Skip to content
Back to skills

Cc Connect

BSecurity

Bridge local AI coding agents (Claude Code, Cursor, Gemini CLI, Codex) to messaging platforms (Slack, Telegram, Discord). Use when: controlling AI agents from team chat, sending coding tasks via Slack/Telegram, building team-accessible AI workflows.

  • 142 stars
  • 0 votes
  • 0 copies
  • 6 views
  • Added May 27, 2026
developmentpythongobashsqlnodecode-reviewgitapidocumentation

Works with

  • claude code
  • cursor
  • terminal
  • cli
  • api

Security analysis

B89/100
  • highContains patterns that look like hardcoded API keys
  • mediumInstalls packages at runtime which could introduce malicious dependencies

Pro scans all 2 files and shows the line behind each finding

Scanned May 27, 2026

npx -y skills add TerminalSkills/skills --skill cc-connect --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Cc Connect?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Cc Connect
[![Security: B — Skills Directory](https://www.skillsdirectory.com/api/skills/terminalskills-cc-connect/badge)](https://www.skillsdirectory.com/skills/terminalskills-cc-connect)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: cc-connect
description: >-
  Bridge local AI coding agents (Claude Code, Cursor, Gemini CLI, Codex) to messaging
  platforms (Slack, Telegram, Discord). Use when: controlling AI agents from team chat,
  sending coding tasks via Slack/Telegram, building team-accessible AI workflows.
license: MIT
compatibility: "Node.js 18+ or Python 3.10+"
metadata:
  author: terminal-skills
  version: "1.0.0"
  category: automation
  tags: [ai-agents, messaging, slack, telegram, bridge]
---

# CC-Connect

## Overview

CC-Connect bridges AI coding agents running on your local machine to the messaging platforms your team already uses. Code review, research, automation, data analysis — anything an AI agent can do, now accessible from your phone, tablet, or any device with a chat app.

**Architecture:** Your local AI agent <-> CC-Connect bridge <-> Messaging platform (Slack/Telegram/Discord/etc.)

Send a message in Slack, CC-Connect routes it to your local Claude Code instance, the agent does the work, and the response comes back to your chat.

## Instructions

### Installation

```bash
npm install -g cc-connect
```

### Configuration

Create a `cc-connect.yaml` in your project:

```yaml
agent:
  type: claude-code  # or: codex, gemini, cursor
  workdir: /path/to/your/project

platform:
  type: telegram  # or: slack, discord, feishu, dingtalk
  token: "your-bot-token"
```

### Platform Setup

**Telegram:** Create a bot via @BotFather, get the bot token, and add it to your config.

**Slack:** Create a Slack App at api.slack.com/apps, enable Socket Mode and Event Subscriptions, add Bot Token Scopes (`chat:write`, `app_mentions:read`, `messages.im`), and install to your workspace.

**Discord:** Create an application at discord.com/developers, create a bot, enable Message Content Intent, and invite the bot to your server.

### Starting the Bridge

```bash
cc-connect init    # Interactive wizard for platform credentials
cc-connect start   # Start routing messages
```

### Session Management

```yaml
session:
  timeout: 30m
  max_concurrent: 3
  continue: true
  auto_compress: true
```

### Multi-Agent Routing

Route different commands to different agents:

```yaml
agents:
  code-review:
    type: claude-code
    workdir: /path/to/project
    trigger: "!review"
  research:
    type: gemini
    trigger: "!research"
```

### Access Control

```yaml
access:
  allowed_users: ["U123", "U456"]
  allowed_channels: ["C789"]
  admin_users: ["U123"]
```

## Examples

### Example 1: Team Code Review via Slack

A team sets up CC-Connect to allow engineers to request code reviews from Slack:

```yaml
# cc-connect.yaml
agent:
  type: claude-code
  workdir: /home/dev/acme-api

platform:
  type: slack
  app_token: "xapp-1-A07QX4R..."
  bot_token: "xoxb-[REDACTED]..."
  channels: ["#code-review"]

session:
  timeout: 10m
  auto_compress: true

access:
  allowed_channels: ["#code-review"]
  allowed_users: ["U0381KDLS", "U0492JFMA"]
```

In Slack `#code-review`, an engineer types: `@agent Review the auth module for SQL injection risks`. Claude Code analyzes the code and responds in the thread with findings.

### Example 2: Scheduled Daily Reports via Telegram

A solo developer configures CC-Connect with cron jobs for automated daily standup reports:

```yaml
agent:
  type: claude-code
  workdir: /home/dev/saas-app

platform:
  type: telegram
  token: "7284619035:AAF-kLm9xPqR..."
  allowed_users: ["198274563"]

cron:
  - schedule: "0 9 * * 1-5"
    command: "Summarize yesterday's git commits and open PRs, highlight blockers"
    platform: telegram
    timeout: 5m
    fresh_session: true
```

Every weekday at 9am, the agent generates a summary of recent activity and sends it to the developer's Telegram chat.

## Guidelines

- Start with one messaging platform and get it working before expanding to others
- Always set `allowed_users` in production to restrict access
- Use threads in Slack/Discord to keep conversations organized
- Set `session.timeout` to prevent runaway agent sessions consuming resources
- Enable `auto_compress` for long conversations to prevent context overflow
- Use `fresh_session: true` for cron jobs to avoid inherited context from previous runs
- Verify your setup with `cc-connect status` if messages are not routing
- See the [GitHub Repository](https://github.com/chenhg5/cc-connect) for full documentation

Files in this skill

  • SKILL.md4.3 KB
  • _scores.json1.3 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.