Resources

Use RailCore in AI coding tools

Connect AI coding tools to these docs over MCP.

What it is

RailCore Developers is an MCP server: it gives an AI assistant these docs, so it can answer from them while you code. Ask it how to make a request, why one failed or which plan fits, and it answers from the same endpoint pages, guides, error codes and plans you read here, with code samples in six languages.

It needs no sign-in and no API key. It answers from the docs and never sends a request to the API, so it spends none of your requests.

Server address

https://mcp.railcore.tech/developers/mcp

Transport: Streamable HTTP. No sign-in.

Claude Code

Run this in a terminal:

Terminal
claude mcp add --transport http railcore-developers https://mcp.railcore.tech/developers/mcp

That adds it for you, in the current project. Add --scope user before the name to have it in every project, or --scope project to save it in the project's .mcp.json for everyone who works on it.

Claude

On claude.ai and in Claude Desktop, add it as a custom connector:

  1. 1. Open Customize > Connectors.
  2. 2. Choose + Add, then Add custom connector.
  3. 3. Name it RailCore Developers and paste the server address.
  4. 4. When asked how people connect, choose No sign in, then Add.

A Free Claude plan can add one custom connector. On Team and Enterprise plans, an owner adds it for everyone under Organization settings > Connectors.

Cursor

.cursor/mcp.json
{
  "mcpServers": {
    "railcore-developers": {
      "url": "https://mcp.railcore.tech/developers/mcp"
    }
  }
}

Save it as .cursor/mcp.json in your project, or as ~/.cursor/mcp.json to have it in every project. If the file already lists servers, add railcore-developers beside them.

VS Code

.mcp.json
{
  "mcpServers": {
    "railcore-developers": {
      "type": "http",
      "url": "https://mcp.railcore.tech/developers/mcp",
      "tools": [
        "*"
      ]
    }
  }
}

Save it as .mcp.json at the root of your workspace. Or run MCP: Add Server from the Command Palette, choose HTTP and paste the server address.

A setup that keeps its servers in .vscode/mcp.json lists them under servers:

.vscode/mcp.json
{
  "servers": {
    "railcore-developers": {
      "type": "http",
      "url": "https://mcp.railcore.tech/developers/mcp"
    }
  }
}

Any other MCP client

Setting, Value table
SettingValue
Addresshttps://mcp.railcore.tech/developers/mcp
TransportStreamable HTTP
Sign-inNone: no API key and no OAuth
MCP version2026-07-28. Clients on the 2025 versions are served too.

What it can answer

Your assistant picks the tool. You can also name one in your question.

Tool, What it gives your assistant table
ToolWhat it gives your assistant
search_docsSearches the guides, the endpoint pages and the error codes, and returns the pages that match.
get_endpointOne endpoint in full: its parameters, what a request to it counts as, the scope its key needs, the plans that include it, whether it must be signed, and the errors it can answer.
get_code_sampleA request to any endpoint in cURL, Node.js, Python, PHP, Go or Java. For a Growth, Scale or Enterprise key, the sample signs the request.
explain_errorWhat an error code or an error response means, and how to fix it. It removes anything shaped like an API key from the text before reading it.
get_plansEvery plan: its price, its monthly requests, its rate limits and what it includes.

As resources, it offers the OpenAPI document, every guide and every endpoint page, for clients that let you attach one to a conversation. Clients that show prompts list three:

Prompt, Use it to table
PromptUse it to
integrate_endpointAdd one endpoint to your code.
debug_api_errorFind out why a request failed and what to change.
choose_planPick a plan for what you are building.

Things to ask

  • Write a Python function that gets the live running status of train 12951 from RailCore.
  • My RailCore request answers 401 SIGNATURE_INVALID. What is wrong with my signing code?
  • Which RailCore plan fits about 20,000 requests a month, and does it need signed requests?

It reads the docs, not the API

The server cannot see your account or your keys. Code it writes runs with your own key, and its requests are counted like any other.

Privacy and limits

  • Keep your API key and signing secret out of the chat: the server never needs them. If a key ends up in an error you ask about, explain_error removes it before reading the error.
  • Each time your assistant uses a tool, the server records the tool's name, your client's name and version, how long the answer took, and your IP address, for rate limiting.
  • Each IP address has a rate limit. If your assistant reaches it, wait a little and ask again.
  • It reads the docs from the docs catalog every 10 minutes, so for a few minutes after the docs change it can still give the earlier answer.

Last updated