Circlfi-MCP
# CirclFi MCP Server
[](https://www.npmjs.com/package/circlfi-mcp)
[](https://modelcontextprotocol.io/)
[](./LICENSE)
The official [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server for **[CirclFi](https://circlfi.com)** — 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 you@example.com`
- 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
<details>
<summary><strong>Claude Desktop</strong></summary>
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`.
</details>
<details>
<summary><strong>Claude Code</strong></summary>
```bash
claude mcp add circlfi -- npx -y circlfi-mcp
```
Or with premium access:
```bash
claude mcp add circlfi -e CIRCLFI_EMAIL=you@example.com -- npx -y circlfi-mcp
```
</details>
<details>
<summary><strong>Cursor</strong></summary>
[**One-click install →**](cursor://anysphere.cursor-deeplink/mcp/install?name=circlfi&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsImNpcmNsZmktbWNwIl19)
Or add to `~/.cursor/mcp.json`:
```json
{
"mcpServers": {
"circlfi": { "command": "npx", "args": ["-y", "circlfi-mcp"] }
}
}
```
</details>
<details>
<summary><strong>VS Code (Copilot / MCP)</strong></summary>
```bash
code --add-mcp '{"name":"circlfi","command":"npx","args":["-y","circlfi-mcp"]}'
```
</details>
<details>
<summary><strong>Windsurf</strong></summary>
Add to `~/.codeium/windsurf/mcp_config.json`:
```json
{
"mcpServers": {
"circlfi": { "command": "npx", "args": ["-y", "circlfi-mcp"] }
}
}
```
</details>
<details>
<summary><strong>ChatGPT (connectors / developer mode)</strong></summary>
ChatGPT connects to remote MCP servers. In **Settings → Connectors → Advanced → Developer mode**, add a custom connector with the URL:
```
https://circlfi.com/mcp
```
</details>
<details>
<summary><strong>Any other MCP client</strong></summary>
- **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>`
</details>
## 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](https://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](https://circlfi.com/disclaimer/)).
## Pricing
- **Free**: 3 models per stock, unlimited tickers
- **Premium** ($39/mo or $299/yr): all 13 models + market screener — [subscribe via Gumroad](https://circlfi.gumroad.com/l/dwaxoj), then use your purchase email as `CIRCLFI_EMAIL` (local) or bearer token (remote)
## Support
Open a [GitHub Issue](https://github.com/negm17111995/Circlfi-MCP/issues) for server problems. For valuation methodology or subscription questions, see [circlfi.com](https://circlfi.com).
TDQS
Scored across 2 tools
The two tools serve completely distinct purposes: one focuses on intrinsic valuation of a single stock, while the other screens the entire market. There is no overlap or potential for confusion.
Both tools use a consistent 'get_' prefix followed by a descriptive noun phrase (get_stock_valuation, get_market_screener), following a clear verb_noun pattern.
With only 2 tools, the server feels underpopulated for its stated domain of stock analysis and screening. While the tools are individually valuable, the scope suggests more tools would be reasonable (e.g., price data, fundamentals).
The server only covers valuation and market screening, leaving out essential stock analysis operations like price history, financial statements, or news. This is a significant gap for a stock analysis server.