How to Integrate Claude 2 into Slack: Setup Guide

Want Claude 2's document analysis and AI capabilities inside Slack? You can build a custom integration using the Claude API and Slack's bot framework. This isn't an official pre-built app. It's a custom integration you build yourself. Here's how to set it up. ## Why This Matters Your team lives in Slack. Context lives in Slack. Documents get shared in Slack channels. **Integrating Claude 2 into Slack means your team can analyze documents, get answers, and automate workflows without switching tools.** No more downloading PDFs, opening claude.ai in a browser, copy-pasting results back. For operations teams, this reduces friction in document review, Q&A, and collaboration workflows. ## What You'll Build This guide walks through building a Slack bot that: - Responds to mentions with Claude 2 answers - Analyzes uploaded files (PDFs, text, CSV) - Maintains conversation context per channel - Handles document summarization - Works with slash commands ## Prerequisites You'll need: - Claude API access (sign up at console.anthropic.com) - Slack workspace admin access - Basic Python knowledge - A server to host the bot (or use a service like Railway or Heroku) ## Step 1: Set Up Claude API Access First, get your Claude API credentials. **Get API Key**: 1. Go to console.anthropic.com 2. Create an account or sign in 3. Navigate to API Keys 4. Generate a new key 5. Save it securely (you'll need it later) **Test API Access**: ```python import anthropic client = anthropic.Anthropic( api_key="your-api-key-here" ) message = client.messages.create( model="claude-2", max_tokens=1024, messages=[ {"role": "user", "content": "Hello, Claude!"} ] ) print(message.content) ``` If this works, you're ready to proceed. ## Step 2: Create Slack App **Create New App**: 1. Go to api.slack.com/apps 2. Click "Create New App" 3. Choose "From scratch" 4. Name it "Claude Assistant" 5. Select your workspace **Configure Bot Permissions**: Under "OAuth & Permissions", add these scopes: - `app_mentions:read` (see @mentions) - `chat:write` (send messages) - `files:read` (access uploaded files) - `channels:history` (read channel messages) - `groups:history` (read private channels) **Enable Events**: Under "Event Subscriptions": 1. Turn on "Enable Events" 2. Add bot user events: - `app_mention` (when someone @mentions the bot) - `file_shared` (when files are uploaded) 3. Save (you'll need a request URL, set that up in Step 3) **Install App**: 1. Go to "Install App" 2. Click "Install to Workspace" 3. Authorize the app 4. Save the "Bot User OAuth Token" ## Step 3: Build the Integration Here's a working Python implementation using Flask: ```python import os import anthropic from flask import Flask, request, jsonify from slack_sdk import WebClient from slack_sdk.errors import SlackApiError app = Flask(__name__) # Initialize clients slack_client = WebClient(token=os.environ["SLACK_BOT_TOKEN"]) claude_client = anthropic.Anthropic( api_key=os.environ["ANTHROPIC_API_KEY"] ) # Store conversation history per channel conversations = {} @app.route("/slack/events", methods=["POST"]) def slack_events(): data = request.json # Verify Slack challenge if "challenge" in data: return jsonify({"challenge": data["challenge"]}) # Handle app mention if data["event"]["type"] == "app_mention": handle_mention(data["event"]) return jsonify({"status": "ok"}) def handle_mention(event): channel = event["channel"] user = event["user"] text = event["text"] # Remove bot mention from text query = text.split(">", 1)[1].strip() if ">" in text else text # Get response from Claude try: response = claude_client.messages.create( model="claude-2", max_tokens=2000, messages=[{"role": "user", "content": query}] ) answer = response.content[0].text # Send response to Slack slack_client.chat_postMessage( channel=channel, text=answer, thread_ts=event.get("thread_ts", event["ts"]) ) except Exception as e: slack_client.chat_postMessage( channel=channel, text=f"Sorry, I encountered an error: {str(e)}", thread_ts=event.get("thread_ts", event["ts"]) ) if __name__ == "__main__": app.run(port=3000) ``` **Set Environment Variables**: ```bash export SLACK_BOT_TOKEN="xoxb-your-token" export ANTHROPIC_API_KEY="your-claude-key" ``` ## Step 4: Add File Analysis Extend the bot to analyze uploaded files: ```python @app.route("/slack/events", methods=["POST"]) def slack_events(): data = request.json if "challenge" in data: return jsonify({"challenge": data["challenge"]}) if data["event"]["type"] == "app_mention": handle_mention(data["event"]) if data["event"]["type"] == "file_shared": handle_file(data["event"]) return jsonify({"status": "ok"}) def handle_file(event): file_id = event["file_id"] channel = event["channel_id"] try: # Download file from Slack file_info = slack_client.files_info(file=file_id) file_url = file_info["file"]["url_private"] # Download file content headers = {"Authorization": f"Bearer {os.environ['SLACK_BOT_TOKEN']}"} response = requests.get(file_url, headers=headers) file_content = response.text # Analyze with Claude analysis = claude_client.messages.create( model="claude-2", max_tokens=3000, messages=[{ "role": "user", "content": f"Please analyze this document and provide a summary:\n\n{file_content}" }] ) # Post results slack_client.chat_postMessage( channel=channel, text=f"Document Analysis:\n\n{analysis.content[0].text}" ) except Exception as e: print(f"Error analyzing file: {e}") ``` ## Step 5: Deploy You need to host this bot somewhere accessible to Slack. **Option 1: Railway (Recommended)** 1. Push code to GitHub 2. Connect Railway to your repo 3. Add environment variables 4. Deploy 5. Use Railway URL in Slack event subscriptions **Option 2: Heroku** 1. Create Heroku app 2. Configure environment variables 3. Deploy with Git 4. Use Heroku URL in Slack **Option 3: Your Server** 1. Set up a server with Python 2. Use nginx or similar for HTTPS 3. Run the Flask app 4. Configure firewall for Slack IPs ## Real-World Use Cases At The Operations Guide, we use this integration for: **Contract Review in #legal** - Team member uploads vendor contract - @mentions Claude with "summarize this and flag any unusual terms" - Claude analyzes and responds in thread - Legal team reviews before formal analysis **Code Review in #engineering** - Developer shares code snippet - @mentions Claude with "review this function for bugs" - Claude provides feedback - Developer iterates in the thread **Document Q&A in #operations** - Someone uploads quarterly report - Multiple team members ask questions in thread - Claude maintains context across questions - Faster than reading full report individually **Meeting Prep in #leadership** - Upload board deck or briefing doc - Ask Claude for key takeaways - Get summary before meeting starts - Everyone arrives better prepared ## Cost Considerations Every Claude query costs API tokens. Estimate your costs: **Typical Slack usage**: - 50 questions per day: ~$2-3/month - 20 document analyses per day: ~$4-6/month - Total: ~$6-9/month Compare to productivity gains from faster document review and Q&A. ## Limitations and Gotchas **File Format Support** Claude works best with plain text. Complex PDFs may need preprocessing. **Response Time** Claude takes 10-30 seconds. Set expectations with your team. **Context Length** Slack messages are short, but document uploads can be large. Watch your token usage. **Rate Limits** Claude API has rate limits. Build queuing if you expect high volume. ## Quick Takeaway Integrating Claude 2 into Slack takes about 2 hours of development work. For teams that share and discuss documents in Slack, it eliminates workflow friction and speeds up document analysis by 3-5x.