pyfolio-performance

Python library + MCP server to read Portfolio Performance XML files. Access your portfolio data via AI agents for investment analysis.

Features

• Parse Portfolio Performance XML export files
• Access accounts, depots, securities, transactions
• MCP server for AI agents (Claude Desktop, Cursor, OpenCode, kiro-cli)
• Price history tracking

Local Development

# Create virtual environment
python3 -m venv .venv

# Activate
source .venv/bin/activate  # Linux/Mac
# or: .venv\Scripts\activate  # Windows

# Install dependencies
pip install -r requirements.txt

Run the regression tests:

```bash
python -m pytest
ruff check .

Portfolio Performance MCP Server

This project includes an MCP (Model Context Protocol) server that exposes your Portfolio Performance data to AI agents for investment analysis and advice.

Quick Setup (native with pyenv/venv)

# 1. Use Python 3.12 via pyenv, then create and activate virtual environment
pyenv install 3.12  # if needed
python3 -m venv .venv
source .venv/bin/activate  # Linux/Mac

# 2. Install dependencies
pip install -r requirements.txt

# 3. Test the server
python mcp_server.py

The devcontainer uses .venv-devcontainer instead. Activate the matching venv before starting opencode, because MCP commands use plain python.

Configuration

The server auto-loads the portfolio on startup from:

1. PORTFOLIO_FILE environment variable
2. Defaults to kommer.xml in project root

Available Tools

Tool	Description
ping	Health check — verify server is running
load_portfolio	Load a portfolio XML file
reload_portfolio	Reload current portfolio without restart
get_portfolio_summary	Full overview: cash, depot values, P/L, per-account and per-depot breakdowns with invested/profit and full holdings
get_accounts	All accounts with cash balances
get_account_by_name	Single account by name
get_depots	All depots with full security holdings
get_depot_by_name	Single depot by name
get_securities	All securities with ISIN, WKN, price, value, ticker, currency, custom attributes
get_transactions	All transactions with depot and account info
get_transactions_by_type	Filter by type (BUY, SELL, DIVIDENDS, etc.)
get_transactions_by_year	Filter by year
get_transactions_for_security	Filter by security, optionally by depot and type
get_security_by_name / by_isin / by_wkn	Look up a security
get_security_price_history	Historical daily closing prices
get_performance_by_year	Yearly totals grouped by transaction type

OpenCode Integration

Add this to your project opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "portfolio": {
      "type": "local",
      "command": ["python", "mcp_server.py"],
      "environment": {
        "PORTFOLIO_FILE": "kommer.xml"
      },
      "enabled": true
    }
  }
}

Then activate the native .venv or container .venv-devcontainer and run opencode from the project directory. The portfolio auto-loads on startup.

ExtraETF MCP Server

A lightweight MCP server that fetches ETF holdings, factsheets, allocations, and dividend data from extraetf.com. No API key needed — data is extracted from the ETF profile page.

Tools

Tool	Description
get_etf_holdings	Top holdings with ISIN, weight, sector, country
get_etf_factsheet	TER, fund size, replication, performance, risk metrics
get_etf_allocation	Country, region, sector, currency, asset class breakdown
get_etf_dividends	Dividend history, yields, CAGR
ping	Health check

OpenCode Integration

Add a second MCP server to opencode.json:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "portfolio": { ... },
    "extraetf": {
      "type": "local",
      "command": ["./venv/bin/python", "extraetf_server.py"],
      "enabled": true
    }
  }
}

yfinance MCP server

yfinance pairs well with this project — get prices, fundamentals, and news for individual stocks alongside your portfolio and ETF data.

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "yfinance": {
      "type": "local",
      "command": ["docker", "run", "-i", "--rm", "narumi/yfinance-mcp"],
      "enabled": true
    }
  }
}

Sample Prompt

You are a portfolio analyst. Analyze this portfolio across:
- Strategy clarity
- Portfolio structure & diversification
- Position quality & redundancies
- Risk analysis (sector, regional, career correlation)
- Performance overview
- Concrete recommendations & target portfolio
- Scenario forecasts (conservative/realistic/optimistic)
- Market valuation & phase
- Holistic wealth perspective
- Financial independence timeline
- Complexity & robustness scores (KPI)
- Long-term sustainability

Use the MCP tools to fetch data, then provide actionable insights.

Tests

The test suite includes small XML fixtures copied from the upstream Portfolio Performance project under tests/fixtures/original_project/. They cover account balances, depot holdings, tax/fee transactions, delivery transactions, portfolio transfers, single-node XML parsing, and class-level cache resets.

Run:

python -m pytest
ruff check .

*.xml files are ignored by default to avoid committing personal portfolio data. Only the public fixture XML files under tests/fixtures/original_project/ are explicitly unignored.

Docs

https://pyfolio-performance.readthedocs.io/en/latest/

Changelog

See FIXES.md for a record of bugs fixed in this fork.