AgentLink is a high-performance Rust bridge that seamlessly connects chat channels to AI programming agents, enabling natural language coding workflows across your favorite messaging apps.
Connect to 15+ chat channels with native adapters:
- Enterprise: Feishu/Lark (WebSocket), DingTalk, Slack (Socket Mode), WeCom
- Social: Telegram (Long Polling), Discord (Gateway), LINE, Weibo
- Messaging: WeChat Personal (iLink), QQ/OneBot, QQ Official Bot, MAX
- Extensible: HTTP Webhook & WebSocket Bridge for custom integrations
Unified interface for popular programming agents:
- Codex (exec & app-server modes)
- Claude Code / Cursor / Gemini CLI
- OpenCode / Kimi / Qoder / iFlow
- Generic CLI adapter for any command-line agent
Tauri 2.0 + Vue 3 cross-platform GUI:
- ๐จ Modern, responsive interface
- ๐ QR code scanning for instant platform setup
- โ๏ธ Visual configuration editor with validation
- ๐งช Built-in message testing with rich media support
- ๐ Real-time service status monitoring
GitHub Actions powered CI/CD:
- โ
Automatic builds on every push to
main - ๐ฆ Platform-specific installers (MSI, DMG, AppImage, DEB)
- ๐ท๏ธ Tagged releases with semantic versioning
- ๐ง Optimized icon configuration for all platforms
- Session Management: SQLite-backed persistent sessions
- Approval Workflow: Built-in
/allowand/denycommands - Rich Media: Images, files, audio, video, location, cards
- Security: Token authentication, signature verification
- Reliability: Message queuing, error recovery, timeout handling
Download the latest release for your platform:
Windows
# Download from GitHub Releases
# Extract and run agentlink.exemacOS
# Download the DMG from GitHub Releases
# Drag AgentLink.app to ApplicationsLinux
# Download AppImage or DEB package
chmod +x AgentLink-*.AppImage
./AgentLink-*.AppImage- Launch the desktop client to configure your first chat channel
- Scan QR code for instant Feishu/Lark/WeChat setup
- Configure your agent (Codex, Claude Code, etc.)
- Start chatting with your AI programming assistant!
- Rust 1.70+
- Node.js 18+
- Platform-specific dependencies (see below)
# Clone the repository
git clone https://github.com/doit9816/AgentLink.git
cd AgentLink
# Build the core bridge
cargo build --release
# Validate a config file
./target/release/agentlink validate-config examples/agentlink.1.toml
# Build the desktop client
cd client/agentlink-desktop
npm install
npm run build
# Outputs:
# - CLI: ../target/release/agentlink
# - Desktop bundles: src-tauri/target/release/bundle/Linux
sudo apt-get install -y \
libwebkit2gtk-4.1-dev \
build-essential \
libssl-dev \
libayatana-appindicator3-dev \
librsvg2-devmacOS
# Xcode Command Line Tools required
xcode-select --installWindows
# Visual Studio Build Tools or MSVC required
# WebView2 runtime (usually pre-installed on Windows 10/11)Windows (All-in-One)
powershell -NoProfile -ExecutionPolicy Bypass -File .\scripts\auto-package.ps1This script:
- โ Runs all tests
- ๐จ Builds core bridge in release mode
- ๐ฅ๏ธ Builds Tauri desktop client
- ๐ฆ Generates platform installers (MSI, NSIS)
- ๐๏ธ Creates distribution archive
macOS/Linux
bash ./scripts/package-unix.sh <version> <target-name>Our CI/CD pipeline automatically builds and publishes releases:
On manual dispatch
- Triggers the cross-platform desktop build workflow without changing the published updater feed
On Version Tag (v*)
git tag v0.2.0
git push origin v0.2.0- Creates stable release
- Generates release notes
- Publishes platform-specific installers:
- Windows:
.zip,.msi,.exe(NSIS) - macOS:
.tar.gz,.dmg - Linux:
.tar.gz,.AppImage,.deb
- Windows:
- Publishes signed desktop installers and
latest.jsonto GitHub Releases (used by the in-app updater)
Each release includes:
- ๐ฆ Core Bridge: Standalone CLI executable
- ๐ฅ๏ธ Desktop Client: Platform-specific installer
- ๐ Documentation: README and examples
- ๐ง Configuration Templates: Sample TOML files
powershell -NoProfile -ExecutionPolicy Bypass -File .\scripts\auto-package.ps1The package script builds the core bridge, the desktop client, and the Windows release archive.
The desktop client's Tauri updater reads latest.json from GitHub Releases. Installers are release assets; only metadata is fetched for update checks:
https://github.com/doit9816/AgentLink/releases/latest/download/latest.json
tauri-action uploads per-platform signatures and download URLs pointing at the same release's assets.
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Chat Channels โ
โ Feishu โ Slack โ Telegram โ Discord โ WeChat โ QQ โ ... โ
โโโโโโโโโโโโโโโโโโโโโโฌโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ
โผ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Channel Adapters โ
โ (Native WebSocket/HTTP/Long Polling) โ
โโโโโโโโโโโโโโโโโโโโโโฌโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ
โผ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ AgentLink Engine โ
โ โข Session Management โข Message Routing โ
โ โข Approval Workflow โข Rich Media Handling โ
โโโโโโโโโโโโโโโโโโโโโโฌโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ
โผ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ Agent Adapters โ
โ Codex โ Claude Code โ Gemini โ Cursor โ Generic CLI โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
The Rust workspace has been refactored from a flat src/*.rs layout into feature-oriented modules:
src/
agents/ Codex integration and generic CLI agent presets
app/ config loading and registry assembly
channels/ native chat channels plus Bridge/HTTP adapters
core/ shared traits, messages, approvals, and session types
engine/ routing, session serialization, approval flow, final replies
setup/ setup commands such as Feishu/Lark bootstrap helpers
store/ SQLite-backed persistence
testing/ mock platform/agent helpers for e2e and integration tests
lib.rs public exports for embedding and tests
main.rs CLI entrypoint and default runtime wiring
src/lib.rs still re-exports the main public modules, so existing imports can keep working while new code moves to the namespaced layout.
- Team Collaboration: Let your team interact with AI agents through familiar chat apps
- Code Review: Get instant code suggestions and reviews in Slack/Discord channels
- DevOps Automation: Trigger deployments and infrastructure changes via chat commands
- Learning & Onboarding: Help new developers with interactive coding assistance
- Multi-Channel Support: Reach users on their preferred messaging channel
AgentLink uses TOML configuration files for maximum flexibility:
[[projects]]
name = "my-project"
default_platforms = ["feishu", "slack"]
[[projects.platforms]]
id = "feishu"
type = "feishu"
connection_mode = "websocket" # No public IP required!
[projects.platforms.options]
app_id = "cli_xxx"
app_secret = "xxx"
[projects.agent]
type = "codex"
[projects.agent.options]
work_dir = "/path/to/project"
backend = "exec"
mode = "suggest"
model = "gpt-5.2"See examples/ directory for complete configuration templates.
We welcome contributions! Here's how you can help:
- ๐ Report Bugs: Open an issue with reproduction steps
- ๐ก Suggest Features: Share your ideas in discussions
- ๐ง Submit PRs: Fix bugs or add new features
- ๐ Improve Docs: Help others understand AgentLink better
- ๐ Add Channels: Implement adapters for new chat channels
- โ Core Engine: Stable
- โ Desktop Client: Stable (Tauri 2.0)
- โ CI/CD Pipeline: Fully automated
- โ 15+ Chat Channels: Production ready
- ๐ง Rich Media: Expanding channel support
- ๐ง Mobile Clients: Planned
Built with amazing open-source technologies:
- Rust - Systems programming language
- Tauri - Desktop application framework
- Vue 3 - Progressive JavaScript framework
- Tokio - Async runtime for Rust
This project is licensed under the MIT License - see the LICENSE file for details.
โญ Star us on GitHub if AgentLink helps your workflow!