RCA-MCP

io.github.dave1362由 GitHub 账号 dave1362 发布v4.1.20更新于 Oct 24, 2026

Causal root-cause analysis MCP server -- 56 tools across graphs, RCA models, PyRCA.

已验证STDIO仅桌面数据与分析 AI 与机器学习

服务详情

标识符: io.github.dave1362/rca-mcp-connector
传输: STDIO
运行时: Desktop only
信任: Verified
发布于: Sep 16, 2026

安全提示

MCP 服务会接收对话中的工具调用请求，并可能执行外部操作。启用前请先了解该服务能做什么，并只授予它所需的权限。

注册中心已验证此条目所属的 GitHub 账号 dave1362。这并不表明该账号就是产品方。

SourceWeft 的检查只涵盖条目是否完整、端点是否响应；不审查服务器的代码，也不审查其工具的行为。

在 README 中添加徽章

向读者展示该 MCP 服务已收录于 SourceWeft，徽章链接到本页。

[![Listed on SourceWeft](https://sourceweft.com/badge/mcp/io.github.dave1362%2Frca-mcp-connector.svg)](https://sourceweft.com/mcp/io.github.dave1362%2Frca-mcp-connector?utm_source=readme&utm_medium=badge)

概览

AI 生成的概览

本地 MCP 连接器，为助手提供 56 个因果根因分析工具：构建因果图、运行 RCA 模型与 PyRCA 算法并生成报告。

可以做什么: 构建并操作用于根因分析的因果图。运行 10 类 RCA 模型及 3 种基于 PyRCA 的算法，并进行多模型共识比较。生成 PDF、HTML、Excel 和 Markdown 格式的报告。
适用场景: 适合排查系统、服务或指标异常的原因，需要助手构建因果图、运行根因模型并输出报告，而不仅是汇总日志的场景。
运行要求: 以本地 Python 包（rca-mcp-connector）运行，可通过 uvx 或 pip 启动。需要 RCA_MCP_API_KEY 中的 API 密钥，并需访问后端网络，后端地址由 RCA_MCP_API_URL 设置。README 说明默认自定义域名尚未启用，需配置当前后端地址。

安装前请注意

需要 API 密钥（RCA_MCP_API_KEY），该密钥会发送给第三方后端，应视为机密。分析数据由该远程服务处理。README 称密钥免费且无需信用卡，但使用前应确认当前条款。

安装

在 SourceWeft 中

打开控制台中的 RCA-MCP，将其添加到工作区。
为需要使用其工具的对话启用该服务。

Desktop only，通过 STDIO。 STDIO 服务会启动本地进程，因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照仓库中的启动说明。

README

RCA-MCP Connector

[Python] [License] [PulseMCP]

Note: the api.rca-mcp.com custom domain isn't wired up yet — point RCA_MCP_API_URL at the current backend URL below instead.

What is RCA-MCP?

The only MCP server purpose-built for causal Root Cause Analysis. 56 tools covering causal graph construction, 10 RCA model families plus 3 dedicated PyRCA algorithms (Salesforce PyRCA, BSD-3-Clause), multi-model consensus, and PDF/HTML/Excel/Markdown report generation. Works with Claude, Ollama, Groq, OpenAI, Gemini, LangChain, Cursor — 10 providers.

Quick Start (2 minutes)

rca-mcp-connector is a published PyPI package — no clone needed. Point any MCP client at it with uvx (or pip install rca-mcp-connector if you'd rather manage the install yourself):

Get a free API key at rcamcp.datalizedglb.cloud — no credit card required — then set RCA_MCP_API_KEY in your MCP client's config (examples below).

Claude Code Setup

Add to .mcp.json in your workspace root:

json

{  "mcpServers": {    "rca-mcp": {      "command": "uvx",      "args": ["rca-mcp-connector"],      "env": {        "RCA_MCP_API_URL": "https://rcamcp-production.up.railway.app",        "RCA_MCP_API_KEY": "your_api_key_here"      }    }  }}

{  "mcpServers": {    "rca-mcp": {      "command": "uvx",      "args": ["rca-mcp-connector"],      "env": {        "RCA_MCP_API_URL": "https://rcamcp-production.up.railway.app",        "RCA_MCP_API_KEY": "your_api_key_here"      }    }  }}

Ollama Setup

bash

go install github.com/mark3labs/mcphost@latestmcphost -m ollama:qwen3:14b --config providers/mcp-servers.json

go install github.com/mark3labs/mcphost@latestmcphost -m ollama:qwen3:14b --config providers/mcp-servers.json

OpenAI Agents SDK

python

from agents import Agent, MCPServerStdioimport asyncio
async def main():    async with MCPServerStdio(        params={            "command": "uvx",            "args": ["rca-mcp-connector"],            "env": {                "RCA_MCP_API_URL": "https://rcamcp-production.up.railway.app",                "RCA_MCP_API_KEY": "your_api_key_here",            },        }    ) as rca_server:        agent = Agent(name="RCA Agent", model="gpt-4o", mcp_servers=[rca_server])        result = await agent.run("Find the root cause of the API latency spike.")        print(result.final_output)
asyncio.run(main())

from agents import Agent, MCPServerStdioimport asyncio
async def main():    async with MCPServerStdio(        params={            "command": "uvx",            "args": ["rca-mcp-connector"],            "env": {                "RCA_MCP_API_URL": "https://rcamcp-production.up.railway.app",                "RCA_MCP_API_KEY": "your_api_key_here",            },        }    ) as rca_server:        agent = Agent(name="RCA Agent", model="gpt-4o", mcp_servers=[rca_server])        result = await agent.run("Find the root cause of the API latency spike.")        print(result.final_output)
asyncio.run(main())

LangChain

python

from langchain_mcp_adapters.client import MultiServerMCPClientfrom langchain_anthropic import ChatAnthropicfrom langgraph.prebuilt import create_react_agentimport asyncio
async def main():    async with MultiServerMCPClient({        "rca-mcp": {            "command": "uvx", "args": ["rca-mcp-connector"],            "env": {                "RCA_MCP_API_URL": "https://rcamcp-production.up.railway.app",                "RCA_MCP_API_KEY": "your_api_key_here",            },            "transport": "stdio",        }    }) as client:        tools = await client.get_tools()        agent = create_react_agent(ChatAnthropic(model="claude-sonnet-4-6"), tools)        result = await agent.ainvoke({"messages": [{"role": "user", "content": "Run an FMEA analysis"}]})        print(result["messages"][-1].content)
asyncio.run(main())

from langchain_mcp_adapters.client import MultiServerMCPClientfrom langchain_anthropic import ChatAnthropicfrom langgraph.prebuilt import create_react_agentimport asyncio
async def main():    async with MultiServerMCPClient({        "rca-mcp": {            "command": "uvx", "args": ["rca-mcp-connector"],            "env": {                "RCA_MCP_API_URL": "https://rcamcp-production.up.railway.app",                "RCA_MCP_API_KEY": "your_api_key_here",            },            "transport": "stdio",        }    }) as client:        tools = await client.get_tools()        agent = create_react_agent(ChatAnthropic(model="claude-sonnet-4-6"), tools)        result = await agent.ainvoke({"messages": [{"role": "user", "content": "Run an FMEA analysis"}]})        print(result["messages"][-1].content)
asyncio.run(main())

See providers/ for ready-to-use config templates and full examples (Groq, Gemini, OpenRouter, Claude Desktop).

Third-Party Licences

Algorithms in rca_pyrca_* tools are independently-written adaptations of PyRCA's published methods (Zheng et al. 2023, arXiv:2306.11417), not direct copies of PyRCA source code, per the private API's models/pyrca_adapter.py.

Citing RCA-MCP

bibtex

@software{rcamcp2026,  title  = {RCA-MCP: An MCP Server for Causal Root Cause Analysis},  author = {dave1362},  year   = {2026},  url    = {https://github.com/dave1362/rca-mcp-connector},  note   = {v4.1.20}}

@software{rcamcp2026,  title  = {RCA-MCP: An MCP Server for Causal Root Cause Analysis},  author = {dave1362},  year   = {2026},  url    = {https://github.com/dave1362/rca-mcp-connector},  note   = {v4.1.20}}

来源：README.md，提交 25a2a64

工具

SourceWeft 尚未在沙箱中启动过该软件包，因此其工具未知。

版本历史

v4.1.20最新Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

近 30 天内可连通25 分
请保持端点能够响应 MCP 握手；SourceWeft 每 30 天探测一次。
已声明认证方式10 分
请在注册中心条目中声明凭据请求头，或保持端点能够响应探测，以便判定其认证方式。
已列出工具15 分
请让服务器在无凭据时响应 tools/list，或发布可在沙箱中无凭据启动的软件包。
描述必需15 分
图标10 分
文档15 分
许可证10 分