sdk_sidecar Codex via sidecar
opencode OpenCode via auto-managed sidecar

Messaging Platform Bridges

Configure bridges to get notifications on your preferred platform:

Telegram

export TELEGRAM_BOT_TOKEN="your_bot_token"
export TELEGRAM_FORUM_GROUP_ID="your_group_id"

Slack

export SLACK_BOT_TOKEN="xoxb-your-token"
export SLACK_CHANNEL_ID="C01234567"

Discord

export DISCORD_BOT_TOKEN="your_bot_token"
export DISCORD_CHANNEL_ID="1234567890"

External Agent API

Tether exposes a WebSocket and REST API for external agents to connect and interact with users through messaging platforms.

REST API Endpoints

Create a Session

POST /external/sessions
Content-Type: application/json

{
  "agent_metadata": {
    "name": "My Custom Agent",
    "type": "custom",
    "icon": "🤖",
    "workspace": "my-workspace"
  },
  "session_name": "Code Review Task",
  "platform": "telegram"
}

Response:

{
  "session_id": "sess_abc123",
  "platform": "telegram",
  "thread_info": {
    "thread_id": "123456",
    "platform": "telegram"
  }
}

Send Output

POST /external/sessions/{session_id}/output
Content-Type: application/json

{
  "text": "Agent output text here",
  "metadata": {}
}

Request Approval

POST /external/sessions/{session_id}/approval
Content-Type: application/json

{
  "title": "Approve Changes?",
  "description": "Ready to commit these changes",
  "options": ["Approve", "Reject", "Review"]
}

Check for Input

GET /external/sessions/{session_id}/input?timeout=30

Response:

{
  "type": "human_input",
  "data": {
    "text": "User's message",
    "timestamp": "2025-01-01T12:00:00Z"
  }
}

WebSocket API

Connect to /external/sessions/{session_id}/ws for bidirectional communication:

Agent → Tether events:

  • output: Send text output to user
  • approval_request: Request user approval
  • status: Update agent status (thinking, executing, done, error)

Tether → Agent events:

  • human_input: User sent a message
  • approval_response: User responded to approval request

Example WebSocket message:

{
  "type": "output",
  "data": {
    "text": "Processing your request...",
    "metadata": {}
  }
}

Development

Run Tests

pytest tests/

Database Migrations

# Create a new migration
alembic revision --autogenerate -m "description"

# Apply migrations
alembic upgrade head

Documentation

For full documentation, see github.com/larsderidder/tether.

License

Apache 2.0. See LICENSE for details.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

tether_ai-0.3.0.tar.gz (344.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

tether_ai-0.3.0-py3-none-any.whl (307.5 kB view details)

Uploaded Python 3

File details

Details for the file tether_ai-0.3.0.tar.gz.

File metadata

  • Download URL: tether_ai-0.3.0.tar.gz
  • Upload date:
  • Size: 344.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for tether_ai-0.3.0.tar.gz
Algorithm Hash digest
SHA256 4c7dad434525dcb8acb35529cac72a65f816b37852c0aa3f5258dec969a9ef75
MD5 71eb3ac71504aee7818700b7097a9bcb
BLAKE2b-256 7869d3d5536936705e7feb9e19fe1ec5cbd95f39e43e2d7a7a16ce62ec124afa

See more details on using hashes here.

Provenance

The following attestation bundles were made for tether_ai-0.3.0.tar.gz:

Publisher: publish.yml on larsderidder/tether

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file tether_ai-0.3.0-py3-none-any.whl.

File metadata

  • Download URL: tether_ai-0.3.0-py3-none-any.whl
  • Upload date:
  • Size: 307.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for tether_ai-0.3.0-py3-none-any.whl
Algorithm Hash digest
SHA256 40020f6886a6130025a22dcda4c78d57f852d2b9b1b616213e08dc0cd4b1b2fc
MD5 149588682170f74348e1d71b413ab870
BLAKE2b-256 659c9e4d99c2827d2ac5f5074d3f24730f5f84fc98220814159d75eca0156ce0

See more details on using hashes here.

Provenance

The following attestation bundles were made for tether_ai-0.3.0-py3-none-any.whl:

Publisher: publish.yml on larsderidder/tether

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page