Nexus Legal Mcp

PRIMARY LEGAL ANALYSIS (Node A — ISO 31000). Reads a document in full: extracts parties, risks, clauses, time limits, applicable law and relevant case law. Applies the jurisdiction (ES/CO/GB/etc.) and the corresponding branch of law. Returns a structured report with certainty locks [L1]/[L2-J]/[L3-NV]/[L4] and blocking signals [L5-C]/[L5-P]. Ends with a NEXUS-AUDIT-TRAIL block for later audit. Cost: ~1-3 credits depending on size and model tier. USE WHEN: the user supplies a legal document and asks for analysis of risk, viability, critical clauses, or a defence at trial.

OPEN LEGAL QUESTION (Node A — Q&A mode). A legal question in natural language, with or without a reference document. Returns an answer with certainty locks and statutory / case-law references where they apply. Supports conversation threads through the `history` parameter. Cost: 1 credit. USE WHEN: the user asks a specific legal question (what is the limitation period? is this clause enforceable? what has the Supreme Court said about X?) and does NOT need a full document analysis.

LEGAL DRAFTING (Node A — Drafter mode). Produces the first draft of a document (appeal, claim, defence, contract clause, opinion) following the format and mandatory locks of MODULE-DRAFT. If the date of service is uncertain it degrades to [L3-NV] and asks for verification in the brief. USE WHEN: the user asks for a specific document to be drafted, usually after a Node A analysis or over a supplied document. Cost: 2-4 credits.

CROSS-AUDIT OF THE NODE A ANALYSIS (Node B — adversarial control). Checks the previous analysis against the original document: detects unsupported assertions, misapplied certainty locks, omissions, internal contradictions, and produces the V-XX verification block. Returns a confidence score 0-100. USE WHEN: the user wants a second layer of validation before putting the analysis in front of a client or a court. Requires `nexus_analyze` to have run first. Cost: 1-2 credits.

MULTI-AGENT CONSENSUS NODE A ↔ NODE B. Node A (the analyst) produces the report and the Node B (adversarial auditor) validates it; if B finds discrepancies (deterministic veto), A re-evaluates taking B's feedback on board and B audits again, until FULL or PARTIAL CONSENSUS (max. 2 iterations). Returns final report + final audit + consensus verdict, with deterministic citation verification against the official corpus. `jurisdiction` is REQUIRED (no default). LONG OPERATION: 3-6 minutes of streaming (this stdio client has no 60s limit unlike the remote connector, so the call waits for the complete result). USE WHEN: the user wants the analysis with the strongest guarantee (analyst + auditor in a loop) in a single call. Cost: ~3-7 credits depending on iterations.

MONTE CARLO SCENARIO SIMULATION (Monte Carlo node — ISO 31000 §6). Stresses a contract's risk profile under user-defined variables. For each variable it simulates three intensities (Moderate P25 / Base P50 / Severe P75), re-grades the risks from the Node A analysis and computes a Contractual Resilience Index (CRI 0-100). Returns impact tables and hardening recommendations. USE WHEN: the client wants to know how the contract holds up under external shocks (M&A, due diligence, ratings). Cost: 4 credits.

ADMINISTRATIVE DOCTRINE SEARCH (Doctrine node — RAG + LLM). Queries the administrative and tax doctrine corpus (vinculantes V-XXXX-YY, TEAC RG XXXX/XXXX, Consejo de Estado dictámenes). Distingue 'Doctrina consolidada' (≥2 fuentes ORIGEN-A/B) vs 'Criterio orientativo' ([L3-NV]). Indica vinculatoriedad y conflictos with case law. USE WHEN: the user asks what the administrative position is on a specific tax or regulatory point. Cost: 1 credit.

SECOND LEGAL OPINION (Opinion node — adversarial control run on a different LLM from Node A). Takes a Node A answer and re-examines it with a different model, looking for: omissions, bias, misapplication of the law, or case law ignored. Useful for high-value case files. USE WHEN: the client requires a second validation of a legal position before acting. Cost: 2 credits.

ADVERSARIAL RED TEAM (Node C — destructive mode). Hostile analysis of the document as if written by the other side's lawyer: legal gaps, trap clauses, power asymmetries, catastrophic risks. Returns structured JSON with vulnerabilities (severity High/Medium) and a catastrophic_risk_score 1-100. USE WHEN: pre-signature of a strategic contract, M&A audit, aggressive due diligence. Cost: 5 credits.

ADVERSARIAL ARGUMENT (Node C — opposing-party mode). For each conclusion of the Node A analysis, builds the strongest argument the other side would run. Useful to prepare litigation or to anticipate objections in negotiation. Different from `nexus_redteam`: adversarial DEBATES the conclusions (prose), redteam DESTROYS clauses (structured JSON). USE WHEN: the client needs to anticipate the other side. Cost: 2-3 credits.

MULTI-JURISDICTION COMPARISON. For the facts or contract supplied, compares how the legal system of each listed jurisdiction (2-15) treats it. Identifies: most favourable applicable law, optimal forum, tax exposure by country (double-taxation treaties), recognition of judgments between jurisdictions, conflicts of laws. Returns a comparison structured by jurisdiction (risk, applicable law, recommendation). USE WHEN: the client operates in several countries and must decide where to litigate, contract or establish. Cost: 2-4 credits depending on the number of jurisdictions.

SEMANTIC CASE-LAW SEARCH. Vector query (voyage-law-2) over the curated Nexus corpus: GB ~138k judgments (Scotland and Northern Ireland) · ES ~141k (CENDOJ) · CO ~106k · more jurisdictions in progress. Returns top-K results with title, source, extract (2,000 chars) and a permanent URL to the full text. USE WHEN: the user asks to find judgments on a specific point, test an argument, or build a bundle of authorities. Cost: free (search only).

Verify whether a legal citation (court judgment or statute article) is accurate. Returns verification status, similarity score, and official source URL. Supports multiple jurisdictions (ES, CO, …) according to the wired verification corpus, without privileging any single one — pass `jurisdiction` to target a specific corpus, or omit it to let the backend infer it. Jurisdictions without a wired corpus return an honest 'not available' status instead of a misleading result. USE WHEN: the user wants to check whether a case-law or statutory citation exists, is correct, or may be hallucinated.

Verify a document's citations —LEGISLATION and CASE LAW— against the corpus of their jurisdiction. Auto-detects the citations in the text (art. 1902 CC; section 80 of the Employment Act 1968; [2007] SGCA 37…), checks force/repeal/existence at a given date and returns a structured per-citation report with canonical text and the official source URL where there is one. Deterministic, no LLM cost. SCOPE: MULTI-JURISDICTIONAL. Each citation is routed by its FORM to its own verifier; there is NO default jurisdiction and nothing falls back to Spain. The response declares `verifiable_jurisdictions` — citations from any other jurisdiction come back as 'unknown' (NOT checked), never as 'not_found'; check that list before reading a `total:0` as a clean document. READING THE RESULT: `found` means the citation EXISTS — which is also true when it resolves to a DIFFERENT decision. Never derive a verdict from `status` alone: read (`status`, `meaning`), or the aggregate `flagged` / `review_required`. USE WHEN: auditing a document before filing it, or detecting hallucinated citations in a text. The PDF dossier export is NOT exposed as a tool (binary); it is only available from the web UI.

Calculate a procedural time limit under the law of a SPECIFIC forum. `jurisdiction` is required and has NO default: a time limit computed under another jurisdiction's rules produces a date the client may act on, and a missed limitation period cannot be undone. Supported: ES (LEC + LOPJ, CGPJ and regional holidays, August as a non-working month) and GB-EAW (CPR r 2.8, clear days, England and Wales bank holidays, and NO August vacation). Scotland and Northern Ireland are refused rather than approximated: their rules and their bank holidays differ. Returns the due date, the rule applied and every adjustment made. USE WHEN: the user needs to compute a procedural time limit (appeal, cassation, defence, filing a defence, appellant's notice…) from a base date.

Cross-reference between statute articles and court judgments. Find which statutes a judgment cites, or which judgments cite a particular statute article. USE WHEN: the user wants to know which provisions a judgment cites, or which judgments apply a particular provision.

Analyze a legal document that lives in the user's Box account WITHOUT downloading it: pass the Box file_id and Nexus retrieves the user's Box token (OAuth), downloads the binary, extracts text (Gemini OCR for PDF/DOCX), runs the legal analysis (Node A) and — by default — writes the result back to the Box file as metadata. Requires the user behind the API key to have an active Box connection. Charges 1 credit only on success (2 for 'auditoria' mode). Max file size 50 MB. USE WHEN: the user wants to analyse a file they already hold in Box and refers to it by file_id.

STRATEGIC NEGOTIATION PLAYBOOK. From a Node A risk analysis (ISO 31000), generates three levels of counter-proposal for each finding (conservative / balanced / aggressive) with suggested clause wording and trade-offs. Returns structured JSON + deterministic citation verification. USE WHEN: you already have an analysis of a contract and need to prepare for the negotiating table. Cost: 3 credits.

VERSION COMPARISON V1↔V2 (Node B — dual audit). Analyses two versions of the same document (V1 original vs V2 returned by the other side) and reports: alarms resolved, new vulnerabilities introduced, and an executive view of the commercial impact, with deterministic citation verification. With `jurisdictionB` it activates the bilateral checks (governing law / forum altered between versions). USE WHEN: the other side returns a redline and you need to know what improved and what traps went in. Cost: 2 credits.

CONFLICT CHECK (conflicts of interest). Deterministic check BEFORE opening a case file: compares the parties given against the parties of the firm's existing case files (pattern matching + trigram similarity + shared tokens). Returns `{ clean, conflicts[] }`. Consumes NO credits. USE WHEN: you are about to open a case and need descartar un conflicto de intereses antes de aceptarlo.

FIRM-WIDE ATTENTION INBOX. Aggregates in one call everything that needs attention across the firm: outputs pending review, evidence unverified or flagged as possibly hallucinated, and time limits overdue or falling due (next 7 days). Firm-scoped, read-only, consumes NO credits. USE WHEN: you want an actionable summary of the firm's state to prioritise the day.

CASE-FILE RAG CHAT (interactive auditor). Multi-turn conversational Q&A over a document and its Node A/B analysis supplied in `context` (BYOD, no persisted case file). Returns the answer with deterministic citation verification annotated. USE WHEN: you want to interrogate a contract that has already been analysed ('which clause caps my liability?', 'what happens if I miss the deadline?'). Cost: 1 credit.

Semantic search over IN-FORCE legislation articles (Voyage-3 embeddings) of the consolidated statute corpus. Filters by in-force status at a date (lex temporis), instrument type and subject. Returns the most similar provisions with their heading, number, instrument id, literal text and official URL. USE WHEN: looking for the provision that applies to a question, without drafting. Consumes no LLM credits (only the query embedding).

Deterministic fetch of the literal text of a specific statute article, by the identifier its jurisdiction uses. Where the lookup is bounded by the date (Spain today) it returns the version in force then, for historical lex temporis queries, with its rubric and vigency window; where it is not (the United Kingdom), the text is the official one but the in-force status is NOT checked and the response says so. USE WHEN: the instrument and provision are known exactly and the official text is wanted, not a search. Consumes no credits.

Stufenbau pyramid of the statute corpus: coverage aggregated by hierarchical level (constitutional, legal, regulatory, …) for a jurisdiction, with totals of instruments/provisions, in-force/repealed and in-force percentage per level. Public discovery (no credits). USE WHEN: a coverage view by hierarchical level is wanted.

Self-discoverable inventory of the indexed jurisprudence corpus: for each jurisprudencia_<iso> table, its inferred jurisdiction, total documents, date range (oldest/newest) and top sources. Pure metadata, no credits. USE WHEN: you want to know which jurisdictions and case-law sources exist before searching.

Coverage of the STATUTE corpus (analogous to nexus_corpus_coverage but for legislation): jurisdictions with corpus, per-jurisdiction breakdown for the jurisdictions this account has active, totals of instruments/versions/articles (relational store only), last update of OUR COPY and ingestion lag. Public discovery, no credits. USE WHEN: you want the state and freshness of the statute corpus.