Demand Discovery AI — Claude Code MCP server: what it does
Demand Discovery AI — a remote MCP server for idea validation: it reads behavioral signals, gives a 0–100 Demand Score and a Build, Pivot or Kill verdict.
no license 635 installs updated 2026-05-02
claude mcp add --transport http demand-discovery-ai https://demand-discovery-ai.run.toolsWhat Demand Discovery AI does
The endpoint is https://demand-discovery-ai.run.tools, over HTTP. Ask it about validating an idea, finding prospects or reading market signals and it answers from its own knowledge base. Seven tools come with it. In Cursor and GitHub Copilot you wire it up through a JSON config instead of a CLI command.
get_validation_framework returns three stages: research the market, get a Demand Discovery Report with the score and verdict, then run 90 days of outreach. get_data_source_categories sorts the evidence into four buckets: what people search for, what they say about the pain, what companies adopt and pay for, and where money and hires move. Scores rest on how people behave, not on text a model made up. That is the claim the server makes for itself.
With start_demand_report you get a deep link to the free market research report, form fields already carrying the startup name, target market, one-line description and email. You confirm and submit it yourself. Paid tiers start at $49 for Starter and go to packs of 5 or 25 ideas, where a pivoted or archived idea frees a slot for a new one.
What it helps with
- Test whether an idea deserves code
- Find people worth interviewing before launch
- See where the demand evidence comes from
- Compare validation methods without the sales pitch
- Start a market research report from the chat
Example prompts for the agent
- Is a room-booking tool for small cafés worth building? Run it through the Demand Discovery framework.
- Show me the three validation stages and how the Demand Score is assembled.
- Start the free market research report for my idea.
How to install Demand Discovery AI
Platform CLI: remote server
claude mcp add --transport http demand-discovery-ai https://demand-discovery-ai.run.toolsPlatform CLI: remote server
codex mcp add demand-discovery-ai --url https://demand-discovery-ai.run.toolsConfig file
{
"mcpServers": {
"demand-discovery-ai": {
"url": "https://demand-discovery-ai.run.tools"
}
}
}Config file
{
"servers": {
"demand-discovery-ai": {
"type": "http",
"url": "https://demand-discovery-ai.run.tools"
}
}
}Platform CLI: remote server
gemini mcp add -t http demand-discovery-ai https://demand-discovery-ai.run.toolsStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Demand Discovery AI
How do I install Demand Discovery AI in Claude Code?
Run claude mcp add --transport http demand-discovery-ai https://demand-discovery-ai.run.tools. The transport is http and no API key is involved.
What does Demand Discovery AI do?
It answers questions about validating a startup idea and scores demand from behavioral signals, returning a number from 0 to 100 plus a Build, Pivot or Kill verdict. The start_demand_report tool launches a report without leaving the chat.
How is it different from a general AI assistant?
The compare_validation_approaches tool lays out where generic assistants, trend aggregators and passive scoring tools stop, and what this one does instead. Answers lean on observed behavior rather than a model's opinion.