pinelabs-online-mcp: Skill for Claude Code

Pine Labs MCP Tool Generator

Generate complete MCP tool implementations from Pine Labs API documentation, curl commands, or request/response examples.

Input Formats

The user may provide any of:

1. Pine Labs docs URL — fetch and extract endpoint details
2. curl command — parse method, path, headers, body, and response
3. Request/response JSON — infer params and types
4. API specification — method, path, parameters

If the tool name is not provided or cannot be inferred, ask the user before proceeding.

Workflow

- [ ] Step 1: Parse input and extract API contract
- [ ] Step 2: Determine tool name and resource file
- [ ] Step 3: Implement tool function
- [ ] Step 4: Register tool in tools.py
- [ ] Step 5: Add Pydantic models (if needed)
- [ ] Step 6: Write unit tests
- [ ] Step 7: Run linter — fix errors
- [ ] Step 8: Run tests — fix errors

Step 1: Parse Input

Extract: HTTP method, endpoint path, required/optional parameters with types, and example response.

Map parameters:

• str → validate_resource_id() or raw string param
• int / float → typed parameter
• dict → Pydantic model
• list → typed list parameter
• bool → boolean parameter

Step 2: Determine Tool Name and File

Naming conventions:

• Fetch single: get_{resource} or get_{resource}_by_{field}
• Fetch list: get_{resources} or list_{resources}
• Create: create_{resource}
• Update: update_{resource}
• Delete: delete_{resource} or cancel_{resource}

Place in pkg/pinelabs/{resource_type}.py. Create new file only if needed.

Step 3: Implement Tool

import json
import logging

from fastmcp import FastMCP

from pkg.pinelabs.client import PineLabsClient, PineLabsAPIError
from pkg.pinelabs.utils.errors import (
    api_error_response,
    validation_error_response,
    unexpected_error_response,
)
from pkg.pinelabs.utils.validators import validate_resource_id

logger = logging.getLogger("pinelabs-mcp-server.{resource}")


def register_{resource}_tools(
    mcp: FastMCP, client: PineLabsClient
) -> None:

    @mcp.tool(
        name="tool_name",
        description=(
            "Action verb + what it does. "
            "When to use / prerequisites. "
            "Constraints or return format."
        ),
    )
    async def tool_name(required_param: str) -> str:
        rid = validate_resource_id(required_param, "param_name")

        try:
            result = await client.get(f"/path/{rid}")
            return json.dumps(result, indent=2)
        except PineLabsAPIError as e:
            return api_error_response(
                e.message, e.code, e.status_code
            )
        except Exception as e:
            return unexpected_error_response(e, "context")