Three things — only one is downloaded
A tunnel is not needed, no VPN, no open ports, no server to run. It is a URL.
1 · Download
An AI assistant that supports MCP — Claude Desktop, Cursor, VS Code with Copilot, Windsurf, Claude Code. This is the "client".
2 · Copy
The Talaix server URL — pasted into the client's MCP settings. One line of configuration.
3 · Optional
A Talaix account — needed only for governance (per-agent permissions, approvals, audit). Read-only analysis does not need one.
The server URL
Public, read-only, rate-limited to 30 requests/minute per client. No API key required.
What is MCP, in one paragraph
MCP (Model Context Protocol) is a standard that lets an AI assistant call external tools instead of guessing. Talaix publishes its evidence engine as such a server, so any MCP-capable assistant can run a real hazard analysis, verify an asset or pull the data-source registry — with the source, date and confidence attached to every figure.
From zero to your first call
Install an MCP-capable app
Claude Desktop, Cursor or VS Code with Copilot — open it and sign in to your account there.
Open its MCP settings
Usually Settings → MCP, an mcp.json file, or "Connectors" in Claude.
Paste the configuration
Add the Talaix server (see the table below), then restart the app.
Ask a normal question
For example: "Analyse flood and wildfire risk at 45.42, 12.33". The assistant calls the Talaix tools and answers with the evidence.
Add it to your client
Exact wording varies by app version — check your client's documentation if something moved.
| Client | How | Configuration |
|---|---|---|
| Claude Desktop / Claude.ai | Settings → Connectors → "Add custom connector" and paste the URL. Or the config file claude_desktop_config.json. | { "mcpServers": { "talaix": { "url": "https://talaix.com/api/mcp" } } } |
| Cursor | Config file ~/.cursor/mcp.json (global) or .cursor/mcp.json (per project), or Settings → MCP. | { "mcpServers": { "talaix": { "url": "https://talaix.com/api/mcp" } } } |
| VS Code (Copilot) | File .vscode/mcp.json — note the key is servers, not mcpServers. | { "servers": { "talaix": { "type": "http", "url": "https://talaix.com/api/mcp" } } } |
| Claude Code (CLI) | One command in your terminal. | claude mcp add --transport http talaix https://talaix.com/api/mcp |
| Windsurf · Cline · Continue · Zed · LM Studio | All support MCP; add the server from their own MCP file or screen, using the same shape (mcpServers → url). | url: https://talaix.com/api/mcp |
Do you need to download a program or a service?
Yes the client app
The user must have an AI assistant that supports MCP. Some are free (Cursor, VS Code, Claude Code); Claude Pro/Max is a subscription.
No from Talaix
Talaix is a cloud service reached by URL. The user installs nothing from Talaix, runs no server, and needs no VPN or open ports.
What to expect
Transport
- Streamable HTTP request/response — no SSE streaming in this version.
- Public read-only, 30 requests/minute per client.
- Verified live: the server answers with protocolVersion 2025-06-18 and 9 tools.
What you can ask
- "What is the wildfire and flood risk at my site?"
- "Verify this asset against EU Taxonomy DNSH."
- "What is the event history for this location?"
- "Which data sources are available, and which are not?"
Ready to try it?
Point your client at the URL above and call the tools now — no account needed for read-only analysis. When you want named agents, per-tool permissions, approvals and an audit trail, sign in — Agent Access is part of your Talaix account.