Amp

A CLI one-liner or a settings.json block — native Streamable HTTP with headers, like Claude Code.

Setup

1. Create a key in the console and export it as UNZOI_KEY.
2. Run the add command below, or add the amp.mcpServers block to settings.json directly.
3. If you used the workspace file, approve the new server when Amp prompts for it.
4. Ask something that needs current reporting and confirm the tool call in the run.

amp mcp add unzoi --header "x-api-key=$UNZOI_KEY" https://api.unzoi.com/mcp

Or, directly in settings.json

          {
  "amp.mcpServers": {
    "unzoi": {
      "url": "https://api.unzoi.com/mcp",
      "headers": { "x-api-key": "nai_your_key_here" }
    }
  }
}
        

What differs here

• A server added in the workspace file (.amp/settings.json) needs explicit approval before Amp runs it; one added in the global file (~/.config/amp/settings.json) does not — put a server you trust across every project in the global file.
• Because this is a hosted endpoint rather than a process Amp spawns itself, the approval step is about trusting the URL and the key, not about sandboxing local code.
• Amp reads the OpenAPI-adjacent tool descriptions the same way Claude Code does, so it picks list_stories over search_news for the same kind of question — nothing client-specific to configure for that.

Next

• Get a key at console.unzoi.com, if you have not.
• The tool reference — what the model is reading when it chooses between the 18 tools.
• Stories for context windows — the one habit that most changes how well an agent works against this API.
• Limits — an agent that loops can spend a month's credits faster than you expect.
• Amp's own MCP documentation — authoritative for anything about the client itself, including config paths, which move.

Endpoint, for copying: https://api.unzoi.com/mcp