Circlfi Mcp

io.github.negm17111995由 GitHub 账号 negm17111995 发布v1.1.0更新于 Oct 24, 2026

13-model stock valuation engine for AI agents - fair values for 5,900+ US stocks, updated daily.

已验证Streamable HTTP可网页运行金融数据与分析

概览

AI 生成的概览

CirclFi MCP 让 AI 助手获取 5,900 多只美股的公允价值数据,涵盖 13 种估值模型、质量评分与市场筛选器。

可以做什么
  • 查询单只美股的内在价值档案:公允价值、隐含上涨空间、各模型置信度、公司质量评分与价值陷阱风险评分
  • 对比多只股票或多个模型的估值结果
  • 按最低质量评分、最高价值陷阱评分或行业筛选全部美股,输出 JSON 或 CSV(付费版)
适用场景
适合需要助手回答美股内在价值、价格与公允价值差距、公司质量或价值陷阱风险等问题,或按质量与风险条件筛选市场时使用。免费版无需账号即可查询每只股票的三种模型。
运行要求
可使用托管的远程端点(Streamable HTTP),或用 npx 运行 npm 包 circlfi-mcp 的本地进程(需 Node.js)。免费版无需账号或密钥;付费版使用 Gumroad 购买邮箱,本地设为 CIRCLFI_EMAIL,远程则作为 Authorization bearer 令牌发送。需要能访问 CirclFi API 的网络。
安装前请注意
付费权限绑定 Gumroad 购买邮箱,该邮箱会作为 bearer 令牌或环境变量发送,应视为凭据。付费价格为每月 39 美元或每年 299 美元。工具调用会转发至 CirclFi API,股票代码与查询内容会离开本机。输出为量化模型数据,仅供教育用途,不构成投资建议。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Circlfi Mcp,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Web executable,通过 Streamable HTTP。 远程服务在工作区中配置后即可从网页运行时运行。

交给智能体安装

把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体,或 SourceWeft 对话中。智能体会读取该服务器的安装指南,先向你展示发布者、端点和登录方式,经你同意后再把它添加到你的客户端。密钥不会写进任何命令。

阅读 https://sourceweft.com/mcp/io.github.negm17111995%2Fcirclfi-mcp/install.md,并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前,先告诉我它的发布者、端点和登录方式,等我确认。不要把 API 密钥写进命令或我会提交的文件;问我密钥应从哪里获取。

查看智能体所依据的安装指南

其他 MCP 客户端

把它添加到你客户端的 mcpServers 配置中。

{
  "mcpServers": {
    "circlfi": {
      "type": "http",
      "url": "https://circlfi.com/mcp"
    }
  }
}

README

CirclFi MCP Server

[npm version] [MCP] [License: ISC]

The official Model Context Protocol (MCP) server for CirclFi — the institutional-grade equity valuation platform.

Give any AI assistant direct access to 13 valuation models across 5,900+ US stocks, recalculated daily from SEC EDGAR, FRED, and GDELT data. Intrinsic values, Quality of Company scores (32 fundamental signals), and Value Trap risk scores — as native AI tools.

Two ways to connect

Option A — Remote server (no install, works everywhere)

Point any MCP-compatible client at the hosted endpoint:

https://circlfi.com/mcp
  • Transport: Streamable HTTP (stateless), current MCP spec
  • Auth: none needed for the free tier. Premium subscribers send their Gumroad purchase email as a bearer token: Authorization: Bearer [email protected]
  • Works with ChatGPT connectors (developer mode), Claude custom connectors (web/desktop), agent frameworks, and any cloud client that can't run local processes.

Option B — Local stdio server (npm)

bash
npx -y circlfi-mcp

(Installing from GitHub also works: npx -y github:negm17111995/Circlfi-MCP)

Install in your client

Claude Desktop

Add to claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json, Windows: %APPDATA%\Claude\claude_desktop_config.json):

json
{  "mcpServers": {    "circlfi": {      "command": "npx",      "args": ["-y", "circlfi-mcp"],      "env": { "CIRCLFI_EMAIL": "" }    }  }}

Restart Claude Desktop and look for the tools icon. Premium subscribers: put your Gumroad purchase email in CIRCLFI_EMAIL.

Claude Code
bash
claude mcp add circlfi -- npx -y circlfi-mcp

Or with premium access:

bash
claude mcp add circlfi -e [email protected] -- npx -y circlfi-mcp
Cursor

One-click install →

Or add to ~/.cursor/mcp.json:

json
{  "mcpServers": {    "circlfi": { "command": "npx", "args": ["-y", "circlfi-mcp"] }  }}
VS Code (Copilot / MCP)
bash
code --add-mcp '{"name":"circlfi","command":"npx","args":["-y","circlfi-mcp"]}'
Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

json
{  "mcpServers": {    "circlfi": { "command": "npx", "args": ["-y", "circlfi-mcp"] }  }}
ChatGPT (connectors / developer mode)

ChatGPT connects to remote MCP servers. In Settings → Connectors → Advanced → Developer mode, add a custom connector with the URL:

https://circlfi.com/mcp
Any other MCP client
  • Local: command npx, args ["-y", "circlfi-mcp"], optional env CIRCLFI_EMAIL
  • Remote: Streamable HTTP at https://circlfi.com/mcp, optional header Authorization: Bearer <gumroad-email>

Available tools

get_stock_valuation

Complete intrinsic valuation profile for any US ticker: fair value, implied upside, and statistical confidence per model, plus the Quality of Company (QOC) score and Value Trap risk score.

Free tier — 3 models, no signup: Bayesian DCF (10,000-run Monte Carlo with jump-diffusion), Earnings Power Value (Greenwald), CUCE Ensemble (meta-model).

Premium — all 13 models: the above plus EROIC Spread, First Chicago, Markov DDM, ML-RIV, Dynamic NAV, PWERM, Regime Cross-Sectional, Sentiment SOTP, FTNN Topology, and RCMH-DCF. Full methodology: circlfi.com/methodology

get_market_screener (Premium)

Screen all 5,900+ US stocks in one call. Filter by minimum QOC, maximum Value Trap score, or industry; return JSON for analysis or CSV for download.

Example prompts

  • "What is the intrinsic value of AAPL according to CirclFi?"
  • "Screen for Software stocks with QOC above 8 and Value Trap below 10, output as CSV."
  • "Compare the Bayesian DCF valuations of MSFT and GOOGL and tell me which trades further below fair value."

How it works (security & architecture)

This repository is a thin client — no proprietary algorithms, no API keys. Tool calls are forwarded to the CirclFi API (https://circlfi.com/api/mcp), which verifies premium subscriptions with Gumroad in real time and returns model output. Free-tier requests need no account at all. The hosted remote endpoint (https://circlfi.com/mcp) exposes the identical tools directly over Streamable HTTP.

All output is quantitative model data for educational purposes — not financial advice (disclaimer).

Pricing

  • Free: 3 models per stock, unlimited tickers
  • Premium ($39/mo or $299/yr): all 13 models + market screener — subscribe via Gumroad, then use your purchase email as CIRCLFI_EMAIL (local) or bearer token (remote)

Support

Open a GitHub Issue for server problems. For valuation methodology or subscription questions, see circlfi.com.

来源:README.md,提交 a407d90

工具

0
SourceWeft 尚未读取此服务器的工具。

版本历史

1
  1. v1.1.0最新Sep 16, 2026

评分

已满足适用检查项的 85%

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 天内可连通必需25 分
  • 已声明认证方式必需10 分
  • 已列出工具必需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