JACTUS MCP Server
Enables AI assistants like Claude to discover, validate, and simulate all 18 ACTUS financial contract types using JACTUS, with tools for contract simulation, risk computation, and portfolio analysis.
README
JACTUS MCP Server
Standalone MCP server for JACTUS — Enables AI assistants like Claude Code to directly access JACTUS financial contract simulation capabilities.
What is this?
Model Context Protocol (MCP) is a protocol that allows AI assistants to interact with external tools. This package gives Claude (and other MCP-compatible assistants) the ability to discover, validate, and simulate all 18 ACTUS financial contract types powered by JACTUS.
Installation
pip install jactus-mcp
This installs the MCP server and pulls jactus as a dependency automatically.
Download Docs & Examples
After installing, optionally download JACTUS documentation and examples from GitHub:
jactus-mcp setup
This downloads the matching version's docs/ and examples/ into ~/.jactus-mcp/data/. You can also specify a tag:
jactus-mcp setup --tag v0.2.0 # specific version
jactus-mcp setup --force # re-download
jactus-mcp status # check what's downloaded
jactus-mcp clean # remove downloaded data
From GitHub
pip install git+https://github.com/pedronahum/JACTUS-MCP.git
For Development
git clone https://github.com/pedronahum/JACTUS-MCP.git
cd JACTUS-MCP
pip install -e ".[dev]"
Configuration
Claude Code
Add to your Claude Code MCP settings (.mcp.json or settings):
{
"mcpServers": {
"jactus": {
"command": "jactus-mcp"
}
}
}
Or using Python module:
{
"mcpServers": {
"jactus": {
"command": "python",
"args": ["-m", "jactus_mcp"]
}
}
}
Claude Desktop
Add to your Claude Desktop configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"jactus": {
"command": "jactus-mcp"
}
}
}
Transport Options
# Default: stdio transport (for local clients)
jactus-mcp
# or
python -m jactus_mcp
# Streamable HTTP transport (for remote clients)
python -m jactus_mcp --transport streamable-http
Features
Core Tools (always available)
These tools work with just pip install jactus-mcp — no source tree needed:
| Tool | Description |
|---|---|
jactus_list_contracts |
List all 18 contract types by category |
jactus_get_contract_info |
Get detailed info about a contract type |
jactus_get_contract_schema |
Get required/optional parameters for all 18 types |
jactus_get_event_types |
List all ACTUS event types |
jactus_list_risk_factor_observers |
List all risk factor observer types |
jactus_simulate_contract |
Simulate a contract and get structured cash flows |
jactus_validate_attributes |
Validate contract attributes before simulation |
jactus_compute_risk |
Compute DV01, delta, gamma, PV01 risk metrics |
jactus_simulate_portfolio |
Simulate a portfolio with aggregation |
jactus_get_topic_guide |
Structured guides on 7 topics |
jactus_get_quick_start |
Built-in PAM quick start example |
jactus_health_check |
Verify installation and JACTUS availability |
jactus_get_version_info |
Version information |
Docs & Example Tools (requires setup)
These tools require JACTUS docs/examples. Enable them with jactus-mcp setup:
| Tool | Description |
|---|---|
jactus_search_docs |
Search across JACTUS documentation |
jactus_get_doc_structure |
Browse documentation files and headers |
jactus_list_examples |
List Python scripts and notebooks |
jactus_get_example |
Retrieve example source code |
jactus_run_example |
Execute an example |
Alternatively, point to a local JACTUS checkout:
export JACTUS_ROOT=/path/to/JACTUS
MCP Resources
| Resource URI | Description |
|---|---|
jactus://docs/architecture |
System architecture guide |
jactus://docs/pam |
PAM contract walkthrough |
jactus://docs/derivatives |
Derivative contracts guide |
jactus://docs/readme |
Project overview |
jactus://contract/{type} |
Dynamic contract info + schema |
MCP Prompts
| Prompt | Description |
|---|---|
create_contract |
Guide to create a new contract |
troubleshoot_error |
Help troubleshoot errors |
understand_contract |
Explain a contract type |
compare_contracts |
Compare two contract types |
Quick Example
With the MCP server running, ask Claude:
"Simulate a $100k PAM loan at 5% interest maturing in 1 year"
Claude will:
- Call
jactus_get_contract_schema("PAM")to get required fields - Call
jactus_simulate_contractwith the attributes - Return structured cash flow events (IED, IP, MD) with payoff amounts
Contract Types
JACTUS implements all 18 ACTUS contract types:
- Principal: PAM, LAM, LAX, NAM, ANN, CLM
- Non-Principal: UMP, CSH, STK
- Exotic: COM
- Derivatives: FXOUT, OPTNS, FUTUR, SWPPV, SWAPS, CAPFL, CEG, CEC
Development
Running Tests
# Unit tests
pytest tests/ -v --ignore=tests/test_mcp_integration.py
# Integration tests (requires MCP stdio client)
pytest tests/test_mcp_integration.py -v
# All tests
pytest tests/ -v
Architecture
JACTUS-MCP/
├── src/jactus_mcp/
│ ├── __init__.py # Package version
│ ├── __main__.py # python -m jactus_mcp entry point
│ ├── server.py # FastMCP server (tools, resources, prompts)
│ ├── models.py # Pydantic response models
│ └── tools/
│ ├── _utils.py # Shared utilities (get_jactus_root, type conversion)
│ ├── contracts.py # Contract discovery & schema (18 types)
│ ├── simulate.py # Contract simulation
│ ├── examples.py # Example retrieval & execution
│ ├── validation.py # Attribute validation
│ ├── documentation.py # Documentation search & topic guides
│ ├── risk.py # Risk analytics (DV01, delta, gamma)
│ └── system.py # Health checks & version info
├── tests/ # Comprehensive test suite
├── pyproject.toml # Package configuration
└── README.md
License
Apache License 2.0
Links
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。