Equity Due Diligence: evidence-first stock notes
Turns the free equity-intel-mcp server into a repeatable research note: insider Form 4, 13F holders, valuation, implied move. Data and caveats, no buy/sell calls.
Free
Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+22 moreClaude CodeCursorCodex CLIMuseDots
OpenClaw
Manus
Grok
Grok Bot+18 agentsEquity Due Diligence: evidence-first stock notes
Example session with this skill installed
Do evidence-first due diligence on AAPL with equity-intel-mcp.
- Read your context and instructions
- Compiled the equity due diligence
Layout of a real run (AAPL, 2026-10-15). The insider row is real tool output (SEC EDGAR data is public). Other rows use placeholders, because Yahoo Finance, Dataroma and Finnhub restrict republishing their content; your own run fills them in.
AAPL — evidence note (2026-10-15, data as returned by equity-intel-mcp)
One line: insiders only selling over 180 days; <one plain sentence on the other rows>.
| Dimension | Reading | Strength of evidence | Caveat |
|---|---|---|---|
| Insiders (Form 4, 180d) | $0 bought vs $181.3M sold, 27 filings | weak | selling only; no open-market buys to read |
| Superinvestors (13F) | N tracked holders; last quarter B buys / S sells | weak | 45-day lag; no copy edge in author's test |
| Valuation | fair value ~$F vs $P (±X %); health H/100 | weak | sector P/E may understate a premium franchise |
| Analysts | no data | — | FINNHUB_API_KEY not set |
| Options (1m implied move) | ±M % to <expiry>; put/call OI R | n/a (risk, not direction) | — |
| Tool blended score | S on a −100…+100 scale, confidence C %, K sources | summary only | not advice |
What would change this picture: an open-market purchase by a director or officer; the next 13F round (Q3 filings due 14 November) showing net adds; a guidance change that moves forward EPS enough to close or widen the valuation gap.
Gaps: analyst consensus (no Finnhub key).
Not financial advice. Data from public sources (SEC EDGAR, Dataroma, Yahoo Finance, Finnhub); may be delayed or wrong. Verify before acting.
Connects securely to your tools. The creator never sees your data.
About this skill
What it does. Gives your agent a fixed, evidence-first procedure for researching a US stock with the open-source equity-intel-mcp server (github.com/cstamigo-droid/equity-intel-mcp; 7 tools: SEC Form 4 insider activity, Dataroma superinvestor holdings, valuation, analyst consensus, options-implied move, quote, and a blended score). The agent collects every source in a set order, records which ones returned nothing, applies interpretation rules (for example: insider selling is weak evidence; 13F holdings are 45+ days old) and writes a one-page note with a "strength of evidence" column and a "what would change this picture" section.
What it is not. It is a data tool, not an advisor. The skill tells the agent never to give buy/sell calls, price targets or position sizes. Note: the free server's composite tool does print an action label (BUY ... AVOID) next to its score; the skill treats that as a mechanical weighting and instructs the agent not to repeat it.
It has no proven edge: in the author's own pre-registered test (13-Oct-2026), copying the best ideas of 24 superinvestors from original 13F filings over 51 quarters did not beat an equal-weight S&P 500 ETF by a statistically significant margin (+0.4-0.6 % per quarter, t = 0.9-1.3, 90 % interval includes zero), and 0 of 15 primary tests across five strategy families passed. Details are in reference/limits.md, inside the skill.
What you need. Python 3.10+, pip install "equity-intel-mcp>=0.1.3" (free, MIT), an EDGAR_IDENTITY contact string (your name and e-mail; the SEC asks for it, it is not a key) and, optionally, a free Finnhub key for analyst consensus. No brokerage account, no paid data. Your machine fetches the data directly; the skill sends nothing to the author. Several sources (Yahoo Finance, Finnhub free tier, Dataroma) are for personal use only; see limits.
Why a skill if the server is free? The server returns raw signals. The skill is the part that makes an agent use them carefully: order, gaps, caveats and a consistent format you can compare across tickers.
Not financial advice.
How to use
Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
1 install
Downloaded by developers to date
Free forever
No account required to browse
Trust & safety
Security scanned
Verified clean 4 days ago
- Free to download with an account
Requires
Python