chenhg5/cc-connect

What is cc‑connect?

cc‑connect is an open‑source bridge that lets you run AI agents (e.g., Kimi, Claude, GitHub Copilot CLI, Cursor, OpenCode, Qoder, Pi, etc.) on your own computer and control them from any instant‑messaging platform you already use – Feishu/Lark, DingTalk, Telegram, Slack, Discord, WeCom, QQ, and more. In practice you can chat with the bot in a group or DM, ask it to review code, run a command, fetch data, or automate a workflow, and the request is forwarded to the local agent which replies back through the same chat.

---

Key Features (as described in the README)

Feature	What it means for you
Multi‑platform support	One connector works with dozens of chat services (Slack, Discord, Telegram, Feishu, DingTalk, QQ, etc.). No need to run separate bots for each platform.
First‑class AI agents	Out‑of‑the‑box support for many popular local agents: Devin CLI, Google Antigravity (agy), GitHub Copilot CLI, Cursor, OpenCode, Qoder, Kimi, Pi, and more. Adding a new agent is a simple plug‑in.
Local execution	The heavy LLM inference runs on your own machine (or on a private server you control). The bridge only forwards messages, keeping your data private.
Rich interactions	Agents can inspect or modify code, run shell commands, troubleshoot errors, perform data analysis, or trigger any automation you have scripted.
Stable releases	v1.3.3 is the first stable release of the 1.3.3 series, consolidating hundreds of pull‑requests and fixing beta‑stage bugs.
Community & sponsorship	Active Discord/Telegram groups, plus a list of sponsors that provide discounted API access to large‑model providers (Kimi, Claude, OpenAI, etc.).
Package distribution	Published on npm (cc-connect) and as a Go module, with CI badges, Go report card, and versioned releases.

---

How it works (high‑level architecture)

1. Connector daemon runs on your computer. It listens for incoming webhook events from the chat platform you configure.
2. When a message arrives, the daemon forwards the text to the selected local AI agent via its CLI or API.
3. The agent processes the request (e.g., runs a code‑review command) and returns a response.
4. The daemon sends the response back to the chat platform, appearing as a normal chat reply.

The README includes a diagram (docs/images/connector.png) that visualises this flow.

---

Getting started (quick steps from the README)

1. Install – npm i -g cc-connect (or build from source with Go).
2. Configure – create a config file linking a chat platform (provide bot token/webhook URL) and select the local agent you want to expose.
3. Run – start the daemon (cc-connect serve). Your bot will appear in the chosen chat app.
4. Interact – type commands like @bot review myfile.go or @bot run tests and watch the local agent handle them.

---

Who might use it?

• Developers who want to keep their LLM‑powered coding assistants on‑premise but still chat with them from mobile or desktop messengers.
• Teams that need a shared AI assistant across many communication tools without deploying separate services.
• Researchers/ hobbyists experimenting with new agents (Devin, Antigravity, etc.) and wanting a simple way to invoke them from chat.

---

License & Community

• Licensed under the MIT License.
• Active Discord and Telegram groups for support.
• The project is sponsored by a range of AI‑model API providers (Kimi, APINebula, Qiniu, etc.), offering users discounted access to large models.

---

In short: cc‑connect is a practical glue layer that makes locally‑run AI agents reachable from any chat app, turning your phone or Slack channel into a remote control panel for code review, automation, and other AI‑driven tasks.