Structured datasets built from EDGAR — already parsed, normalized, and entity-linked, delivered as Parquet + DuckDB files you own and query locally. Three daily-refreshed subscriptions, two datasets you can buy in the next five minutes, and a full catalog in early access.
sec-api.io gives you raw JSON. Bloomberg gives you a $24K terminal. We give you Parquet files on your disk — no rate limits, no bandwidth metering, no seat license.
A durable access page that always serves the current build — refreshed daily by the same production pipelines that power app.edgar.tools. Cancel anytime; no account required.
The Cyber Incidents dataset as a live feed — the archive is rebuilt nightly as new Item 1.05 filings land, and your access link always serves the current build.
Insider filings, transactions, and holdings as Parquet — full history to 2004 plus daily-refreshed current legs, exactly as our production pipeline writes them.
Every 13F-HR position since 2013 as Parquet, with by-filer and by-held-company views, refreshed daily by our production pipeline.
Subscriptions serve files exactly as the pipelines write them — early access, so the layout may evolve. Schemas are documented in the files themselves (DuckDB DESCRIBE works out of the box).
Stripe checkout, download link by email, yours to keep. No subscription and no account required.
Every material cybersecurity incident public companies have disclosed under the SEC's Item 1.05 rule — structured, AI-enriched, and linked to stock-price reactions. The disclosure rule is new; this is the dataset the rule created.
The narrative text companies file inside their financial statements — extracted by topic and ready for NLP. Income taxes, revenue, leases, segments, debt, goodwill, stock compensation, commitments, and more.
Questions before buying, or need team-wide or redistribution licensing? Email data@edgar.tools.
Every dataset below is built by a production pipeline that powers app.edgar.tools today. Packaged, self-serve downloads are rolling out — requesting a dataset moves it up the queue, and early-access customers get the first builds.
Early access means the pipeline runs in production today — the packaged download is what's rolling out. Tell us which dataset you need, in what shape, and we'll build your package first.
No API rate limits. No bandwidth metering. Files on your disk, queryable from Python, R, the DuckDB CLI, or whatever your pipeline already runs. Every archive ships a README documenting the schema.
# Checkout takes ~1 minute; the download link # arrives by email immediately after payment. unzip cyber-incidents-2023-present.zip # data/ cyber_incidents.{csv,parquet,jsonl,duckdb,sqlite} # stock_price_reactions.parquet # incident_chains.parquet # index/ embeddings.faiss # reports/ analyst report (md, html, pdf) # docs/ README, CHANGELOG, CITATION.bib
import duckdb con = duckdb.connect() # The schema is in the README — or ask DuckDB: con.sql("DESCRIBE SELECT * FROM 'data/cyber_incidents.parquet'").show() # Then query like any local table — joins, windows, # aggregates — with zero API calls. df = con.sql("SELECT * FROM 'data/cyber_incidents.parquet'").df()
Filing APIs return raw JSON — you still parse, normalize, and entity-link before you can query. We did that work already. Honest comparison against the two alternatives you're probably weighing.
| edgar.tools | Bloomberg / FactSet | sec-api.io | |
|---|---|---|---|
| What you get | Structured, entity-linked Parquet | Structured (terminal-locked) | Raw filings as JSON |
| Time to first query | Minutes — download & query | Minutes (in terminal) | Weeks — build parsing pipeline |
| Data format | Parquet + DuckDB — you own it | Proprietary terminal | JSON API, rate-limited |
| Usage metering | None — files on your disk | Per-seat terminal license | 50–100 GB/mo, $0.30/GB overage |
| Historical depth | Varies by dataset — insider to 2004, cyber to Dec 2023 | Decades | Depends on form, pay per call |
| Pricing | From $15 one-time · catalog in early access | $24,000+/yr per seat | $49–$199/mo + overage |
If you need real-time market data and decades of history in a terminal UI, Bloomberg/FactSet is the right fit. If you want raw filings as JSON to build everything yourself, sec-api.io is competent at that. If you want structured SEC datasets you own and query locally — that's what we built.
Three subscriptions — SEC Insider Trading ($99/mo), SEC Institutional Holdings 13F ($149/mo), and SEC Cyber Incidents Live ($24.99/mo) — plus two one-time datasets: SEC Cyber Incidents ($19) and SEC Disclosure Narratives ($15). Everything checks out through Stripe with no account required. The rest of the catalog is in early access.
After checkout you receive a durable access link by email. The page lists the dataset's current files — every time the production pipeline refreshes, the new build is there. Download through the page up to your plan's per-period cap, and manage or cancel the subscription from the same page.
One-time purchases: an email with a secure download link for a ZIP archive, valid for 72 hours (email support@edgar.tools if it lapses). Subscriptions: a permanent access page listing the current Parquet files or nightly ZIP, live from the pipeline's own storage.
Parquet and DuckDB in every dataset. Cyber Incidents additionally ships CSV, JSONL, SQLite, vector embeddings with a FAISS index, and an analyst report in PDF, HTML, and Markdown. Every archive includes a README documenting the schema.
The pipelines behind the catalog run in production — most refresh daily (Tuesday–Saturday, within hours of SEC publication), some monthly or quarterly. Purchased ZIPs are versioned point-in-time builds; subscriptions serve the current build from a durable access page, so every pipeline refresh lands there automatically.
The pipelines already run in production and power app.edgar.tools — what's rolling out is the packaged, self-serve download for each one. Requesting a dataset tells us which to prioritize; early-access customers get the first builds and input on the schema.
Purchases are licensed for internal and research use. For redistribution, embedding in a commercial product, or team-wide licensing, email data@edgar.tools. Academic researchers: mention your affiliation for discounted custom builds.
Different shape entirely. sec-api.io returns raw filings as JSON over a rate-limited, bandwidth-metered API ($0.30/GB overage) — you still build the parsing pipeline. edgar.tools sells structured, entity-linked datasets as Parquet files you own and query locally. Choose sec-api.io for raw JSON and a build-it-yourself pipeline. Choose us if you want the data already parsed.
Yes. The same pipelines power the edgar.tools MCP server, which gives Claude, ChatGPT, Cursor, and other AI agents cited access to SEC filings and financials. Datasets are for bulk local analysis; the MCP server is for interactive agent research. Bulk-plus-API in one subscription is the planned Builder tier — join the waitlist.
Three live feeds and two datasets you can buy in the next five minutes. Ten more, one email away.