RealTest MCP Server
Provides structured access to RealTest backtesting documentation and example scripts to help LLM agents generate accurate RealScript code. It offers tools for semantic search, authoritative function references, and verified script retrieval to prevent hallucinations.
README
RealTest MCP Server
An MCP server that gives LLM agents structured access to RealTest backtesting documentation and example scripts. Fixes RealScript hallucinations by providing authoritative function references, searchable docs, and verified example scripts at query time.
Requirements
- Python 3.11+
- RealTest installed (for the User Guide PDF and example scripts)
Setup
1. Clone and install
git clone <repo-url>
cd realtest-mcp
pip install -e .
2. Configure paths
Edit config.toml to match your RealTest installation:
[realtest]
pdf_path = "C:\\RealTest\\RealTest User Guide.pdf"
[scripts]
examples = "C:\\RealTest\\Scripts\\Examples"
user_scripts = []
[database]
path = "%LOCALAPPDATA%\\RealTestMcp\\chromadb"
All paths support %VAR% environment variable expansion. Individual fields can
also be overridden via environment variables: REALTEST_MCP_PDF_PATH,
REALTEST_MCP_EXAMPLES, REALTEST_MCP_DB_PATH.
3. Ingest docs and scripts
python -m realtest_mcp ingest
python -m realtest_mcp status
4. Configure your MCP client
Add to .mcp.json in your project (or pass via --mcp-config):
{
"mcpServers": {
"realtest": {
"command": "python",
"args": ["-m", "realtest_mcp", "serve"]
}
}
}
5. (Optional) Add your own scripts
Add directories to user_scripts in config.toml, then re-run ingest.
Commands
| Command | Description |
|---|---|
serve |
Start MCP server (StdIO transport, managed by client) |
ingest |
Extract PDF docs, index scripts into ChromaDB |
status |
Show database statistics (chunk counts, timestamps) |
MCP Tools
Six tools are exposed to MCP clients:
| Tool | Description |
|---|---|
get_primer |
Load the RealScript mental model (call once per session) |
get_reference |
Look up exact element docs with alias resolution |
get_section |
Fetch narrative documentation sections by title |
list_elements |
Browse element categories and their contents |
search_docs |
Semantic search over documentation |
search_scripts |
Find example scripts by concept or technique |
See API Reference for full parameter details and usage examples.
Documentation
- Architecture Overview -- system design, components, data flow
- API Reference -- tool parameters, resolution logic, workflows
- Roadmap -- proposed enhancements
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。