Startvest-LLC/idealift-mcp-server

Decision intelligence for AI assistants via Model Context Protocol.
Connect Claude, ChatGPT, and other AI assistants to your product backlog. Capture feedback, track decisions, and manage ideas without leaving your AI workflow.
What it does
- **Normalize ideas** — Transform raw text into structured feature requests, bug reports, or tasks
- **Signal aggregation** — Capture product feedback from Slack, Teams, Discord, and GitHub
- **Decision tracking** — Record who decided what, when, and why with full audit trails
- **RICE scoring** — Prioritize ideas with Reach, Impact, Confidence, and Effort scores
- **Ticket creation** — Push structured ideas to GitHub Issues, Jira, or Linear
MCP Tools
| Tool | Auth Required | Description |
|------|:---:|-------------|
| normalize_idea | No | Transform raw text into a structured idea |
| check_auth | No | Check if user is linked to IdeaLift |
| list_destinations | Yes | Get available GitHub/Jira/Linear projects |
| create_ticket | Yes | Create a ticket in a destination system |
| connect_destination | Yes | Connect a new destination |
Plus proxy tools for full IdeaLift API access (ideas, signals, decisions, roadmap).
Setup
1. Install dependencies
2. Configure environment
Required environment variables:
| Variable | Description |
|----------|-------------|
| IDEALIFT_APP_URL | IdeaLift API base URL |
| INTERNAL_API_KEY | Service-to-service auth key |
| DATABASE_HOST | SQL Server host |
| DATABASE_NAME | Database name |
| DATABASE_USERNAME | Database user |
| DATABASE_PASSWORD | Database password |
| OPENAI_API_KEY | For idea normalization |
3. Build and run
Or for development:
The MCP server starts on port 3001 (configurable via PORT).
4. Connect to your AI assistant
**SSE endpoint:** http://localhost:3001/mcp
Add this URL as an MCP server in Claude Desktop, ChatGPT, or any MCP-compatible client.
Transport
Uses Server-Sent Events (SSE) transport with automatic keep-alive pings every 15 seconds.
OAuth