openclaw-china-docker: A Docker Image for OpenClaw with Chinese IM Platform Plugins

What openclaw-china-docker Solves

OpenClaw is an AI agent framework, but configuring it for Chinese IM platforms requires installing and configuring separate plugins for each platform. openclaw-china-docker packages all of this into a single Docker image. The image includes pre-installed plugins for Feishu (official team plugin and an older built-in version), DingTalk, QQ Bot, WeCom, and WeChat. It also integrates Agent Reach, which extends the bot's reach to Twitter, Xiaohongshu, Weibo, Douyin, and other platforms.

The primary users are teams in China who want to deploy a unified AI bot gateway for multiple IM channels without performing individual plugin installations on each deployment. The README describes this as an out-of-the-box OpenClaw experience with deep enhancements for domestic IM scenarios.

Deploying the Image with Docker Compose

The image is available on Docker Hub at `justlikemaki/openclaw-docker-cn-im:latest`. Pull it with:

docker pull justlikemaki/openclaw-docker-cn-im:latest

The recommended deployment path is to clone the repository, copy `.env.example` to `.env`, fill in the required values, and then start the stack:

docker compose up -d

The README recommends cloning the full repository rather than downloading only `docker-compose.yml` and `.env.example` separately, because subsequent upgrades require syncing the README, docker-compose.yml, `.env.example`, and any new documentation files. The `.env.minimal` file is available as a shorter alternative to `.env.example` for minimal configurations.

Configuration Through Environment Variables

The `.env.example` file is the central configuration document. It controls the model endpoint, API key, and AI protocol in use. The `BASE_URL` and `API_KEY` variables point to the AI model gateway. The `API_PROTOCOL` variable selects the protocol: `openai-completions`, `openai-responses`, or `google-generative-ai`. Up to six model providers are configurable as MODEL through MODEL6, each with their own base URL, API key, and protocol.

Plugin synchronization on startup is controlled by three variables: `SYNC_EXTENSIONS_ON_START` enables or disables it, and `SYNC_EXTENSIONS_MODE` selects the sync strategy. The `seed-version` mode (recommended) updates only when the image's internal seed version has changed. The `overwrite` mode replaces all plugins on every startup, and `missing` adds only plugins that are absent from the extensions volume.

The `SYNC_OPENCLAW_CONFIG` variable controls whether environment variables are written into the `openclaw.json` configuration file automatically. Teams who maintain `openclaw.json` manually should set this to `false` to prevent the container from overwriting their changes.

The Docker Sandbox Mode for Code Execution

The image supports a Docker-in-Docker sandbox mode for running Python code and shell scripts in isolation from the host. To enable it, set `OPENCLAW_SANDBOX_MODE=all` (or `non-main`) in the `.env` file and uncomment the `/var/run/docker.sock` mount line in `docker-compose.yml`.

The README notes this feature uses an official minimal Docker image with no usage examples provided. The trade-off of mounting `/var/run/docker.sock` is significant: it gives the container full access to the host Docker daemon, meaning a compromised container could control the host's Docker environment. The README lists this as a trade-off without recommending a specific mitigation.

Bundled Tools Beyond IM Plugins

The Dockerfile installs several tools beyond the IM plugins. The base image is `node:22-slim` with Python 3.12 copied from `python:3.12-slim-bookworm`. Additional tools installed globally include opencode-ai, clawhub, Playwright, playwright-extra, puppeteer-extra-plugin-stealth, and `@steipete/bird`. FFmpeg is installed via apt for audio and video processing. Chinese CJK fonts (`fonts-noto-cjk`) and emoji fonts are included for display compatibility.

The npm registry is set to `https://registry.npmmirror.com` inside the Dockerfile, which routes package downloads through a Chinese mirror. This is a practical choice for deployment inside China but means the build will use the npmmirror copy of each package rather than the canonical npm registry version.

Bun and uv (a Python package manager) are installed for additional runtime flexibility. The Playwright browser installation targets Chromium with its dependencies, supporting the Agent Reach integrations that use browser automation.

Limitations and the Suspended Update Notice

The README begins with a header stating that updates have been suspended ("暂停更新"). The last push was on 2026-04-14. Teams adopting this image should treat it as a snapshot, not an actively maintained project.

The image's npm install command pins `[email protected]` by version, which means bug fixes or security patches to OpenClaw after that version are not included unless the Dockerfile is updated manually. The project's GPL-3.0 license allows forking and modification, so teams with operational requirements can maintain their own build.

Agent Reach, the component that enables Twitter, Xiaohongshu, and similar integrations, is a separate repository (Panniantong/Agent-Reach). Its availability and maintenance status are independent of this project.

openclaw-china-docker Versus Building from the Upstream OpenClaw Image

The upstream OpenClaw project provides its own Docker image without the Chinese IM plugins included. The difference is one of pre-integration versus self-assembly: the upstream image requires users to install and configure each IM plugin separately, while openclaw-china-docker pre-installs Feishu, DingTalk, QQ Bot, WeCom, and WeChat plugins and provides a configuration-driven environment variable system for quick setup.

For teams outside China or teams using only one or two of the supported platforms, the upstream OpenClaw image with manual plugin installation may be preferable, since it avoids the weight of plugins that are not needed. For teams in China deploying across multiple IM platforms simultaneously, the pre-integrated approach saves significant configuration time on each new deployment.

The AGPL-3.0 license on the upstream OpenClaw and the GPL-3.0 license on this project both require distributing source modifications when deployed as a network service.

Editorial conclusion

openclaw-china-docker is appropriate for teams in China who need an AI bot gateway for Feishu, DingTalk, QQ Bot, WeCom, or WeChat and want to avoid assembling these integrations by hand. Before deploying, read the .env.example file carefully: it requires valid API keys, model endpoint URLs, and IM platform credentials. The README also notes that updates have been suspended, and the last push was on 2026-04-14. Teams who cannot accept a stalled upstream should plan to fork and maintain the Dockerfile themselves. The Docker sandbox mode (OPENCLAW_SANDBOX_MODE=all) requires uncommenting the /var/run/docker.sock mount in docker-compose.yml, which grants the container access to the host Docker daemon.

Frequently asked questions

Official sources

1. Issues
2. justlovemaki/openclaw-china-docker on GitHub
3. License: GPL-3.0
4. README