vaaya / get started
A “connector” just links your AI app to Vaaya — no code, no API key. Two things make the rest obvious, then pick your app above.
Claude Desktop, Claude on the web, ChatGPT, Cursor, and VS Code. You add Vaaya from a settings screen inside the app — no typing of commands.
Claude Code, Codex, Gemini CLI. You paste one line into their text window and press Enter. Here’s how to open that window.
Only needed for the terminal tools. It’s a built-in app on every computer — here’s how to find it and paste into it.
Press ⌘ Cmd + Space, type Terminal, press Enter. A window opens — click in it, paste with ⌘ Cmd + V, press Enter to run.
Press the ⊞ Windows key, type cmd, press Enter. In the window, paste with Ctrl + V (or right-click), press Enter to run.
You do not need “npx”, “mcp-remote”, or any helper. Vaaya is hosted, so every app connects straight to https://vaaya.ai/mcp.
Add Vaaya from a settings screen — paste the address, sign in, done. No terminal.
Vaaya, paste the address below into the URL box, and click Add.Paste this as the URL
https://vaaya.ai/mcpVaaya, paste the address below, and click Add.Paste this as the URL
https://vaaya.ai/mcpmcp.json file.{
"mcpServers": {
"vaaya": { "url": "https://vaaya.ai/mcp" }
}
}
Vaaya.mcp.json instead.)Paste this as the URL
https://vaaya.ai/mcp{
"servers": {
"vaaya": { "type": "http", "url": "https://vaaya.ai/mcp" }
}
}
Vaaya, paste the address below, set authentication to OAuth, and click Create. Sign in with Google when the browser opens.Paste this as the MCP server URL
https://vaaya.ai/mcpChatGPT is cautious with paid tools — it may describe a plan instead of running it. Just tell it: “call the use tool now to execute” and it will run.
You already have these tools — just paste one line to add Vaaya.
claude. (New to it? See How to open the typing window above.)claude), type /mcp, choose vaaya, and pick Authenticate — your browser opens, sign in with Google and approve.Add Vaaya (same on Mac and Windows)
claude mcp add --transport http --scope user vaaya https://vaaya.ai/mcpgemini.gemini) and type /mcp to finish signing in to Vaaya in your browser.Add Vaaya
gemini mcp add --transport http vaaya https://vaaya.ai/mcp.codex/config.toml in your home folder. Mac: ~/.codex/config.toml · Windows: C:\Users\<you>\.codex\config.toml.codex. The first time it uses Vaaya it opens your browser to sign in with Google.experimental_use_rmcp_client = true [mcp_servers.vaaya] url = "https://vaaya.ai/mcp"
Newer Codex versions can add it from the terminal — run codex mcp --help to check.
One quick test confirms Vaaya is connected before you rely on it.
Send this message to your AI:
Use the Vaaya consult tool and tell me what you can do.You’re good to go when it actually runs the tool and replies with what Vaaya can reach (images, web, voice, compute…) — no “not connected” or sign-in error. If it only talks about Vaaya, add: “call the tool now.” (This uses the free consult tool — nothing is charged.)
In the terminal, type /mcp and look for Vaaya listed as connected:
vaaya ✔ connected · 21 tools
If it shows connected with tools, you’re set. Not there? Re-run the add command above, then /mcp again.
The four things people ask most.
Remove the connector and add it again. A plain “Reconnect” reuses old, limited permissions — a fresh add fixes it.
Copy https://vaaya.ai/mcp exactly — it’s the only address you ever paste into any app.
No. Vaaya is hosted, so every app connects straight to the address. You never install a bridge.
You start with $100 in free credit, and you’re only charged for calls that actually run — failed calls aren’t billed.