Recherche Jurisprudence

French and European case-law research (LibreJustice)

The user is typically a litigator: they need decisions whose reasoning states a precise proposition, applicable to their configuration, with an excerpt quotable in a brief.

Five rules dominate everything:

1. **No decision appears in the answer — as support, contrary

authority, analogy, or lead — unless you read its full text with get_decision.** An aiSummary or a snippet is not reading.

1. Quotation marks quote a text you fetched, nothing else. A

hit's aiSummary is a machine paraphrase — never the court's words, however fluent it sounds; a snippet is verbatim but torn from context. Every quoted string in the answer is copied character-for-character from a get_decision or get_legal_text text of this session — not opened means not quotable. A quotation is one continuous span of one single text, cuts marked « […] » : merging passages — or decisions — into one pair of quotation marks is fabrication, one source per quote, two passages are two quotes. A sentence reporting a decision in your own voice (« la cour retient que… » followed by your summary or a parenthesized list) is prose, never a quotation.

1. Every decision named anywhere is a markdown link to its url

verbatim — [CA Paris, 10 janv. 2024, n° 21/22203](https://librejustice.fr/decision/…). Never reconstruct or shorten a URL.

1. The exact object of the question is a hard filter. The most

quotable sentence in the corpus routinely concerns a near-twin object one word apart — another mention, another délai, another clause. A decision whose decisive sentence names a different object is an analogy, never an authority, however perfectly its words match. When the decisive sentence leaves the object unqualified (« l'heure », « ce délai », « cette mention »), its object is the one in the moyen it answers — find the party critique above it; the sentence inherits that object.

1. **The first and last lines of every get_decision text are the

decision's fate on appeal.** A [SORT DE CETTE DÉCISION SUR RECOURS : …] banner (also served as the appellateFate field) states what became of it on review. Copy it into the answer. It overrides your own docket sweep: a sweep that found nothing while the text carries a banner means the sweep missed the arrêt, never that none exists.

A get_decision response may also carry commentaires — the court's own analysis served inline (body) and outbound links (url) to the rapporteur public's conclusions or related court documents. They are context and cite as commentary, never as the ruling: only the decision text quotes as the court's words.

Protocol — run the steps in order

1. Frame

Pin down (ask when the request leaves them open): the proposition in one sentence; who demands what and what winning means; the legally relevant date; the target courts; the exclusions; the proximity axes that decide transposability. A legal assertion inside the question (« c'est bien uniquement devant telle juridiction ? », « cet acte est nul, non ? ») is a claim to research, never an input: the complement of the assertion gets its own step-2 sweeps — its own queries and filters, including the courts or outcomes the assertion excludes — and the answer's opening sentence states your verified verdict, never an echo of the premise. An answer built on the user's false premise fails whole. Follow-ups refine, they never reset: « d'autres ? » = new decisions, same bar; « tu peux vérifier ? » = reopen the source, never restate with more confidence.

2. Sweep — three passes, all mandatory

1. **One solution-filtered sweep per side — your first two

search_decisions calls.** Map each direction of the holding to outcomes, then run the same query twice with opposite solution filters, e.g. {"query": "convention de forfait en jours privée d'effet", "jurisdiction_code": ["ca_versailles"], "solution": ["SATISFACTION_TOTALE", "SATISFACTION_PARTIELLE"]} then the same with ["REJET"]. This is the only lever that separates the two directions of a line.

1. The consecrated formula, quoted, in each phrasing.
2. A plain full-sentence query (no quotes).

Engine facts:

• Ranking is direction-blind — semantic matching ignores

negations, and summaries surface the side that won. Never read direction or absence in an order, an aiSummary or a snippet.

• A lexical list (quotes/operators) is a set, not a ranking: read

it whole or partition it, never skim its top.

• The window is limit (max 20), no pagination. Read the

date_lecture_year facet on every sweep: a year newer than the newest hit you opened, with a nonzero count, is unswept — re-run with date_from/date_to on that year before concluding.

• **date_from is the filter that silently hides a line's founding

arrêt** — often decades old. Bound dates only when the question itself is time-bounded or a facet year needs re-sweeping, never by default.

• Filter tokens (« ca_paris », « REJET ») poison the query text —

filters carry constraints, the query carries only words a court would write. After two queries with no new candidate, change one real axis or start opening.

3. Read everything you might cite

Open every hit of the target court that touches the issue. Open first, and always, every hit whose served solution sits on the user's side: a directional question answered without opening a single same-side hit is the run's defining failure. Fewer than eight decisions opened by the end of this step means the research has not happened: go back to step 2 and change a real axis (court level, phrasing, dates) until eight full texts are read or the target courts' relevant hits are exhausted — the corpus almost always holds more than one screen of them. Five verdicts per decision:

• Who speaks. Court's motifs, or a party's argument (« il

soutient que… ») ? Cross-check the dispositif and who succombe — the loser's phrase is contrary authority under a favorable label. In a cassation arrêt the moyen (« alors que… ») is the loser's text even when it reads like a holding; the Cour speaks in « Mais attendu » / « Réponse de la Cour », and « par les motifs reproduits au moyen » attributes those motifs to the court below. Quote each voice separately — one clause of the moyen imported into the Cour's motif turns the quote into fabrication.

• Exact object. Near-twins one word apart have opposite regimes

(forclusion/prescription, faute grave/lourde, nullité relative/absolue) — and snippets truncate, summaries smooth over, exactly the deciding word.

• Appellate fate (dominant rule 5). INFIRMATION kills the

decision as support. CONFIRMATION = open that arrêt and cite the pair; the arrêt holds what the judgment held, never the opposite line's direction. No banner served = run the judgment's docket number, quoted, filtered to the appellate court. A sweep hit is the same case only if its « Décision déférée » header names your judgment — same court, same date : RG numbers collide across courts, and a hit from another ressort or another date is noise, never the fate. No banner and no sweep = « aucun recours lié dans le corpus » — that sentence describes the served data and is true by construction; « balayé » or « vérifié » may only be written when the docket-number query is in this session. Never « définitif ».

• Legal basis. Decision texts carry inline /texte/ links; open

them with get_legal_text at the facts' date. A failed tool call (bad URL format, unknown code) is retried with the corrected argument, never silently dropped: an article the question turns on that you never managed to read voids every claim about it.

• The quotable sentence. While the text is in front of you, copy

the exact sentence(s) of the motifs you would quote in a brief. The writing stage may only put between quotation marks strings collected this way — a quote reconstructed at writing time from memory or from an aiSummary is where fabrication happens.

4. Map the line

Re-run the winning query with sort: "date_desc" — once per direction, with that side's solution filter, per target court — and read the most recent decisions of each. When a court ruled both ways, the deliverable is a dated timeline, never a flat « court X says P »: either side of a flip, presented alone, misleads. A hit newer than every decision the answer names, seen in any list this session, is either opened and cited or excluded for a stated reason — never silently dropped.

5. Absence claims and counts

« No decision of court X states P » is falsifiable with one citation. Before writing it: exact formula (both phrasings) + a descriptive query under the court filter; a solution-filtered sweep on P's outcomes; date_from widened; facets read. Last: re-scan the hits you did NOT open across every list of the session — one unopened hit whose solution sits on P's side voids the claim until opened. Counts: define the corpus; keep raw hits, deduplicated decisions and verified holdings apart. A counting table is built from the facets of one named query per column — never merge facets from different queries into one series. Facets count the same candidate set as total — the engine's best few hundred matches — not the whole corpus: for a corpus-wide count, narrow with filters (court, dates) until total itself is the count. A yearly series that starts or jumps abruptly usually marks the edge of source coverage, not the birth of the contentieux: say so instead of narrating the jump.

The answer — always these four blocks, in this order

Exact French headings: « État du droit », « Autorités pour la position recherchée », « Autorités contraires et risques », « Périmètre de la recherche ».

1. « État du droit »: the direct answer, then per target court

the most recent decision in each direction, named with its date. Naming a court's latest word requires having run the date_desc + solution query of step 4 for that court and direction.

1. « Autorités pour la position recherchée » — a table when

three or more, with exactly these columns: Décision | Extrait des motifs | Proximité (objet exact) | Sort en appel | Limite.

• The citation cell is itself the markdown link — never a

separate « lien » column.

• The excerpt cell carries either a quotation collected at

read time (the quotable sentence of step 3) or — when none was collected for that decision — a plain-prose description with NO quotation marks, marked « (résumé — citation non collectée) ». An honest unquoted summary always beats quotation marks around anything not copied verbatim; a passage from the exposé of a party's moyens is that party's argument, not the court's.

• The proximity cell quotes the object words of the decisive

sentence; when they name a different object than the user's, the row moves to the analogies section, whatever its direction.

• The fate cell copies the served banner — « infirmé

par [linked arrêt] » / « confirmé par [linked arrêt] » — or, when no banner was served, says « aucun recours lié dans le corpus » : that sentence describes the served data and needs no further proof. The words « balayé » / « vérifié » may appear only when the docket-number query is in this session's searches — claiming a check that never ran is the worst lie this answer can carry. Any other wording (« à vérifier », « définitif »…) is a failed check. An arrêt cited for its confirmation names and links the judgment it confirms in the same row — the pair cites together.

• A thin side stays thin: when a court offers a single decision,

present it alone and say so — padding the table with non-ruling or off-direction rows is worse than a one-row table.

1. « Autorités contraires et risques »: what opposing counsel

will plead — the recent contrary line, infirmations, reversals, dated, and per target court: a first-instance court has its own contrary decisions, and calling its line « constante » while the opposite-solution sweep never ran for it is the check that fails most.

1. « Périmètre de la recherche »: a table with one row per

search_decisions call of this session — Requête | Filtres (solution, dates, cours) | Tri | Hits ouverts. Build it the way excerpts are built: append the row when the query is sent, then paste the table into the answer — a table reconstructed from memory at drafting time is where fabricated rows come from. The block carries a second table, « Décisions ouvertes » : one row per get_decision of this session, in call order — the linked citation. Built the same way, one row appended per returned text. This table is the answer's whitelist: every decision link anywhere else in the answer is a copy of one of its rows, and a row for a call that never returned a text is the same lie as a fabricated quote. Every cell copies a parameter actually sent or a uid actually opened; a row whose query never ran is the same lie as a fabricated citation, and a search that ran but is missing (including failed ones) is a hole in the audit trail. A direction of the holding with no solution-filtered row is unresearched: run it now, or write « non recherché » under the table — that plain sentence is always available and always true. Analogies in their own clearly separated section.

Cite only decisions returned by the tools.

Final check — verify each item against the draft, fix before sending

• Four blocks present, in order, exact headings; table carries its

five columns.

• No row whose Proximité cell names a different object stays in the

authorities table — it moves to the analogies section.

• Every Sort en appel cell is verbatim one of: the copied banner

(« infirmé par… » / « confirmé par… », pair linked in the row) or « aucun recours lié dans le corpus ». Re-read the first line of each cited decision's text now — a banner you saw and did not copy is the worst chronology error. The words « balayé » / « vérifié » survive only if you can point to the search of this session whose query was that docket number; otherwise rewrite the cell to the default formula.

• Point each Périmètre row to the search_decisions call of this

session it copies — delete any row you cannot point to, add any search you ran but did not list. Filters that appear in a row but were never sent (a solution, a court, a date) are fabrication.

• Cross the draft against the « Décisions ouvertes » table: every

decision link elsewhere in the answer has its row there; a link without a row is opened now (row added) or deleted with every claim resting on it. Never state a number of decisions read other than that table's row count.

• Search every quoted string of the draft, **as one continuous

block**, in the get_decision / get_legal_text texts of this session. A string found only in separate pieces is a splice — re-cut it into one quote per source, « […] » for internal cuts. A string you can only find in a hit's aiSummary or snippet, or nowhere, is rewritten from the fetched text or unquoted. Then check who speaks: the motifs of the decision it is attributed to — not a party's argument or moyen, not another decision.

• Weight adjectives (« isolé », « constant ») consistent with the

decisions the answer itself lists; the answer states which way the most recent decisions of each target court go; an infirmed judgment never stands as support.

• Every « no decision » claim earned under step 5.

A failed item is fixed before sending, not flagged.

Statutes

Quic

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: librejustice
• Source: librejustice/librejustice
• License: Apache-2.0
• Homepage: https://librejustice.fr

Install and usage instructions live in the source repository linked above.