Documentation
Two models behind one key. QED writes academic prose from your notes; Sic scores sentences for machine authorship. Use them over HTTP, or give them to an agent as tools.
- 1Quickstart5 minutesCreate a key, score a paragraph, draft a passage, check it against your notes.
- 2API referenceReferenceEvery endpoint, field and error: write, detect, compare, disclosure, models, usage.
- 3MCP serverAgentsConnect Claude, Cursor or another agent, and what each tool does.
- 4Models and limitsReferenceModel strings, sizes, speed, allowances and rate limits, and what we keep.
- 5Responsible usePolicyDisclosure, checking, and what a detector score may and may not be used for.
At a glance
| Base URL | https://api.etal-labs.com/v1 |
| Authentication | Authorization: Bearer etal_…, with a key from the console |
| Format | JSON in, JSON out. Text is Markdown: math as $…$, citations as [@key]. |
| Models | qed (QED 1, qed-2026b) and sic (Sic 1, sic-2026a) |
| MCP | https://mcp.etal-labs.com/mcp, Streamable HTTP, sign-in or key |
| Stored | Counts, sizes and timings of your requests. Not the text you send, and not the text returned. |