Anny AI Help Center
Sign in

    Connect Anny AI — For Agents

    Add Anny AI as a remote MCP server with the exact address below. Do not substitute another hostname, ask for an API key, or add a bearer token.

    https://www.anny.ai/api/mcp

    The server uses remote Streamable HTTP. Authentication opens in the user’s browser. Prefer a user-level configuration unless the user explicitly asks to share the connection with a project.

    Claude Code

    Run:

    claude mcp add --transport http --scope user anny-rates https://www.anny.ai/api/mcp

    Start Claude Code and open /mcp. Select anny-rates and complete the browser sign-in if Claude has not already opened it.

    Check the current Claude Code MCP guide.

    Codex

    Run:

    codex mcp add anny-rates --url https://www.anny.ai/api/mcp
    codex mcp login anny-rates

    The second command opens the Anny AI sign-in. Codex CLI, the Codex IDE extension, and the ChatGPT desktop app share this MCP configuration on the same Codex host.

    Check the current Codex MCP guide.

    Cursor

    Merge this entry into the user’s ~/.cursor/mcp.json. Use .cursor/mcp.json only when the user wants the connection scoped to one project.

    {
      "mcpServers": {
        "anny-rates": {
          "url": "https://www.anny.ai/api/mcp"
        }
      }
    }

    Open Cursor Settings → Tools & MCP, enable anny-rates, and complete the browser sign-in.

    Check the current Cursor MCP guide.

    VS Code

    Run MCP: Open User Configuration from the Command Palette, then merge this entry into the servers object:

    {
      "servers": {
        "anny-rates": {
          "type": "http",
          "url": "https://www.anny.ai/api/mcp"
        }
      }
    }

    Start the server from the configuration editor or run MCP: List Servers. Confirm trust and complete the browser sign-in.

    Check the current VS Code MCP guide.

    Verify before reporting success

    1. Confirm anny-rates is enabled and connected.
    2. Ask it to list the user’s Anny AI organizations.
    3. Ask the user to choose an organization when more than one is available.
    4. List the visible hotels in that organization.

    You can also ask the assistant to search the Anny AI Help Center. Use a question such as “Find the Help Center guide for adding a competitor and open it.” The assistant can use knowledge_search to find the article and knowledge_get to open the current guide. Set absoluteLinks to true when the returned Markdown will be read outside Anny AI so its Help Center links are ready to open.

    Read Estimated demand

    Rate results can include estimatedDemand for the matching stay date and length of stay. A rate row or summary row returns one object or null; matrix and trend tools return arrays. The same public shape is available through the rates API, Lighthouse-compatible rates, and MCP rate tools:

    {
      "arrivalDate": "2026-08-21",
      "losNights": 1,
      "level": "high",
      "availabilityRatioPct": 42,
      "eligibleCompetitorCount": 4,
      "totalCompetitorCount": 5,
      "unavailableReason": null,
      "updatedAt": "2026-08-08T14:00:00.000Z"
    }

    level is low, medium, high, or null. A null level means no demand level is available for that stay date and length of stay. availabilityRatioPct counts the room options the Comp Set is listing as a percentage of its own typical number, not a share of unsold rooms, so 100 is an ordinary day rather than an empty market. Below 100 means competitors are listing fewer room options than usual, which is the signal behind a higher level. Report demand from level and use the ratio only as supporting detail; never convert it into an occupancy percentage or present it as how full the market is. Keep arrivalDate and losNights with the result. Rate summaries include current demand, and trend tools include demand history for the selected stay date.

    Do not ask the user for organization, hotel, Comp Set, shop, or report IDs. Resolve names with the Anny AI list and read tools. Read current state before a change, and ask for confirmation before starting a rate shop.

    Help improve the tools

    When a tool is missing, awkward, fails, or gives an unclear or incorrect result, explain what happened and offer to send Anny AI a suggestion. Use suggestion_submit only after the user agrees or explicitly asks you to submit feedback.

    Keep the suggestion practical: include the task you attempted, what went wrong, and the change that would make the workflow easier. Do not include passwords, access tokens, or customer data that the recommendation does not need. Anny AI identifies the connected user and routes the request automatically.

    See What your AI assistant can do for a plain-language guide to every available Anny AI tool.

    For product questions and Help Center guidance, see Get help from Ask Anny.

    For Claude or ChatGPT’s browser setup, send the user to Connect Anny AI — For Humans.