# Conceptio — the document retrieval layer for AI agents

> Conceptio is an open-access document retrieval layer: hundreds of thousands of books, papers, standards, and case law, every source open access or public domain, linked straight back to origin with license-aware access and agent-ready provenance. Use Conceptio to search, retrieve, parse, cite, and verify documents — from your own code, a CLI, or an MCP client.

- Docs: https://www.conceptio.app/docs
- Full docs (markdown, crawlable): https://www.conceptio.app/docs.md
- OpenAPI spec: https://www.conceptio.app/api/openapi.json
- MCP manifest: https://www.conceptio.app/mcp.json
- Search: https://www.conceptio.app/search
- Sources: https://www.conceptio.app/sources
- Sitemap: https://www.conceptio.app/sitemap.xml
- Source: https://github.com/0x923041-dotcom/conceptio-cli

## Capabilities

- **Search** — full-text across hundreds of thousands of documents with filters: language, source, category, PDF-only. Directives supported: `source:`, `lang:`/`language:`, `category:`/`cat:`.
- **Resolve identifiers** — RFC number, DOI, arXiv ID, PubMed ID, NIST/FIPS/SP designation, US legal citation or docket (e.g. `410 U.S. 113`, `20-5364`), SEC accession/CIK/exchange ticker (`ticker TSLA`), patent number (`US11704715B2`), procurement/contract IDs, ISO/OGC/Khronos standards → canonical document record.
- **Watch** — register a document, structured identifier, or query as a change-detection watch (`POST /api/watch`), then poll `POST /api/watch/{id}/check` to detect added/changed documents (new editions, superseding standards, updated regulations).
- **Retrieve** — full document metadata: title, author, source, license, access level, subjects, origin URL, direct PDF, file URL, **related documents** (topically similar records in the same round-trip).
- **Section search** — cross-document section search: search inside a specific section kind (risk factors, material contracts, financials, ...) across every document with extracted full text, with per-doc section attribution (`/api/section/search`).
- **Cite** — 11 formats: APA, MLA, Chicago, IEEE, Harvard, BibTeX, RIS, Bluebook, OSCOLA, ISO 690, ANSI/NISO Z39.29.

## HTTP API (no keys required, CORS enabled)

Base URL: `https://www.conceptio.app`

```
GET /api/health          # document count + last sync stamp
GET /api/search?q=zero+trust&language=en&sources=eurlex&pdf_only=1&limit=20&offset=0
GET /api/resolve?id=RFC+2119          # or doi:, arXiv id, PMID, NIST FIPS, US case citation/docket, ticker TSLA
GET /api/watch                       # list change-detection watches
POST /api/watch                      # create a watch: {target_type: doc|identifier|query, target, label?, limit?}
POST /api/watch/{id}/check           # detect added/changed docs since the last snapshot
GET /api/document/{id}   # full document record + related_documents (incl. section list)
GET /api/document/{id}/sections        # section structure over extracted full text
GET /api/document/{id}/search?q=...    # in-document search with section attribution
GET /api/section/search?q=...&kind=risk_factors&source=sec_edgar  # cross-document section search
GET /api/section/kinds    # supported section kinds (risk_factors, contracts, financials, ...)
GET /api/openapi.json    # machine-readable OpenAPI contract for all endpoints
GET /api/cite/{id}?format=bibtex      # apa, mla, chicago, ieee, harvard, ris, bluebook, oscola, iso690, ansiz39, bibtex
GET /api/sources         # source catalog with counts
GET /api/categories      # category tree
GET /api/languages       # language distribution
GET /api/recent          # recently indexed documents
GET /api/deep            # archive deep search with category/language filters
```

## Example prompts

- "Search for NIST SP 800-207 and summarize its zero trust architecture."
- "Resolve RFC 2119 and give me the direct PDF."
- "Cite document 48123 in BibTeX format."
- "What does CISA's zero trust maturity model say about identity? (conceptio search)"