Bring Farina Prima to your agentBETA

Endpoint

A single keyless, public Streamable HTTP endpoint serves every MCP client:

POST https://pandough.app/api/mcp

Stateless JSON-RPC 2.0 over HTTP with SSE streaming. No auth or API key required.

Tools

6 tools are available. All are read-only except plan_bake, which performs a calculation (no data is written).

Connect Claude Desktop (stdio)

Add Pandough to your claude_desktop_config.json using the mcp-remote bridge (it proxies the HTTP endpoint to Claude Desktop's stdio transport):

{
  "mcpServers": {
    "pandough": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://pandough.app/api/mcp"]
    }
  }
}

Restart Claude Desktop. The Pandough tools will appear in the tools menu.

Connect ChatGPT & HTTP clients

Point any HTTP MCP client (ChatGPT custom connectors, your own agent runtime, etc.) at the endpoint and speak JSON-RPC 2.0. List the available tools with:

POST https://pandough.app/api/mcp
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list",
  "params": {}
}

Example: plan_bake

Plan four Neapolitan pizzas at 65% hydration against a flour blend (70% Caputo Pizzeria + 30% Manitoba Oro), with the baker's real kitchen and fridge temperatures, a short warm bulk, an overnight cold ferment, and an Italian calculator link:

Request

POST https://pandough.app/api/mcp
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "plan_bake",
    "arguments": {
      "recipeSlug": "neapolitan-pizza",
      "flourBlend": [
        { "flourSlug": "caputo-pizzeria", "percent": 70 },
        { "flourSlug": "manitoba-oro", "percent": 30 }
      ],
      "hydrationPercent": 65,
      "servings": 4,
      "roomTempCelsius": 22,
      "fridgeTempCelsius": 4,
      "roomTempHours": 4,
      "fridgeTempHours": 20,
      "locale": "it"
    }
  }
}

Sample response

{
  "jsonrpc": "2.0",
  "id": 2,
  "result": {
    "content": [
      {
        "type": "text",
        "text": "# Bake Plan\n\n**Total dough:** 1120g\n**Yeast:** 0.4g (0.06%) instant-dry\n\n## Ingredients\n  Pizzeria-ORO: 671g (100%)\n  Water: 436g (65%)\n  Salt: 19g (2.8%)\n  Yeast: 0.4g (0.06%)\n\n## Schedule\n  Warm bulk fermentation: 4h at 22°C\n  Cold fermentation (fridge): 20h at 4°C\n  Final proof: 2h at 22°C\n\n## Warnings\n  - 65% hydration is above Pizzeria-ORO's recommended range (57–64%) — expect a slack, hard-to-handle dough.\n\n## Open in Pandough\nOpen this calculator link to continue editing the bake:\nhttps://pandough.app/it/calculator?recipe=neapolitan-pizza&n=4&w=280&h=65&rt=4&ft=20&y=0.06&blendInline=caputo-pizzeria:70,manitoba-oro:30&source=mcp&aid=1a2b3c4d"
      }
    ]
  }
}

The yeast amount, schedule, and warnings are computed by the same engine that powers the Pandough calculator — and they're flour-aware: passing a flourSlug or flourBlend tailors the recommended hydration band and surfaces a warning when your target hydration falls outside it. Room and fridge temperatures (roomTempCelsius / fridgeTempCelsius) are the dominant driver of the yeast amount and timing, so pass them whenever the user knows them. The Open in Pandough link prefills every parameter — including the blend via blendInline — so the user can keep editing in the app, even when signed out.

Rate limits

The public endpoint applies a best-effort per-IP rate limit:

• •30 requests / minute per IP
• •300 requests / hour per IP

Exceeding a limit returns HTTP 429 with a Retry-After header and a JSON-RPC error (code -32029). Back off and retry after the indicated number of seconds.