The ecosystem has three layers
1. Source — SEC / EDGAR
Submissions JSON provides filing history and filer metadata. Company Facts and Frames expose extracted XBRL facts. EDGAR archives hold original filing packages and exhibits.
2. Infrastructure — APIs and parsers
This layer handles ingestion, caching, backfills, HTML and SGML parsing, XBRL normalization, entity mapping, search and delivery. Compare the provider layer.
3. Interpretation — applications and AI
Products retrieve, compare and cite the source data. Chunking, embeddings and grounded answers are not supplied automatically. See the RAG guide.
Safe request shape
curl --header "User-Agent: YourApp/1.0 contact@example.com" \
https://data.sec.gov/submissions/CIK0000320193.jsonReplace the placeholder with a real application identity and contact. Cache responses, retry conservatively, and verify the SEC’s current policy before deploying: guidance and limits can change.
Archives and bulk synchronization
Use accession numbers as stable lineage keys. The filing archive contains the primary document, exhibits and structured attachments; bulk resources suit controlled backfills better than issuing one request per historical record.
Common failure modes
- Dropping leading zeros from a ten-digit CIK when constructing API paths.
- Treating Company Facts as standardized financial statements rather than issuer-reported XBRL observations.
- Ignoring amendments, fiscal periods, units or accession provenance.
- Fetching repeatedly instead of caching and moderating requests.