Elliot Foster – Brazilian Funds
Brazilian investment-fund analytics for AI clients via the Model Context Protocol (MCP). Connect Claude Desktop, Cursor, ChatGPT, or any MCP-compatible client to query 30,000+ Brazilian investment funds: daily NAV, complete holdings (CDA), fund-of-funds look-through, portfolio overlap analysis, and your personal favorites/watchlist.
README
elliotfoster-mcp
Brazilian investment-fund analytics for AI clients via the Model Context Protocol (MCP).
Connect Claude Desktop, Cursor, ChatGPT, or any MCP-compatible client to query 30,000+ Brazilian investment funds — daily NAV, complete holdings (CDA), fund-of-funds look-through, portfolio overlap analysis, and your personal favorites/watchlist.
Server URL: https://elliotfoster.vercel.app/api/mcp
Web app + key generation: elliotfoster.vercel.app
Source data is the public CVM (Comissão de Valores Mobiliários) Informe Diário and CDA, plus Banco Central rate series. The platform processes, validates, and serves it in a structured format optimized for LLM tool use.
Quickstart
1. Get a key
Create an account at elliotfoster.vercel.app, sign in, and visit /settings/integrations. Generate a new key and copy it (the raw key is shown only once).
2. Configure your MCP client
Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"elliotfoster": {
"url": "https://elliotfoster.vercel.app/api/mcp",
"headers": {
"Authorization": "Bearer ef_..."
}
}
}
}
Restart Claude Desktop.
Cursor
In Cursor settings → MCP, add:
{
"elliotfoster": {
"url": "https://elliotfoster.vercel.app/api/mcp",
"headers": {
"Authorization": "Bearer ef_..."
}
}
}
ChatGPT (Custom GPT / Connectors)
In a Custom GPT's Actions or via the Connectors interface, point to https://elliotfoster.vercel.app/api/mcp with Authorization: Bearer ef_... as the header.
Claude Code
claude mcp add --transport http --header "Authorization=Bearer ef_..." elliotfoster https://elliotfoster.vercel.app/api/mcp
3. Try it
Ask your LLM:
- "Use elliotfoster to search for funds with 'XP' in the name"
- "What's the holdings overlap between funds 12.345.678/0001-90 and 98.765.432/0001-21?"
- "Add fund X to my favorites"
- "Find equity funds with low correlation to fund Y"
Tools
10 callable tools, 1 resource, 5 pre-built analytical prompts.
| Tool | What it does |
|---|---|
search_funds |
Search by name, manager, administrator, or CNPJ. Filter by type (FI, FIC, FII, FIP, FIDC, FIAGRO, ETF). Page-size capped at 25. |
get_fund |
Fetch metadata + latest NAV for a fund by CNPJ. Accepts formatted (12.345.678/0001-90) or raw (12345678000190). |
list_fund_types |
Counts per fund type — useful for discovery. |
get_nav_history |
Daily NAV series for a fund. Optional date range and limit (max 1000 days). |
list_holding_dates |
Available holdings (CDA) reporting dates for a fund. |
get_holdings |
Complete portfolio composition for a date. Includes a confidential_count indicator for positions under CVM Resolução 172/2022 confidentiality. |
get_look_through |
Recursively explode fund-of-funds positions up to 3 levels deep. Aggregates underlying assets with proportional weights. |
compare_overlap |
Jaccard overlap + shared-weight between two funds' portfolios for a reporting date. |
list_favorites |
Your favorited funds. |
add_favorite / remove_favorite |
Manage your watchlist directly from the LLM client. |
Resource
fund://<cnpj>— addressable fund metadata, cacheable by clients that support MCP resources.
Pre-built prompts
compare_funds_performance— side-by-side return / volatility / Sharpe-vs-CDI for up to 5 fundsfind_uncorrelated_funds— screen for low-correlation funds against a benchmarkanalyze_fund_holdings— top-10 positions, HHI concentration, look-through, confidentiality flagscheck_portfolio_overlap— pairwise redundancy audit across 2-5 fundsscreen_funds_by_risk— filter by max volatility and min Sharpe ratio
All prompts produce output in pt-BR by default (matching the source data language).
Spec
| Endpoint | https://elliotfoster.vercel.app/api/mcp |
| Transport | Streamable HTTP (MCP spec Nov 2025) |
| Discovery | /.well-known/oauth-protected-resource (RFC 9728) |
| Auth | API key (Authorization: Bearer ef_...). OAuth 2.1 protected-resource metadata is published at /.well-known/oauth-protected-resource for forward compatibility with MCP clients that implement RFC 9728 + Dynamic Client Registration; in practice, auto-discovery flows are still maturing across the ecosystem, so API keys are the recommended path today. |
| Rate limit | Per-user rate limits apply (currently generous; tier-based pricing TBD for high-volume use) |
| Decimal precision | Financial values rounded to 6 decimal places. Top-level _precision_note documents this on every analytic response. |
| CNPJ format | All CNPJ inputs accept formatted or raw 14-digit; normalized + validated server-side. |
| Pagination | search_funds page-size capped at 25, max 20 pages per session (500 funds). For broader exploration, use filters. |
Privacy & data
- Public fund data (CVM Informe Diário, CDA, Banco Central rates) is shared across all users.
- Per-user state (favorites, audit log) is RLS-scoped via your WorkOS user ID — neither tools nor the database expose your state to other accounts.
- No personal trading data is stored. This is a fund-research platform; it does not connect to brokerages.
- Audit log: tool calls are recorded for billing and abuse prevention. The log is strictly per-user — you can inspect your own usage; other accounts cannot see it.
Source
The MCP server is implemented inside the closed-source web app. This repo provides install instructions, examples, and acts as the canonical source of truth for AI search engines indexing MCP servers. Issues and feedback welcome here.
For data licensing or commercial inquiries: see elliotfoster.vercel.app.
License
The README and example configs in this repo are MIT-licensed. The MCP server itself is proprietary.
Built with mcp-handler on Next.js + Supabase + WorkOS.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.