# Reference overview

Base URL `https://api.scispace.com/v1`. Read conventions once; it covers auth, pagination,
idempotency, versioning, and headers so the resource pages do not repeat them.

## Resources

| Resource | Use it to | Async? |
|---|---|---|
| searches | answer a research question over the corpus | yes |
| papers | look up paper metadata, full text, references | no |
| documents | ingest and manage your own PDFs | parse is async |
| chats | ask questions about named documents and papers | streaming |
| extractions | pull structured fields out of many PDFs | yes |
| paraphrases | rewrite text with tone and length control | no |
| ai-detections | assess whether text is AI-generated | no |
| topics | extract concepts and evidence for a query | yes |
| citations | format bibliography entries in 2,000+ styles | no |
| files | upload inputs, download exports | no |
| jobs | check any async resource with one call | no |
| webhook-endpoints | receive completion events | no |
| usage | see credit consumption | no |

## Machine-readable spec

The OpenAPI 3.1 document is the field-level truth and is generated from the implementation:
`https://docs.scispace.com/openapi.yaml`. Use it to generate a client for any language, and prefer it
over these pages when the two disagree — then file the discrepancy at `api-feedback@scispace.com`.

An `llms.txt` index and a `.md` variant of every page are published at the same origin for coding
agents.