MCP

Crowe Logic is a remote Model Context Protocol server at https://api.crowelogic.com/mcp. Any MCP client connects with a Crowe Logic personal access token as the Bearer header and gets the same control plane the console, the desktop app and the CLI use. The plan on your Crowe ID is the plan the server honours: a free account answers on CroweLM Flash with the daily cap, a paid plan unlocks the tiers and the frontier engines it names. Plans are compared at /pricing.

Tools

ToolWhat it does
crowelm_askOne prompt to a CroweLM tier or a frontier engine offered by name. Returns the answer. Metered against your plan exactly as the console meters it.
crowelm_modelsThe models your plan unlocks, each with its engine, maker, host and minimum plan, plus the default tier for your plan.
crowe_accountYour plan and this month's usage against its flagship and standard budgets, with the pricing and account links.

Every tier names the engine it runs on, the same way the model table does. Asking for a model above your plan returns an error that names the plan required rather than a bare refusal.

Get a token

  1. Sign in at api.crowelogic.com/login with your Crowe ID. A new account is the free tier; no card.
  2. Open api.crowelogic.com/account and create a personal access token. It starts with crowe_pat_ and is shown once.

A Crowe ID access token also works as the Bearer token for the length of its lifetime. The personal access token is the one to put in a config file, and you can revoke it from the same page.

Claude Code

claude mcp add --transport http crowe-logic https://api.crowelogic.com/mcp \
  --header "Authorization: Bearer crowe_pat_..."

VS Code and GitHub Copilot

.vscode/mcp.json in a repository, or the user level mcp.json. The input prompt keeps the token out of the file.

{
  "inputs": [
    { "id": "crowe-logic-token", "type": "promptString",
      "description": "Crowe Logic personal access token", "password": true }
  ],
  "servers": {
    "crowe-logic": {
      "type": "http",
      "url": "https://api.crowelogic.com/mcp",
      "headers": { "Authorization": "Bearer ${input:crowe-logic-token}" }
    }
  }
}

GitHub Copilot app

Add server, type HTTP, URL https://api.crowelogic.com/mcp, one header named Authorization with the value Bearer crowe_pat_.... Leave the OAuth client ID empty.

Cursor

{
  "mcpServers": {
    "crowe-logic": {
      "url": "https://api.crowelogic.com/mcp",
      "headers": { "Authorization": "Bearer crowe_pat_..." }
    }
  }
}

GitHub Copilot coding agent

Repository settings, Copilot, Coding agent, MCP configuration. Store the token first as an Agents secret named COPILOT_MCP_CROWE_LOGIC_TOKEN in the copilot environment.

{
  "mcpServers": {
    "crowe-logic": {
      "type": "http",
      "url": "https://api.crowelogic.com/mcp",
      "headers": { "Authorization": "Bearer $COPILOT_MCP_CROWE_LOGIC_TOKEN" },
      "tools": ["*"]
    }
  }
}

Check it from the shell

Without a token the server answers 401 and points at its metadata. With one, initialize and then tools/list.

$ curl -si -X POST https://api.crowelogic.com/mcp \
  -H 'Accept: application/json, text/event-stream' -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}' | head -3
HTTP/2 401
www-authenticate: Bearer error="invalid_token", error_description="Authentication required",
  resource_metadata="https://api.crowelogic.com/.well-known/oauth-protected-resource/mcp"

$ TOKEN=crowe_pat_...
$ curl -s -X POST https://api.crowelogic.com/mcp -H "Authorization: Bearer $TOKEN" \
  -H 'Accept: application/json, text/event-stream' -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}'

How the sign-in works

The server follows the MCP authorization specification. Its metadata at /.well-known/oauth-protected-resource/mcp names Crowe ID, https://id.crowelogic.com/realms/crowe, as the authorization server, and the Bearer token is resolved by the same code path as every other Crowe Logic surface, so the plan gate, the free daily cap, the token budgets and the usage ledger are shared. Clients that try the browser sign-in flow need Crowe ID to accept their client registration; until that is switched on, use the personal access token header shown above.

Registry

The server is listed in the official MCP Registry as io.github.MichaelCrowe11/crowe-logic, so clients that browse the registry find it by name. Source, tests and this page's reference copy live in the crowe-logic-foundry repository.