Ask Claude about any public company and it answers from the actual SEC filing — with a source you can check.
Benchmarked 10–20× cheaper than agentic web search — and more correct. Free tier included. No credit card required.
A real exchange, captured from the live server — not a mock-up. One free-tier tool call, every figure traceable to the filing it came from.
"Give me a brief on Nvidia — what do their latest filings say about how the business is doing?"
Fiscal year ended 2026-01-31. Nasdaq. Semiconductors & Related Devices.
Source: NVIDIA FY2026 10-K, filed 2026-02-25 · market data prior close 2026-08-07 · sample captured 2026-08-08
That was company_brief — a Free-tier tool. The assistant is told to cite the source filing on every answer, so you can always click through and check.
Each card is a recipe — a prompt that walks your assistant through a real analysis, step by step. All free to read. On Analyst they also arrive in your client as slash commands.
One filing, one ticker, fast. Get through a 10-K or an earnings release without reading all of it.
The 10-K is 180 pages and the meeting is in 20 minutes.
10-K Speedrun →Item 1A changed. The company won't tell you where.
Risk-Factor YoY Diff →You know the stock reaction. You don't know why.
Earnings Release Decoder →A watchlist on a cadence. Many tickers, run weekly, triaged down to what moved.
Forty filings hit your watchlist this week. Three matter.
8-K Event Triager →You read Form 4 headlines, not Form 4s.
Insider Activity Interpreter →Long-form, framework-driven analysis across several filings and peers. The deepest thing the assistant will do unaided.
Your compounder checklist ignores SBC and per-share economics.
F.O.R.G.E. Compounder Analysis →You have a bad feeling and no falsifiable bear case.
Short-Seller Forensics →Browse the full recipe library → — every recipe lists the exact tools it calls, so you can check tier requirements before running one.
Trust first. Every data point is parsed from SEC EDGAR (the SEC's public filing database) — XBRL read directly, never transcribed — and every answer carries a citation back to the source document.
Then cost. Without MCP, Claude fetches a 300-page filing and burns 100,000+ tokens parsing legal prose. With analyze_filing, it calls one tool and gets structured data back in 200ms. Measured, not claimed: in our published benchmark, the same SEC research tasks used 10–21× fewer tokens through MCP than agentic web search — at higher judged correctness, with per-run records public.
Same verb ladder as /pricing. Every tool lives in exactly one tier, grouped by the six categories the live server uses — what you see here is what tools/list returns.
100 API calls/day, resets daily·no credit card
Find any company, browse its filings, get a one-call snapshot. The proof transcript above ran entirely on this tier.
search_companiessearch_entitiescompany_briefcompany_filingsmanager_holdingsedgar_recipes10,000 API calls/day·14-day trial, card on file
Everything a tracked list needs: full financial statements, insider trades, institutional ownership, real-time material events, full-text and disclosure search, and the fund + adviser reference surface.
search_filingsdisclosure_searchedgar_notessearch_fundssearch_advisersfund_profileadviser_profilelive_filingsfiling_sectionfinancial_snapshotfinancial_trendsfinancial_statementscompare_companiesmaterial_eventsinsider_activityinstitutional_ownership50,000 API calls/day·7-day trial, card on file
Depth, not volume. Whole-filing AI analysis, plus aggregation bundles that answer across a position, a peer set, or an entire portfolio in one call — and the slash commands that chain them into finished write-ups.
analyze_filingfiling_detailsposition_intelpeer_factsportfolio_eventsperson_intelaccount_dossier/edgar:position_thesis/edgar:peer_comparison/edgar:portfolio_pulse/edgar:earnings_postmortem/edgar:filing_red_flagsPick your client. Copy the config. Start asking questions.
Already signed in? See your tier-grouped tools on the Plugin Hub.
Not using an AI agent? Use edgar.tools directly in the browser — no setup required.
The short version: edgar.tools is the self-serve, SEC-native option — a first-party MCP server with a full research toolkit, published prices, and a free tier. The head-to-head pages carry the detail.
https://app.edgar.tools/mcp to your config — OAuth handles authentication automatically via Google sign-in. For ChatGPT, Gemini, and custom agents: sign up free at app.edgar.tools, get an MCP token from Settings > API Keys, and pass it as a Bearer token in your config headers.
The full EDGAR universe — structured, cited, 10–20× cheaper than web search. 14-day trial.