The corpus
Everything on this site is derived from one pinned copy of the USPTO Manual of Patent Examining Procedure. It does not track the USPTO site: it is a fixed snapshot, identified by the version string below, so that a citation to a section here means the same thing tomorrow as it does today. USPTO guidance issued after that revision is kept separately and never folded into the corpus text.
What it is
- Source
- USPTO Manual of Patent Examining Procedure, uspto.gov
- Edition
- Ninth Edition
- Revision
- Revision 01.2024
- Version identifier
mpep-e9r01-2024-html- Parsed
- Publisher
- Patent Agentic LLC
- Access terms
- Free to read and cite. Terms of use — programmatic or bulk access requires permission.
Coverage
- Chapters
- 29
- File-level sections
- 736 — the whole manual, all present as source text
- Structurally annotated
- 425 of 736 (58%)
- Cited authorities
- 2,411
- Cross-references
- 3,691
Every section is readable. “Structurally annotated” means a section has additionally been parsed into typed citation edges — its statutes, regulations, cases, cross-references and form paragraphs as machine-readable relations with verified quotes. Annotation is still in progress, so a section page shows which of the two it is rather than implying the whole manual is structured.
Getting the data
- MCP server and JSON API — per-section JSON with the cited-authority digest, search, citation resolution and reverse lookups.
- llms.txt — a single-file summary of the corpus for language models.
- Bulk export — not a public download. Ask on the contact page.
Corrections
The source text is reproduced as published; it is not edited. See methodology for which parts of this site are generated and which are not, and corrections for what is worth reporting. If a parse looks wrong — a mis-split section, a citation edge pointing at the wrong authority, a heading in the wrong place — that is a bug in this site, not in the manual, and it is worth reporting. Send it via contact with the section number.