Coding agents are good at tickets. Give one a repo and a task and it will produce a pull request. Ask it whether a market is worth entering and it will run a web search, read a few blog posts, and hand you adjectives instead of numbers.
I run 01, a studio that ships a new product every day or two. The slow part is rarely the code. It is the work around it: sizing the market, checking who ranks for the keywords, finding out if the domain is free, standing up the same SaaS scaffold again, and then doing the launch chores. We built tooling for our own use, and StartupKit is that tooling, exposed so an agent can call it directly.
StartupKit is an MCP server for startups. There is also a CLI that hits the same API.
What it does
The docs describe four product areas on one account and one API key.
Stack. One command scaffolds a production-ready SaaS on Cloudflare: Next.js 16, better-auth on D1, Stripe billing, email, analytics and SEO, with agent skills included. Stack needs no account.
npx startupkit init "My SaaS"
Research. Market research, competitor SEO, keyword intelligence, trend analysis, domain intelligence, and app-store data. The point is structured output the agent can reason over. The research tools (research, keywords, trends, seo, serp, domains, apps) return a summary table plus a JSON tail, backed by DataForSEO, Google Trends, Appfigures, and registrar APIs. Research tools draw down credits, and market research costs more than the rest because it aggregates several sources.
Launch. A launch checklist you can work through with the agent. Directory and press submissions are listed as coming soon, so I am not counting them as shipped.
Ads. Generating ads for your product. This is also marked coming soon.
In total the MCP server exposes 15 tools. It works with Claude Code, Cursor, OpenCode, Hermes, Codex, and any other MCP client.
Connect it
Get a free API key at startupkit.com, then run:
npx startupkit login
Add this to your agent's MCP config:
{
"mcpServers": {
"startupkit": {
"url": "https://mcp.startupkit.com/mcp",
"headers": {
"Authorization": "Bearer sk_..."
}
}
}
}
The endpoint is https://mcp.startupkit.com/mcp. Replace sk_... with your key. The exact config file and key name differ by client, so check yours; the block above is the format from the StartupKit docs.
Once connected, you ask in plain language: is there demand for this, who ranks for it, is the .com available. The agent calls the tools and the structured results land in its context. You review the output instead of copying numbers out of four different dashboards.
If you prefer a terminal, the CLI covers the same ground:
npx startupkit keywords "ai tools"
npx startupkit domains "ai tools"
npx startupkit seo example.com
Pricing
Free gives you one project, 10 research credits per month, and every product area. The Stack, the MCP server, and the CLI are free.
Pro is $39/month: unlimited projects, 1,000 research credits per month, priority support, and directory and press submissions once they ship. Credit packs can be added on Pro if you run through your monthly credits.
We charge for the research data and launch work, which only matter once you are building something, and give away the scaffold. Boilerplates commonly charge $200 or more for a starter kit, and I would rather the starter be free.
Limits worth knowing
- 10 research credits a month is enough to try the tools, not to do sustained research. That is what Pro is for.
- Directory and press submissions and Ads are not live yet.
- The output is data, not a verdict. An agent can tell you a keyword's volume and who ranks. Whether to build the thing is still your call.
Where to find it
- Site and docs: https://startupkit.com
- MCP endpoint: https://mcp.startupkit.com/mcp
- CLI:
npx startupkit - X: @01stdio
If something in the docs is wrong or a tool does not behave the way it says, tell me on X.
Ian Hunter, CEO, 01
Top comments (0)