pequod-mcp

pequod-mcp

MCP server providing complete US ETF holdings from SEC EDGAR N-PORT filings and identifier crosswalk tools (ticker, CUSIP, FIGI) for AI agents.

Category
Visit Server

README

pequod-mcp — "the Gam"

<!-- mcp-name: io.github.ABWerks/pequod-mcp -->

SEC-primary-source ETF holdings + identifier crosswalk for agents.

An MCP (Model Context Protocol) server that gives AI agents the complete holdings of any US-registered ETF straight from SEC EDGAR N-PORT filings — not a top-10 scrape — plus a set of identifier-crosswalk tools (ticker, CUSIP/SEDOL, FIGI → resolved symbol) backed by OpenFIGI and SEC reference data. Runs locally over stdio; no vendor market-data subscriptions required.

Tools

Tool What it does Sources
lookup_cik Ticker → SEC CIK + entity metadata (name, SIC, exchanges, fiscal year end) SEC company_tickers.json + submissions API
resolve_ticker Raw ticker + optional exchange code → Yahoo-convention symbol, verified Exchange-suffix map (pure string mapping) + SEC + OpenFIGI
resolve_cusip CUSIP (or SEDOL) → symbol, with cross-source verification OpenFIGI (one batched request) + SEC name/ticker agreement
resolve_figi Bloomberg FIGI → symbol, with cross-source verification OpenFIGI + SEC name/ticker agreement
get_etf_holdings Any US ETF → complete holdings list from its latest monthly N-PORT filing (name, ticker, ISIN, FIGI, LEI, shares, USD value, % of net assets, country, currency) SEC EDGAR N-PORT XML

All tools are read-only.

How verification works

Crosswalk results are verified without any market-data vendor:

  1. SEC registry check — if the mapped ticker is a US SEC registrant, the entity title in company_tickers.json must agree with the OpenFIGI security name. A gross name mismatch vetoes the mapping (verification: "name_veto") rather than returning a plausible-looking wrong symbol.
  2. OpenFIGI mapping — non-US symbols (outside SEC's registry) verify on the OpenFIGI mapping itself, labeled verification: "openfigi" so you can tell the confidence tiers apart.

Yahoo Finance symbology (.SW, .T, .L, … suffixes) is applied as a pure string convention — the default path makes no network calls to Yahoo.

Install

# with uv
uv pip install pequod-mcp

# or with pip
pip install pequod-mcp

Run it directly (stdio):

pequod-mcp

Optional: OpenFIGI API key

The server works without a key (OpenFIGI anonymous rate limits apply). For heavier use, get a free key at https://www.openfigi.com/api and set:

export API_OpenFIGI="your-key-here"

Optional: Yahoo enrichment extra

An opt-in extra adds a live Yahoo quote block to crosswalk results. It is off by default and never used otherwise; enabling it means you accept the Yahoo Finance terms of use yourself.

pip install 'pequod-mcp[yahoo]'
export PEQUOD_YAHOO_VERIFY=1

Configure in Claude

Claude Desktop (claude_desktop_config.json) or any MCP client with a stdio transport:

{
  "mcpServers": {
    "pequod": {
      "command": "pequod-mcp",
      "env": {
        "API_OpenFIGI": "your-key-here"
      }
    }
  }
}

Claude Code:

claude mcp add pequod -e API_OpenFIGI=your-key-here -- pequod-mcp

Agent workflow examples

1. "What does SCHD actually hold?"

Agent: lookup_cik("SCHD")
  → { cik: "0000884394", entity_name: "SCHWAB STRATEGIC TRUST", found: true }
Agent: get_etf_holdings("SCHD", cik="0000884394")
  → { fund_name, report_date, source_url (the actual EDGAR filing),
      total_holdings: ~100, holdings: [ { name, ticker, isin, figi,
      balance, value_usd, pct_val, ... } ] }

The agent gets every position with weights, plus the EDGAR URL for citation — auditable back to the primary source.

2. "I have a CUSIP from a 13F — what stock is this?"

Agent: resolve_cusip("594918104")
  → { yahoo_symbol: "MSFT", verified: true, verification: "openfigi+sec",
      company_name: "MICROSOFT CORP", cik: "0000789019",
      openfigi: { figi: "BBG000BPH459", ... } }

If OpenFIGI's mapping disagrees with SEC's registry on who the ticker belongs to, the tool refuses (verification: "name_veto") instead of handing back a wrong symbol.

3. "Compare two dividend ETFs' overlap."

Agent: get_etf_holdings("SCHD")   → complete holdings A
Agent: get_etf_holdings("VYM")    → complete holdings B
Agent: joins on isin (or ticker), computes overlap weight, reports the
       shared names — all from two tool calls, both citing SEC filings.

Data provenance

  • All holdings data comes from SEC EDGAR primary sources (Form N-PORT filings, the submissions API, and company_tickers.json). Identifier crosswalks come from the OpenFIGI API (FIGI is an open OMG standard).
  • No CUSIPs in output. Tool outputs carry ticker, ISIN, FIGI, and LEI; CUSIP fields present in the underlying filings are deliberately stripped.
  • SEC fair access respected: a declared User-Agent on every request, a throttle well under EDGAR's published rate limit, and an on-disk cache (.cache/ next to the module, override with PEQUOD_CACHE_DIR) for company_tickers.json (24 h) and N-PORT documents (30 days — filed documents are immutable).
  • Fail-loud policy: if EDGAR is unreachable or a filing cannot be parsed, tools return an error. There is no degraded or partial fallback source — you either get filing-backed data or a clear failure.
  • N-PORT filings are published on a lag (holdings are typically ~1–2 months behind). The report_date and source_url fields tell you exactly which period you are looking at.
  • Not investment advice. This server relays public regulatory filings and open identifier mappings; verify independently before making investment decisions.

Environment variables

Variable Default Purpose
API_OpenFIGI (unset) OpenFIGI API key (higher rate limits)
PEQUOD_CACHE_DIR .cache/ next to the module On-disk cache location
PEQUOD_YAHOO_VERIFY (off) Opt-in Yahoo enrichment (needs the [yahoo] extra)
LOG_LEVEL INFO Server log verbosity

Development

pip install -e '.[dev]'
pytest            # fully offline — fixture filings, no live network

License

Apache-2.0 — see LICENSE. Copyright 2026 AB Evergreen Analytics LLC.

Recommended Servers

playwright-mcp

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.

Official
Featured
TypeScript
Magic Component Platform (MCP)

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.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

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.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

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.

Official
Featured
TypeScript
Kagi MCP Server

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.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

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.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured