CoolFace
Datasetpublic

docketx/texas-caselaw

Texas Case Law Code & tools: github.com/docketx — legal-scrambler pseudonymises a case file on your own hardware before a frontier model sees it; claude-for-legal is the Claude Code plugin (docketx-open-law) that loads these datasets and checks citations against them. Source & credit — Free Law Project / CourtListener Every opinion in this dataset comes from the Free Law Project / CourtListener bulk export of 2026-06-30 (10,798,347 opinions). CourtListener… See the full description on the dataset page: https://huggingface.co/datasets/docketx/texas-caselaw.

sourceHugging Facecc0-1.0updated 19h agoView on Hugging Face
0likes306downloads
Dataset Card

Texas Case Law

Code & tools: github.com/docketxlegal-scrambler pseudonymises a case file on your own hardware before a frontier model sees it; claude-for-legal is the Claude Code plugin (docketx-open-law) that loads these datasets and checks citations against them.

Source & credit — Free Law Project / CourtListener

Every opinion in this dataset comes from the Free Law Project / CourtListener bulk export of 2026-06-30 (10,798,347 opinions). CourtListener gathered, cleaned, deduplicated and published that record; Free Law Project — the 501(c)(3) non-profit behind it, building free public access to American law since 2010 — pays for it. We sliced and reformatted; they did the hard part. Go to the source: <https://www.courtlistener.com>.

Free Law Project is a non-profit, and it is worth supporting: <https://free.law/donate/>. If this data is useful to you, credit Free Law Project and CourtListener too, and consider going straight to the source: CourtListener runs its own API and MCP server.

616,651 Texas appellate decisions, 15 January 1840 through 11 September 2026, full text. The base snapshot is 615,391 decisions in 13 parts of 50,000 (deterministic doc-id order), sourced from the CourtListener bulk export of 2026-06-30. data/tx-delta-2026-06-27-to-2026-09-11.jsonl.gz ADDS 1,260 lead/combined opinions filed after that snapshot, harvested directly from CourtListener's live search API and its own object storage (never txcourts.gov's search portal, which disallows automated access) — added 2026-09-14. The date range above is the newest and oldest opinion held across BOTH files; it is not a claim that every court is current to it — see "Coverage is not uniform" below.

Coverage (base snapshot): Supreme Court of Texas (33,108), Court of Criminal Appeals (70,462), the Courts of Appeals (508,165), the Commission of Appeals 1918-1945 (3,547), and the Business Court (109). Memorandum opinions included; a decision's majority, concurring and dissenting opinions are held together under its document where CourtListener's own record already combines them.

Each row: id, title, citation, court, date, text (full opinion body), source, license, extra.

Coverage is not uniform — three courts of appeals stop early

One date for 615,391 documents hid a real hole, so it is stated here plainly. Measured 2026-09-13 per court against this corpus:

courtdocumentsnewest opinion heldopinions in 2026
Fifth COA (Dallas)38,4422025-08-260
Fourteenth COA (Houston)36,3572024-12-060
Twelfth COA (Tyler)11,8582024-11-270
First COA (Houston) (for comparison)40,0742026-06-26526
Third COA (Austin) (for comparison)46,1102026-06-26432

86,657 documents — 14.1% of this dataset — belong to courts that have contributed essentially nothing for ten to nineteen months, and Dallas and Houston-14 cover the two largest metropolitan areas in the state. A search of this dataset that returns nothing from those courts for a recent period is a gap in the record, not an absence of authority.

The gap is upstream, not in our packaging. The same three courts stop at the same dates in CourtListener's own 2026-06-30 bulk export and in its live API, both checked directly — so a refresh of this dataset will not close it. The shape is a decay through late 2024 rather than a clean edge (Dallas fell 160 → 93 opinions a month while Houston-1 ran flat), which looks like a scraper failing rather than a policy change. If you work with Texas appellate data, this is worth reporting to Free Law Project.

615,391 here, 662,717 in us-caselaw-tx — why the two numbers differ

Both are measured and they count different things. This dataset groups a decision's majority, concurring and dissenting opinions into one document; `docketx/us-caselaw-tx` publishes one row per opinion and applies no minimum length, so it reports 662,717. That card carries the arithmetic in full: 615,391 + 38,008 separately filed opinions + 9,318 documents below this build's 500-character floor = 662,717.

Component counts here, verified against the serving index on 2026-09-13: Supreme Court of Texas 33,108 + Court of Criminal Appeals 70,462 + Courts of Appeals 508,165 + Commission of Appeals 3,547 + Business Court 109 = 615,391.

Load it

python
from datasets import load_dataset
ds = load_dataset("docketx/texas-caselaw")

Provenance and license

Judicial opinions are edicts of government: uncopyrightable works of the public domain (Banks v. Manchester, 128 U.S. 244 (1888); Georgia v. Public.Resource.Org, 590 U.S. 255 (2020)). This packaging is released under CC0 1.0. Source: CourtListener bulk export (2026-06-30), Free Law Project — https://free.law. The export is itself incomplete for some historical courts and, as measured above, for three Courts of Appeals over the last 10-19 months; absence from this dataset does not mean no authority exists.

Part of the DocketRouter legal corpora: https://huggingface.co/docketx — the same corpus powers the DocketRouter legal-grounding API.