Most LLM/assistant integrations assume standard webhooks or western messaging APIs; China’s IM ecosystem fragments delivery semantics, media limits, and auth flows. OpenClaw China tackles that friction by presenting a unified channel layer that maps OpenClaw’s assistant replies into platform-specific patterns (WeCom, WeChat MP/kf, QQ, DingTalk, Feishu) so you can reuse the same agent logic across multiple Chinese IM endpoints without re-implementing platform idiosyncrasies.
What Sets It Apart
- Unified channel package: a single OpenClaw extension that registers multiple channel adapters (dingtalk, qqbot, wecom, wecom-app, wecom-kf, wechat-mp, feishu-china) so teams can enable only the channels they need while keeping consistent reply pipeline behavior.
- Production-oriented delivery: handles media upload/download, byte-aware splitting for WeChat message limits, exponential backoff retries, message deduplication and stable local media persistence to avoid transient /tmp loss during downstream processing.
- Platform-specific optimizations: QQ C2C streaming (打字机式实时更新), WeChat MP active/passive delivery modes and ASR integration (Tencent Flash ASR), and WeCom long‑connection mode to avoid public IP requirements for some setups — each adapter exposes pragmatic config flags to match platform constraints.
- Low operational barrier: interactive setup helper and npm packages plus source-mode install steps; documentation lists per-channel guidance and recommended policies (dmPolicy, account mapping, chunk strategies) to reduce common integration mistakes.
Who it's for + tradeoffs
Great fit if you maintain an OpenClaw-based AI assistant and need to expose the same assistant across Chinese IM platforms with minimal per-platform rework. It’s useful for teams shipping AI customer service, internal bots, or multi-channel assistants that must honor platform limits (e.g., WeChat byte limits, QQ markdown quirks) and need reliable media handling.
Look elsewhere if you don’t use OpenClaw as the host (this repo is a plugin collection for OpenClaw specifically), if you need a turnkey hosted SaaS connector, or if your target platforms are outside the Chinese IM ecosystem. Some channels still require public endpoints or platform credentials (service accounts, app secrets), and a few adapters note maintenance/feature gaps — check the repo’s per-channel docs and changelog before production rollout.
