consultance.ai
Finance and data

Claude Equity Research Vault — Goodbye FactSet

A sell-side-grade equity research desk that runs inside Claude on live market data and real SEC filings, for the price of a Claude seat instead of a $12,000 to $25,000 FactSet terminal. Pull the comp set, build the DCF, read the 10-K for what management buried, run bull/base/bear scenarios, and write the research note, every figure tied to a pulled source. Built around an actual data connection (Alpha Vantage and SEC EDGAR, both free, both verified working), not screenshots.

$12,000 to $25,000 a seat for FactSet, or the same research for a Claude seat.

Analysts pay $12K to $25K a seat for FactSet, S&P CapIQ, or PitchBook. This is a sell-side-grade research desk inside Claude on live market data and real SEC filings — comp set, DCF, 10-K read, bull/base/bear, and the research note, every figure tied to a pulled source, built around a verified free data connection.

Built on open source
OpenBB + FinRobot (open-source data + agent stack, full install guide included)

The code is public and free. The setup prompt below installs and wires it for you.

Step 1

Paste this setup prompt — Claude installs it for you

Easy mode · paste this into Claude

Claude installs it for you, step by step.

Never used Claude before? It is free to start. Open it in a new tab, copy the prompt, paste it in. It asks one question, then walks you through everything.

  1. Step 1
    Open claude.ai ↗

    Sign up free. No card. Takes 30 seconds.

  2. Step 2

    One click. Lands on your clipboard.

  3. Step 3
    Paste + send

    Claude asks what you need + guides you the rest of the way.

Open claude.ai ↗
Tune the prompt for your level (optional)
Preview the prompt (you do not need to read it)
I want to INSTALL the real data stack so Claude runs a sell-side-grade equity research desk on live market data, instead of pasting numbers in by hand. Walk me through the full install, step by step, and do not skip the technical parts. I would rather spend 30 minutes installing the real thing than get a watered down version. Treat me like an equity analyst or FP&A lead who has never used a Terminal, and define every command before I run it.

Three paths. Recommend Path 1 (full delivery). Fall back only if I get stuck.

## Path 1 (RECOMMENDED, full delivery) — OpenBB live data inside Claude Desktop via MCP

This puts live market-data tools directly inside Claude Desktop. Once installed, Claude pulls quotes, fundamentals, and comps itself. This is the FactSet killer, the real deliverable, not a shortcut.

Walk me through ONE step at a time, waiting for me to confirm each:

1. **Install a Python toolchain.** In Terminal run `python3 --version` (need 3.10 or 3.11). If missing, install from python.org. Then install `uv` (a fast Python runner): `curl -LsSf https://astral.sh/uv/install.sh | sh` (macOS/Linux) or the documented Windows command. Define: Terminal = the command window (macOS: Spotlight → "Terminal"; Windows: "PowerShell").
2. **Install the OpenBB MCP server:** `pip install openbb-mcp-server` (this pulls the OpenBB Platform too). Explain: this is the bridge that lets Claude call live financial data.
3. **Add free data-provider keys** in OpenBB (`~/.openbb_platform/user_settings.json`) — tell me which free providers (e.g. FMP, Polygon free tier) cover quotes + fundamentals and how to paste each key.
4. **Wire it into Claude Desktop.** Settings → Developer → Edit Config (`claude_desktop_config.json`; macOS `~/Library/Application Support/Claude/`). Add this block, save, fully restart Claude Desktop:
   ```json
   {
     "mcpServers": {
       "openbb-mcp": {
         "command": "uvx",
         "args": ["--from", "openbb-mcp-server", "--with", "openbb", "openbb-mcp", "--transport", "stdio"]
       }
     }
   }
   ```
5. **Confirm:** after restart, Claude Desktop shows an "openbb-mcp" tool source. Have Claude pull one live quote to prove the connection before any analysis.
6. Pin **Opus 4.8**, create a private Project "Equity Research", run prompt 01, pick data source (A) LIVE, run 02 to 08.

Reference: OpenBB MCP docs `docs.openbb.co/odp/python/extensions/interface/openbb-mcp`; Claude Desktop local MCP guide `support.claude.com/en/articles/10949351`.

## Path 2 (ADVANCED) — FinRobot autonomous research pipeline

FinRobot is a separate open-source Python agent platform that runs the research pipeline itself (DCF, projections, peer comp, report gen). It calls an LLM API on its own; it is NOT an MCP tool inside Claude. Offer only if I want the heavier automated pipeline:

1. Requires **Python 3.10 or 3.11** (NOT 3.12). Confirm my version.
2. `git clone https://github.com/AI4Finance-Foundation/FinRobot.git`, `cd FinRobot`, `pip install -U finrobot` (or `pip install -e .`).
3. Config: rename `OAI_CONFIG_LIST_sample` → `OAI_CONFIG_LIST` (add LLM API key); rename `config_api_keys_sample` → `config_api_keys` (add Finnhub + FinancialModelingPrep + SEC-API keys; partial works with a subset).
4. Honest note: power-user pipeline, separate from Claude. Most readers stop at Path 1.

## Path 3 (QUICK START, 5 minutes, zero install) — Alpha Vantage + EDGAR by paste

To run the prompts TODAY before installing:
1. Prove the feed: open `https://www.alphavantage.co/query?function=GLOBAL_QUOTE&symbol=IBM&apikey=demo`; confirm a live quote.
2. Free Alpha Vantage key at `alphavantage.co/support/#api-key` (free tier 25 calls/day, 5/min). Add the Alpha Vantage MCP app in Claude, OR paste pulls when a prompt asks.
3. SEC EDGAR (free, no key): 10-K/10-Q/8-K at `data.sec.gov/api/xbrl/...` with a User-Agent header; full-text search at `efts.sec.gov`.
4. Pin Opus 4.8, private Project, run prompt 01, pick the live or paste source.

## Path B — my firm's licensed data
If my firm licenses FactSet or S&P Capital IQ, enable the connector via **Claude for Financial Services** (Enterprise admin step), then run the same prompts with data source (D). Data stays under firm controls.

## Rules for walking me through this
- One step at a time. Tell me exactly what to type or click. Define every term once: Terminal, Python, pip, uv, uvx, MCP server, MCP config, API key, CIK, 10-K, comp set, DCF, WACC, terminal growth, `<review_gate>`, `{{TOKEN}}`.
- Do NOT water it down to avoid the Terminal. Delivery beats ease. If a step fails, debug with me (Python version, valid config JSON, fully quit/reopen Claude Desktop). Drop to Path 3 only if I explicitly want no-install.
- Never tell me a connection is "not possible." Give me the next diagnostic step.
- A named human analyst signs every research note. Claude builds and ties sources; a person signs.

First message: ask me "Path 1 (install OpenBB live data into Claude, ~30 min, the real desk) or Path 3 (5-minute no-install quick start)? And what is your operating system?" Then start step 1.

When the connection returns live data AND prompts 01 to 03 run on a real ticker, switch into "desk mode" and help me lock my coverage list and note format.
Step 2

Step 1 installed it. Now run these 8 prompts on your own data.

the vault

The 8 prompts

Tap a prompt to jump to it. Hit copy. Replace the tokens. Paste into Claude Opus 4.7.

where it breaks

Before you connect live data

  • • Run dummy data first. Real client data is not a test bed.
  • • API keys never go in a public repo. Use env vars and a secrets manager.
  • • Add logging, access control, monitoring, and a rollback path before launch.
  • • Read the license. Forking a repo without checking is how lawsuits start.
license note

Credit the original author

Prompt set and build authored by consultance.ai. FactSet, S&P Capital IQ, and PitchBook are referenced as the incumbent being replaced, no affiliation implied. Alpha Vantage and SEC EDGAR are free third-party data sources used under their own terms. Your analysis stays in your own Claude tenant; we never see it. This is not investment advice; a named human analyst signs every research note before it is published.

the newsletter

AI news worth opening.

The AI tools, launches, and shifts that actually matter, in plain English. New library drops the moment they land.

100% freeNo paywall, everUnsubscribe anytime

Read this far? You want terminal-grade research without the $25K seat. Let us build the desk — source-tied and signed before anything publishes.