Skip to content

README.en.md

Latest commit

 

History

History
128 lines (105 loc) · 2.82 KB

README.en.md

File metadata and controls

128 lines (105 loc) · 2.82 KB

Molta

✨ Use OpenClaw anywhere 🚀

中文

Molta is a lightweight HTTP gateway that forwards OpenAI-style v1/chat/completions requests to a local OpenClaw (ClawdBot) gateway over WebSocket and returns compatible responses, so you can use MoltBot in existing clients.

Features

  • OpenAI-style endpoints: /v1/chat/completions, /v1/models
  • Built-in auth: Bearer token via TOKEN
  • Streaming responses (SSE)
  • Session reuse + quick new-session command

Quick Start

Requires Node.js 20+ (for local dev or npm install)

Option A: Install from npm

npm i -g molta
molta

Option B: Run with Docker

docker run --rm -p 8090:8090 \
  -e TOKEN="<Your token>" \
  -e CLAWD_TOKEN="<Your Clawd Token>" \
  -e CLAWD_HOST="<Clawd host>" \
  -e CLAWD_PORT=<Clawd port> \
  ghcr.io/ve-ria/molta:latest

Option C: Local development (Yarn 4.12.2)

yarn install
yarn dev

Default listening address: http://127.0.0.1:8090.

Environment Variables

Molta reads .env in the current directory and validates it (see schema.json).

Required:

  • TOKEN: HTTP auth token
  • CLAWD_TOKEN: OpenClaw gateway auth token

Optional:

  • HOST: bind host, default localhost
  • PORT: bind port, default 8090
  • CLAWD_HOST: OpenClaw gateway host, default localhost
    • Both Molta and OpenClaw in Docker: use OpenClaw container name or IP
    • Molta in Docker, OpenClaw on host: use host.docker.internal
    • Both on host: use localhost
  • CLAWD_PORT: OpenClaw gateway port, default 18789

Example:

TOKEN="<Your token>"
HOST="127.0.0.1"
PORT=8090
CLAWD_HOST="127.0.0.1"
CLAWD_PORT=18789
CLAWD_TOKEN="<Your Clawd Token>"

API

List models

GET /v1/models

Example response (created is current time):

{
  "object": "list",
  "data": [
    {
      "id": "molta",
      "object": "model",
      "created": "2025-01-01T00:00:00.000Z",
      "owned_by": "molta"
    }
  ]
}

Chat completions

POST /v1/chat/completions

Request body (OpenAI-compatible):

{
  "model": "molta",
  "messages": [
    { "role": "user", "content": "Hello" }
  ],
  "stream": false
}

Auth:

Authorization: Bearer <TOKEN>

Streaming: set stream=true to receive SSE.

Sessions

  • Session ID is derived from user or id (falls back to http)
  • Send /clawd-new or clawd-new to force a new session

Run & Build (dev)

yarn build
yarn start

SEA build (single executable)

Requires Node.js 20+ (22+ recommended)

yarn build:sea

Output: dist/molta-sea (append .exe on Windows). Intermediate files in dist-sea/.

Project Layout

  • src/router/chat/completions.ts: main API logic
  • src/services/gateway.ts: Clawd gateway WebSocket client
  • schema.json: env validation schema