About Full Disclosure
Public companies explain their results, plans and risks in long SEC filings that few people read in full. Full Disclosure reads them and reports what changed, in a few minutes’ reading.
How it works
- 1
Find the filings
For each company on the watchlist, the pipeline looks up its SEC identifier and lists its recent filings from EDGAR.
- 2
Keep the ones that matter
Every annual (10-K) and quarterly (10-Q) report qualifies. Current reports (8-K) qualify only for material events, such as earnings, acquisitions, executive changes or restatements.
- 3
Read them with Claude
The filing's text, plus any attached press release, goes to Claude, which returns a summary, the most important changes, the biggest risks and a sentiment score from −5 to +5, based only on the filing itself.
- 4
Publish
Each analysis is stored in Postgres and appears here. A scheduled job checks for new filings every day.
Built with
- Pipeline
- Python, Requests, Beautiful Soup, Pydantic, pytest
- Analysis
- Claude via the Anthropic API, with structured output
- Data
- Postgres on Supabase
- Website
- Next.js, React, TypeScript, Tailwind CSS, on Vercel
Limits
Summaries are generated by AI and can miss details or get them wrong. Sentiment describes the tone of a filing, not where a stock is headed. Always check the original filing, linked from every analysis. Nothing here is investment advice.