AI Mission Control for Engineering Teams
MCP is an intelligent coordination system that maintains context of everything happening in your team, routes information intelligently, and provides managers with real-time visibility.
- Information overload: Slack has 100+ channels, critical updates get lost
- Scattered context: No single source of truth for team activities
- Blind managers: Hard to know what's blocking the team
- Inefficient help: People @ mention the wrong person or nobody responds
One AI entity that:
- 🧠 Knows everything - Maintains context of all team activities
- 🎯 Routes intelligently - Connects the right people based on expertise
- 📊 Provides visibility - Managers get real-time insights
- 🤖 Learns continuously - Gets smarter from every interaction
┌─────────────────────────────────────────┐
│ MCP Core (The Brain) │
│ ┌───────────────────────────────────┐ │
│ │ Vector DB (Qdrant) │ │
│ │ - All context as embeddings │ │
│ │ - Semantic search capability │ │
│ └───────────────────────────────────┘ │
│ ┌───────────────────────────────────┐ │
│ │ Intelligence (Gemini) │ │
│ │ - Parse standups │ │
│ │ - Route help requests │ │
│ │ - Generate insights │ │
│ └───────────────────────────────────┘ │
│ ┌───────────────────────────────────┐ │
│ │ Database (PostgreSQL) │ │
│ │ - Users, tasks, hierarchy │ │
│ │ - Structured data │ │
│ └───────────────────────────────────┘ │
└─────────────────────────────────────────┘
↕️ REST API
┌──────────┬───────────┬──────────┐
│ │ │ │
┌───▼────┐ ┌──▼─────┐ ┌───▼────┐ ┌──▼─────┐
│ Slack │ │ Web │ │ Future │ │ Jira │
│ Bot │ │Dashboard│ │Clients │ │Webhook │
└────────┘ └────────┘ └────────┘ └────────┘
- Docker and Docker Compose
- Google Gemini API key (Get it here)
- Slack workspace (for bot integration)
git clone <repository-url>
cd Dandy-Hack-25
# Copy environment template
cp .env.example .env
# Edit .env with your credentials
nano .env- Go to api.slack.com/apps
- Click "Create New App" → "From scratch"
- Name it "MCP" and select your workspace
Required Bot Token Scopes:
chat:writeim:writeim:historyusers:readchannels:readgroups:writempim:write
Event Subscriptions:
- Enable Socket Mode
- Subscribe to:
message.im,app_mention
Slash Commands:
/standup- Trigger daily standup/mcp-summary- Get team summary (managers)/mcp-assign- Assign tasks (managers)
- Copy Bot Token (
xoxb-...) to.envasSLACK_BOT_TOKEN - Enable Socket Mode and copy App Token (
xapp-...) to.envasSLACK_APP_TOKEN
# Start all services
docker-compose up -d
# Check logs
docker-compose logs -f
# Verify services are running
docker-compose psServices:
- Backend API: http://localhost:8000
- Swagger Docs: http://localhost:8000/docs
- Dashboard: http://localhost:3000
- Qdrant: http://localhost:6333
- PostgreSQL: localhost:5432
In Slack:
- DM the MCP bot
- Type
/standup - Answer the questions
- Watch MCP process your update!
Trigger via command:
/standup
Or just DM MCP:
Yesterday: Completed user auth API
Today: Working on Redis caching, need help with session management
Blockers: Waiting on design mockups
MCP will:
- ✅ Parse and extract structured data
- ✅ Route help requests to experts
- ✅ Alert manager about blockers
- ✅ Update task statuses
In your standup, mention help needed:
Working on OAuth integration, need help from someone with OAuth experience
MCP will:
- 🔍 Search for team members with OAuth expertise
- ✅ Find available expert (not overloaded)
- 💬 Create 3-person group chat (you + expert + MCP)
- 📝 Learn from the conversation
Get team summary:
/mcp-summary 7
Assign a task:
/mcp-assign @john Implement Redis caching high
Ask questions:
@MCP what's blocking Sarah?
@MCP who knows about Docker?
@MCP summarize this week
Just ask MCP anything:
- "Who worked on the authentication system?"
- "Show me tasks blocked more than 2 days"
- "What are the top blockers this week?"
mcp-product/
├── backend/ # FastAPI backend (MCP Core)
│ ├── main.py # FastAPI app
│ ├── config.py # Configuration
│ ├── services/ # Core services
│ │ ├── mcp_core.py # MCP intelligence
│ │ ├── vector_db.py # Qdrant interface
│ │ └── database.py # PostgreSQL models
│ └── routers/ # API endpoints
├── slack-bot/ # Slack integration
│ └── app.py # Bolt app
├── dashboard/ # Web dashboard
│ └── index.html # Minimal UI
└── docker-compose.yml # Infrastructure
Add a new context source:
# Any service can add context to MCP
await mcp_core.add_context({
"type": "github_commit",
"user_id": "john",
"text": "Fixed OAuth bug in user service",
"metadata": {"repo": "backend", "commit": "abc123"}
})Query context semantically:
# Semantic search across all context
results = await mcp_core.query(
query="Who fixed OAuth bugs recently?",
user_id="manager_id"
)Health Check:
curl http://localhost:8000/healthSubmit Standup:
curl -X POST http://localhost:8000/api/standups/submit \
-H "Content-Type: application/json" \
-d '{"user_id": "john", "message": "Working on feature X"}'Query MCP:
curl -X POST "http://localhost:8000/api/mcp/query?query=who%20knows%20React&user_id=john"- Core MCP backend
- Vector DB for context storage
- Gemini integration
- Slack bot (standups, help routing)
- Minimal dashboard
- Advanced help routing (workload balancing)
- Proactive blocker detection
- Recurring issue tracking
- Velocity analytics
- Jira two-way sync
- GitHub webhook integration
- Calendar integration
- Linear, Asana support
- Multi-team support
- Advanced analytics dashboard
- Dependency graph visualization
- Predictive insights
This is a YC application project. Contributions welcome!
- Fork the repo
- Create feature branch
- Make changes
- Submit pull request
MIT License
- Issues: GitHub Issues
- Slack: Join our community
- Email: support@mcp.dev
Built with ❤️ for engineering teams who want to move faster.
MCP - One AI that knows everything, so you don't have to.