What is Agent Slack

Agent-Slack is an AI-powered agent designed for Slack, utilizing the first-party Slack MCP Server generated by OpenAPI code generation. It integrates LangGraph and LangChain MCP Adapters and is accessible through various agent transport protocols such as AGNTCY Slim, Google A2A, and MCP Server.

Use cases

Use cases for Agent-Slack include automating responses to frequently asked questions, providing real-time data insights, facilitating team discussions, and enhancing user engagement through interactive AI-driven conversations.

How to use

To use Agent-Slack, you need to integrate it with your Slack workspace. This involves setting up the necessary configurations and utilizing the provided APIs to interact with the agent. Detailed instructions can typically be found in the project’s README or documentation.

Key features

Key features of Agent-Slack include seamless integration with Slack, support for multiple transport protocols, and the ability to leverage advanced AI capabilities through LangGraph and LangChain. It also includes evaluation badges for various AI models, ensuring performance and reliability.

Where to use

Agent-Slack can be used in various fields such as customer support, team collaboration, and project management, where AI-driven assistance can enhance productivity and communication within Slack.

🚀 Slack AI Agent

🧪 Evaluation Badges

Claude	Gemini	OpenAI	Llama

• 🤖 Slack Agent is an LLM-powered agent built using the LangGraph ReAct Agent workflow and MCP tools.
• 🌐 Protocol Support: Compatible with A2A protocol for integration with external user clients.
• 🛡️ Secure by Design: Enforces Slack API token-based authentication and supports external authentication for strong access control.
• 🔌 Integrated Communication: Uses langchain-mcp-adapters to connect with the Slack MCP server within the LangGraph ReAct Agent workflow.
• 🏭 First-Party MCP Server: The MCP server is generated by our first-party openapi-mcp-codegen utility, ensuring version/API compatibility and software supply chain integrity.

🏗️ Architecture

flowchart TD
  subgraph Client Layer
    A[User Client A2A]
  end

  subgraph Agent Transport Layer
    B[Google A2A]
  end

  subgraph Agent Graph Layer
    C[LangGraph ReAct Agent]
  end

  subgraph Tools/MCP Layer
    D[LangGraph MCP Adapter]
    E[Slack MCP Server]
    F[Slack API Server]
  end

  A --> B --> C
  C --> D
  D -.-> C
  D --> E --> F --> E

✨ Features

• 🤖 LangGraph + LangChain MCP Adapter for agent orchestration
• 🧠 Azure OpenAI GPT-4 as the LLM backend
• 🔗 Connects to Slack via a dedicated Slack MCP agent
• 🔄 Multi-protocol support: Compatible with A2A protocol for flexible integration and multi-agent orchestration
• 📊 Comprehensive Slack API Support:
  • Channel Management (list, join, leave)
  • Message Management (send, update, delete)
  • User Management (list, info, status)
  • File Management (upload, list, delete)
  • Reaction Management (add, remove)
  • Thread Management (reply, follow)

🚀 Getting Started

Running it via Docker:

1️⃣ Configure Environment

Ensure your .env file is set up as described in the cnoe-agent-utils usage guide based on your LLM Provider.

Example .env configuration:

############################
# Agent Configuration
############################
LLM_PROVIDER=azure-openai
AGENT_NAME=Slack

## A2A Agent Configuration
A2A_AGENT_HOST=localhost
A2A_AGENT_PORT=8000

## MCP Server Configuration
MCP_HOST=localhost
MCP_PORT=9000

############################
# Azure OpenAI Configuration
############################
AZURE_OPENAI_API_KEY=<your-azure-key>
AZURE_OPENAI_API_VERSION=2025-04-01-preview
AZURE_OPENAI_DEPLOYMENT=gpt-4.1
AZURE_OPENAI_ENDPOINT=<your-azure-endpoint>

############################
# Slack Configuration
############################
SLACK_BOT_TOKEN=<your-bot-token>
SLACK_APP_TOKEN=<your-app-token>
SLACK_SIGNING_SECRET=<your-signing-secret>
SLACK_CLIENT_SECRET=<your-client-secret>
SLACK_TEAM_ID=<your-team-id>

2️⃣ Start the Agent (A2A Mode)

1. Pull the A2A image:

docker pull ghcr.io/cnoe-io/agent-slack:a2a-stable

2. Run the agent in a Docker container using your .env file:

docker run -p 0.0.0.0:8000:8000 -it \
  -v $(pwd)/.env:/app/.env \
  ghcr.io/cnoe-io/agent-slack:a2a-stable

3️⃣ Run the Client

Use the agent-chat-cli to interact with the agent:

uvx https://github.com/cnoe-io/agent-chat-cli.git a2a

Alternative: Running Locally

You can also run the agent locally without Docker:

First clone the repo

git clone https://github.com/cnoe-io/agent-slack.git
cd agent-slack

make run-a2a

In a new terminal, start the A2A client:

make run-a2a-client

Quick Demos

▶️ Test with Slack API

🏃 Quick Start: Set Up Slack Access

1. Create a Slack App:
   If you don’t have one, create a new app at Slack API.

2. Configure Bot Token Scopes:

   • Go to OAuth & Permissions
   • Add the following scopes:
     • app_mentions:read
     • channels:history
     • chat:write
     • users:read
     • groups:history
     • im:history
     • mpim:history

3. Enable Socket Mode:

   • Go to Socket Mode in your app settings
   • Enable Socket Mode and generate an app-level token
   • Save the token as SLACK_APP_TOKEN

4. Install the App:

   • Install the app to your workspace
   • Save the Bot User OAuth Token as SLACK_BOT_TOKEN

Example Interactions

1. Channel Management

User: list all channels
Agent: Here are all the channels in the workspace:

1. #general - Company-wide announcements
   - Members: 150
   - Purpose: General discussion and updates

2. #dev-team - Development team discussions
   - Members: 45
   - Purpose: Technical discussions and updates

3. #random - Random conversations
   - Members: 120
   - Purpose: Non-work related chats

Would you like to join any of these channels or get more details?

2. Message Management

User: send a message to #general
Agent: I'll help you send a message. What would you like the message to say?

User: Hello team! Just testing our new Slack bot.
Agent: I'll send that message to #general. Here's what I'm going to send:

"Hello team! Just testing our new Slack bot."

Would you like me to:
1. Send it as is
2. Add formatting or emojis
3. Make any changes first?

User: Send it as is
Agent: Message sent successfully to #general!
Message ID: 1234567890.123456
Timestamp: 2024-04-10 14:30:00 UTC

Let me know if you'd like to:
- Add reactions to the message
- Start a thread
- Edit or delete the message

🔍 Troubleshooting

Common Issues

1. Slack API Issues

   • Verify bot token scopes are correct
   • Check if app is properly installed to workspace
   • Ensure Socket Mode is enabled if required

2. Port Conflicts

   • If port 8000 is in use, modify port in .env
   • Check for other services using required ports

3. Environment Variables

   • Verify all required variables in .env
   • Check API keys and tokens are valid
   • No trailing spaces in values

4. Client Connection Issues

   • Server must be running before client
   • Port numbers should match
   • API keys must match between server and client

📚 Documentation

For more detailed information about the project, please refer to:

• API Documentation - Detailed API reference
• Architecture Overview - System design and components
• Development Guide - Setup and development workflow
• Deployment Guide - Production deployment instructions
• Contributing Guide - How to contribute to the project
• Security Policy - Security practices and vulnerability reporting
• License - Project license details
• Changelog - Version history and changes
• Code of Conduct - Community guidelines

🔐 Security Notes

• Never commit your .env file to version control
• Keep your API keys and tokens secure
• Use environment variables or secret managers in production
• Regularly rotate your API keys and tokens

👥 Maintainers

See MAINTAINERS.md for the list of maintainers.

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details on how to get started.

📄 License

Apache 2.0