mcp-bcrp

MCP Server and Python library for the Banco Central de Reserva del Perú (BCRP) Statistical API. Access over 5,000 macroeconomic indicators directly from your AI agent or Python environment.

Table of Contents

• Overview
• Features
• Requirements
• Installation
• Configuration
• Usage
  • MCP Server
  • Python Library
• Available Tools
• Key Indicators
• Search Engine
• Architecture
• Limitations and Warnings
• Contributing
• License
• Acknowledgments

Overview

The mcp-bcrp package provides a standardized interface to the BCRP statistical database through the Model Context Protocol (MCP). It supports both direct Python usage and integration with AI assistants such as Claude, Gemini, and other MCP-compatible agents.

The library implements:

• Asynchronous HTTP client for efficient data retrieval
• Deterministic search engine with fuzzy matching capabilities
• Spanish language processing for query canonicalization
• Automatic frequency detection (daily, monthly, quarterly, annual)

Features

Requirements

• Python 3.10 or higher
• Internet connection for API requests
• Dependencies: httpx, pandas, fastmcp, rapidfuzz, matplotlib

Installation

From PyPI (when published)

pip install mcp-bcrp

From Source

git clone https://github.com/YOUR_USERNAME/mcp-bcrp.git
cd mcp-bcrp
pip install -e .

With Optional Dependencies

pip install "mcp-bcrp[charts]"  # Include matplotlib for chart generation
pip install "mcp-bcrp[dev]"     # Include development dependencies

Configuration

MCP Server Configuration

Add the following to your MCP configuration file (e.g., mcp_config.json):

{
  "mcpServers": {
    "bcrp-api": {
      "command": "python",
      "args": ["C:/absolute/path/to/mcp_bcrp/run.py"]
    }
  }
}

Environment Variables

Usage

As MCP Server

Once configured, the server can be invoked by MCP-compatible AI assistants:

User: What is the current policy interest rate in Peru?
Agent: [calls search_series("tasa politica monetaria")]
Agent: [calls get_data(["PD04722MM"], "2024-01/2025-01")]

As Python Library

import asyncio
from mcp_bcrp.client import AsyncBCRPClient, BCRPMetadata

async def main():
    # Initialize metadata client
    metadata = BCRPMetadata()
    await metadata.load()
    
    # Search for an indicator (deterministic)
    result = metadata.solve("tasa politica monetaria")
    print(result)
    # Output: {'codigo_serie': 'PD04722MM', 'confidence': 1.0, ...}
    
    # Fetch time series data
    client = AsyncBCRPClient()
    df = await client.get_series(
        series_codes=["PD04722MM"],
        start_date="2024-01",
        end_date="2025-01"
    )
    print(df.head())

asyncio.run(main())

Available Tools (MCP)

Available Prompts

Key Indicators

The following are commonly used indicator codes:

Search Engine

The search engine implements a deterministic pipeline designed for high precision:

Query Input
    │
    ▼
┌─────────────────────────────┐
│  1. Canonicalization        │  Lowercase, remove accents, filter stopwords
└─────────────────────────────┘
    │
    ▼
┌─────────────────────────────┐
│  2. Attribute Extraction    │  Currency (USD/PEN), horizon, component type
└─────────────────────────────┘
    │
    ▼
┌─────────────────────────────┐
│  3. Hard Filters            │  Eliminate series not matching attributes
└─────────────────────────────┘
    │
    ▼
┌─────────────────────────────┐
│  4. Fuzzy Scoring           │  Token sort ratio using RapidFuzz
└─────────────────────────────┘
    │
    ▼
┌─────────────────────────────┐
│  5. Ambiguity Detection     │  Return error if top matches are too close
└─────────────────────────────┘
    │
    ▼
Deterministic Result or Explicit Ambiguity Error

Architecture

mcp_bcrp/
├── __init__.py          # Package initialization and version
├── server.py            # FastMCP server with tool definitions
├── client.py            # AsyncBCRPClient and BCRPMetadata classes
└── search_engine.py     # Deterministic search pipeline implementation

run.py                   # MCP server entry point
bcrp_metadata.json       # Cached metadata (17MB, auto-downloaded)

Limitations and Warnings

Known Limitations

1. Date Format: The BCRP API returns dates in Spanish format (e.g., "Ene.2024"). The library handles this automatically, but custom date parsing may be required for edge cases.

2. Series Availability: Not all series are available for all time periods. The API returns empty responses for unavailable date ranges.

3. Metadata Size: The complete metadata file is approximately 17MB. Initial load may take several seconds on slow connections.

4. Frequency Detection: The library attempts to auto-detect series frequency, but some series may require explicit specification.

Contributing

Contributions are welcome. Please follow these guidelines:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/improvement)
3. Commit changes with descriptive messages
4. Ensure all tests pass (pytest)
5. Submit a pull request

See CONTRIBUTING.md for detailed guidelines.

License

This project is licensed under the MIT License. See LICENSE for the full text.

Acknowledgments

• Banco Central de Reserva del Peru for providing the public statistical API
• FastMCP for the Model Context Protocol framework
• RapidFuzz for fuzzy string matching
• usebcrp for inspiration on BCRP API integration

See Also

Both libraries can be used together to build comprehensive macroeconomic analysis pipelines combining Peru-specific BCRP data with global World Bank indicators.

Disclaimer: This software is provided "as is" without warranty of any kind. The authors are not responsible for any errors in the data or any decisions made based on the information provided by this library.