API sandbox
Run the tools your agent will call.
38 tools behind one MCP endpoint and one bearer token. Pick one, run it with your own free key, and copy the call your client will make.
Live today: 10.5M+ US patents, 10.3K examiners, 8.5M CPC classifications, MPEP and EPO Guidelines, 3.5K sections. Case law, Tax, Contracts, Litigation and Corporate are on the roadmap; each arrives as more tools on this same endpoint, so nothing about the integration changes.
Sign in and your free key is issued on the spot. Then these examples fill in with it and run.
Sign inAsk Antecedint
askStart here. Pass the question as the user asked it, or a claim, and it is routed to the questions that answer it, each a walk over one or more data gardens. A garden is the tuned data: one arrangement of a corpus (the prior-art index, the citation graph, the patent map, claim text, examiner profiles, CPC routing, the MPEP and EPO sections, form paragraphs, design codes), built and measured for the questions it suits. Application numbers, MPEP sections, form paragraph codes, CPC symbols, art units, examiner names and claim shape in the text are read as routing signals; a question that spans walks runs several in parallel. Every answer names the gardens it walked and returns documents; every garden walked states what it holds (holds: value, n, date, source), what is being built on it (roadmap) and what serves it (served_by). Pass asOf with the priority date of the claim: a walk over the prior-art index bounds art to before that day and reports time_policy, and each document carries the date it was decided on and whether it predates the day. Pass excludeFamilyOf to drop the application's own family; the answer reports leaks and the rows that carried no family id. routing says how the questions were chosen; not_covered lists what was not run and why, including gaps that bear on the question, such as non-patent literature. The prior-art index is tuned for what an examiner will cite; litigation-grade §102 and §103 retrieval is being built and the response says so where it applies. Pass questions to pin the set from list_gardens. A walk that needs a cold encoder is skipped with a HUB_COLD error and a remedy, and one that asks for rerank while the reranker is cold or still warming answers RERANKER_COLD, both at once rather than after a load; call get_status with warm true (and rerank true) first, or set waitForHub. Each answer that used a model reports hub: the time per model, and served_cold, true when a model loaded or was still warming inside the call.
0 credits a call · times out at 120s
The question as asked, or claim text. Claim text routes to prior_art_claim; a numbered claim set with dependencies also routes to claim_defects.
Question types from list_gardens (prior_art_claim, examiner_citations, patent_landscape, claim_defects, examiner_profile, cpc_routing, examination_rule, form_paragraph, design_code). Pins the set and skips routing.
Documents per answer (default 10).
Character budget for the documents across all answers (default 24000). Documents are clipped from the tail, never reordered; truncated says when.
false returns ids, refs, scores and dates only (default true). What an evaluation or a dedup pass reads.
The day art must predate, strictly (YYYY, YYYY-MM or YYYY-MM-DD; a year or a month resolves to its first day). Pass the priority date of the claim being searched. Without it, results may postdate the claim and the answer says so.
The older name for asOf, honoured when asOf is absent.
US application number whose patent family the index walks drop. The answer reports whether the family resolved and how many returned rows carried no family id.
Restrict the index walks to these CPC subclasses. Excludes unclassified references, so leave it unset on a first pass.
Run the ColBERT reranker in prior_art_claim (slower; loads a second model on a cold hub).
prior_art_claim reads past the abstracts by default: the independent claims of the top three hits and, for the top two, the passage closest to each element, in answer.dig. false skips it for a faster first look. Included in the flat price either way.
Force the rewrite of the question into patent language on or off. Unset lets the walk decide: on for a short description, off for a claim.
Wait for a cold encoder instead of skipping the index walks (default false; about 75 s on the evaluation tier).
Plug it in
curl -sS https://mcp.antecedint.com/api/v1/mcp \
-H 'Authorization: Bearer ak_live_YOUR_KEY' \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"ask","arguments":{"question":"What does MPEP 2106 require an examiner to establish before rejecting a claim under 101?","maxDocuments":5}}}'Load your key above and these fill in with it.