Inspect this project's Convex deployment
Using the Convex MCP, show my dev deployment status, list tables in the schema, and suggest indexes for the mcps table.
Query and manage your backend from inside your AI agent
The Convex MCP server ships with the Convex CLI and gives Cursor, Claude Code, and Codex tools to inspect deployments, run queries, and optimize your real-time backend without leaving the editor.
AI agents can write code. But they don't always know what works. Convex MCP fixes that.
Using the Convex MCP, show my dev deployment status, list tables in the schema, and suggest indexes for the mcps table.
Cursor
Inspect this project's Convex deployment
Used Convex MCP integrationResults ready. Pulling the strongest patterns for synthesis.
Convex connects your AI agents to real-world references.
Schema inspection
List tables, fields, and indexes in your Convex deployment without opening the dashboard. Your agent reads the live schema and explains relationships between documents. Catch missing indexes or optional fields before they cause production queries to slow down. Understand backend shape while you are still writing the UI that depends on it.
Your AI agents can search, reference, and reason about all of it.
Ask anything. Get real results back.
Works with
and more.
Quick start
Using Convex MCP, walk me through a first query I can run in Cursor to see what this connector can do.
Design research
Use Convex to research design patterns relevant to my product and summarize the top findings.
Competitive scan
With Convex MCP, compare how 3 leading products handle design and list actionable takeaways.
Feature inspiration
Find inspiration for my next design feature using Convex and suggest 3 directions with examples.
Workflow audit
Audit my current design workflow with Convex MCP and recommend concrete improvements.
Research brief
Build a one-page research brief using Convex for an upcoming launch in the design space.
Summarize findings
Pull data from Convex, summarize the strongest patterns, and propose next steps for my team.
Best practices
What are the best practices teams follow when using Convex MCP inside Cursor or Claude Code?
Micro-decision help
I'm stuck on a design micro-decision — use Convex to surface examples and recommend an approach.
When to reach for Convex MCP
Building from scratch
Your agent analyzes successful patterns and applies them using Convex.
Building from scratch workflow preview
Stuck on a micro-decision
Let agents explore best practices from products that already figured it out.
Stuck on a micro-decision workflow preview
Workflow feels outdated
Pull what top products are doing right now so you're building from fresh patterns.
Workflow feels outdated workflow preview
Set up in under a minute.
- 1
Set up in Cursor
Open Cursor Settings → Tools & MCP → New MCP Server Add: "command": "npx", "args": ["-y", "convex@latest", "mcp", "start"] Enable the server and wait for the green status indicator
- 2
Set up in Claude Code
Run: claude mcp add convex -- npx -y convex@latest mcp start Run /mcp to confirm Convex is connected
- 3
You're set
Convex MCP is connected. Start prompting with real references.
Frequently asked questions
Does Convex MCP access production by default?+
No. Production access is disabled by default for security. Add flags to your MCP config only if you explicitly need prod access.
What can I do with Convex MCP?+
Convex MCP lets your agent query query and manage your backend from inside your ai agent without switching tools or copy-pasting data.
What design data can my agent access?+
Your agent can pull design insights from Convex — search, compare, and summarize results in the context of your project.
Can my agent compare results across sources?+
Yes. Ask your agent to compare patterns, metrics, or examples from Convex side by side and summarize actionable takeaways.
What workflows does this support?+
Research, audits, competitive scans, and design planning — your agent uses Convex as live context while you build.
Your AI agents are guessing. They don't have to.
Query and manage your backend from inside your AI agent