The DeepVista MCP server lets any compatible AI client read, search, create, edit, and delete your context cards — without leaving your coding environment or chat interface. All actions are automatically scoped to your currently active DeepVista project. No extra configuration needed.Documentation Index
Fetch the complete documentation index at: https://docs.deepvista.ai/llms.txt
Use this file to discover all available pages before exploring further.
MCP (Model Context Protocol) is an open standard for connecting AI clients to external data sources. Read the spec →
Endpoint
The MCP server uses Streamable HTTP transport with OAuth 2.1 authentication.Setup
- Cursor
- Claude Desktop
- Claude Code
- VS Code
- Windsurf
- Other clients
Add the following to your Cursor MCP config file at
~/.cursor/mcp.json. Create the file if it doesn’t exist.~/.cursor/mcp.json
Sign in to DeepVista
A browser window will open — complete the OAuth flow to authorize the connection.
Verify it’s working: In Cursor’s AI chat, type
List my DeepVista context cards. You should see a summary of cards from your active project.What you can do
Once connected, talk to your AI naturally. Here are some example prompts to get started.| Action | Example prompt |
|---|---|
| Find | Find my context cards about machine learning pipelines |
| Search | Which cards mention series A fundraising? |
| Create | Create a new context card for Acme Corp — Q3 partnership notes |
| Update | Update my card on Sarah Chen with: joined the advisory board in May |
| List | Show me all context cards I've updated this week |
| Delete | Delete the context card titled Old vendor contract |
The AI will always confirm before deleting a card. To switch projects, change your active project in DeepVista — the MCP connection reflects it immediately.
Using an API key instead of OAuth
You can authenticate with a DeepVista API key instead of going through the interactive OAuth flow. This is useful for scripted workflows, read-only access, or integrating with an existing OAuth application. Generate a key from Settings → Security & Access, then configure your client to send:Troubleshooting
I'm getting an authentication error
I'm getting an authentication error
Open deepvista.ai in your browser and confirm you’re signed in to the correct account. Then return to your AI client and try connecting again.If the issue persists, clear the stored auth cache and retry:
My client can't find or reach the server
My client can't find or reach the server
Double-check that the endpoint URL in your config is exactly
https://mcp.deepvista.app/mcp — no trailing slash.If you’re using the mcp-remote bridge, confirm Node.js 18+ is installed:The AI connects but can't see any of my cards
The AI connects but can't see any of my cards
The MCP server returns cards from your currently active project. Open DeepVista and confirm the correct project is selected in the project switcher. An empty list is expected if the active project has no cards yet.
I'm on Windows using WSL and seeing errors
I'm on Windows using WSL and seeing errors
Use this WSL-compatible configuration instead:
I see an internal server error when connecting
I see an internal server error when connecting
Clear saved auth info and try again:You may also need to update to a newer version of Node.js if your current version is below 18.
CLI Quickstart
Capture notes and run workflows from your terminal using the DeepVista CLI.
Get Early Access
Sign up to start connecting your AI tools to DeepVista.
.png?fit=max&auto=format&n=9107aCYFqrRP7oCT&q=85&s=a6dd625dc2a8af905b9f9dac63e02797)
.png?fit=max&auto=format&n=9107aCYFqrRP7oCT&q=85&s=ee063c9f4442fd8912d78b3151b264b7)