45 prompts built for attorneys and optimized for Anthropic Claude (Opus, Sonnet, Haiku) — drafting, document review, legal research, and client communication, with anti-hallucination safeguards on every prompt.
Drafts a commercial agreement from the term sheet you paste, with every price, term and cap taken from your words, and a redline list against the other side's draft if you paste one. Terms nobody supplied stay blank for you.
Drafts a confidentiality clause and any restrictive covenant clause for an agreement with a California-based counterparty. Every California rule comes from the statute or decision text you paste; a restraint the pasted text does not clear is held back for you.
Two heavy-research prompts in one: an Am Law 100 research memorandum and a section-by-section statute and regulation breakdown with compliance obligations.
Drafts a pre-litigation demand letter or an intellectual property cease-and-desist letter from the facts, documents and governing text you paste. The relief and response date are yours; no law, damages theory or deadline is added from memory.
Drafts a plaintiff-side demand letter in a New York commercial construction dispute from the contract, payment records and New York statute or case text you paste. Balance, interest and cure period come only from your inputs and recomputed figures.
Turns an appellate decision you paste into an IRAC-style research memo in which every statement carries a page or paragraph marker. The holding is quoted, dicta only where the court marks it, and later history is left to your citator.
Generate a clause-by-clause risk matrix for a New York commercial lease. Tested on Claude vs Gemini — Claude caught a hidden assignment/subletting conflict.
Two client-comm prompts in one: a sophisticated 1-2 page advisory letter and a sub-500 word case status email with bulleted action items.
Assesses a pending motion argument by argument from the briefs, record excerpts and authorities you paste. Each cited case is checked against pasted text, and probabilities and values appear only as you supply them.
Research the requirements for a valid Texas mechanic's lien on residential property. Tested on Claude and Gemini — see which model wins on accuracy vs currency.
Use Claude to score an O-1A candidate against all eight regulatory criteria, recommend the lead 3-5 criteria, and identify documentation gaps.
Turns pasted medical records into a dated, source-quoted treatment chronology with recomputed visit counts and durations, treatment gaps listed without invented explanations, and conflicts between records shown, never resolved.
Builds a phone-ready intake screening form for a new PI inquiry: liability and damages questions, red-flag prompts, document requests, evidence-preservation steps and a scoring grid with your threshold applied exactly, and every limitation date left for you to verify.
Drafts numbered interrogatories to a personal injury defendant, each tied to its source and the element it serves, sized against the numerical limit only if you paste the governing rule, with limits, timing and format left for you to confirm.
Builds a defendant driver deposition outline in XML-structured form from the police report, statements and records you paste. Impeachment points quote their source, no prior testimony is invented, and the questions to be asked aloud are kept clean.
Generate a full personal injury demand letter with recomputed itemized damages, using only the valuation method, demand and response date you supply.
Run an internal personal injury valuation that recomputes economic damages and builds scenarios only from the comparables, ranges and rule text you supply.
Standing instructions that make Claude draft numbered commercial contracts only from the term sheet you paste, with missing terms left blank, UCC points raised as questions, and review notes for the supervising attorney.
Turns your instructions into a first draft under the law of England and Wales in British legal English. Terms you did not supply stay blank, and statute text comes only from what you paste.
Standing instructions that make Claude review each contract you paste for one named party: a 14-topic checklist, tiers set by your stated priorities, and redlines that never add a burden for your side.
Configure Claude as a research assistant with confidence tagging, citation-fabrication bans, IRAC output, and a built-in verification checklist for every claim.
Configure Claude as a legal communication specialist that switches tone and structure across five recipient profiles, with built-in formatting and ethics guardrails.
A meta-guide on the four most common Claude system prompt pitfalls for law firms — bloat, contradictions, over-constraining, ignoring model updates — plus versioning advice.
Drop in a messy client email or transcript and get a clean six-section intake brief — parties, terms, issues, missing info, next steps — ready to feed into drafting.
Aggregate a clause review into a CEO-ready dashboard — heat map, top three risks with financial ranges, negotiation priorities, and a three-sentence executive summary.
Convert raw legal analysis into two polished deliverables in one shot — a one-page traffic-light executive memo and a full legal memo with redlines and appendix.
Configure Claude as a multi-jurisdictional legal analyst that compares US, EU, UK, and Canadian rules, flags conflicts, and outputs a structured risk matrix.
Walk through every clause with explicit reasoning, a 1-10 risk score, suggested redlines, and a confidence rating you can verify line by line.
One prompt, six stages, full audit trail — runs the entire intake-to-deliverable pipeline inside a single Claude session with global anti-hallucination rules.
Turns intake notes and the documents in hand into a candid internal evaluation memo: liability by element, comparative fault exposure, documented versus undocumented damages, collectability, evidence to preserve, and an accept / decline / investigate recommendation.
Separates the client's documented baseline from post-incident findings, builds a quoted before/after comparison for each body part, anticipates the defense reading of the same records, and drafts the questions to put to the treating physician.
Drafts a time-limited policy-limits demand that can be accepted exactly as written, and checks every term against the governing statute or case text you paste in. It states no state requirement or bad-faith standard on its own.
Audits every lien and reimbursement claim on a settlement (Medicare, Medicaid, ERISA or other health plan, provider liens, med-pay), isolates unrelated charges and arithmetic errors, then drafts the reduction request. It never says a doctrine applies without the governing text.
Compares the requests you served with the responses and objections you received, classifies every deficiency with both texts quoted, and drafts the meet-and-confer letter with your response date.
Builds a deposition outline for a defense IME physician, biomechanical engineer or accident reconstructionist from the report and CV: bias, materials not reviewed, methodology, and every assumption set against the quoted record.
Drafts the plaintiff's mediation statement: liability in evidence order, damages with documented totals and visible arithmetic, the defense's best arguments answered from the record, and a separate confidential section for the mediator.
Turns first-consultation notes and the documents received into an internal assessment memo: safety concerns first, the issues in play, what is documented versus only told, urgent matters, documents to collect, client questions and immediate tasks. It states no law and predicts no outcome.
Drafts a parenting plan from the parents' work schedules, the child's school and activity calendar and the terms you supply: schedule, holidays and vacations, exchanges, decision-making, communication, travel and dispute resolution. Every term comes from your inputs; a topic with no input is left for you to complete.
Drafts the client's declaration or affidavit supporting temporary orders from the client's own account: first-hand facts only, each dated and specific, in the client's voice, tied to the exhibits you list. What the client learned from others is held out and listed for your decision. No conclusions, no characterization of the other parent, no legal standard.
Reviews a proposed settlement agreement or term sheet against your client's goals and the financial documents: missing, ambiguous and inconsistent terms, figures that do not reconcile, obligations with no enforcement mechanism, and a redline list of clarifications.
Turns the charging document, arrest report and intake notes into an internal memo: each charge with elements from statute text you paste, the facts alleged for each element and their source, custody status, immediate tasks and client questions.
Compares police reports, supplemental reports, witness statements and any transcript: one timeline with every time stamp and source, each inconsistency quoted side by side, facts asserted with no stated basis of knowledge, items referred to but not produced, and follow-up discovery to request.
Builds a suppression motion outline from the record: the evidence at issue, the sequence of police conduct with quotes and time stamps, each ground with the authority you paste or an authority-needed marker, the prosecution's likely facts, hearing witnesses and exhibits, and proposed findings.
Builds an internal memo comparing a plea offer with trial, using only the offer, the statute or guideline text and the evidence assessment you paste. Collateral consequences come back as questions with an authority-needed marker, and the model predicts no conviction odds and no sentence.
Builds a cross-examination outline for a police witness from the officer's reports, prior testimony and the recordings transcript. Leading, one-fact questions, impeachment chapters built on quoted inconsistencies, and omissions only where the documents show them.
Pro members get the Prompt Runner: fill in variables, choose your model, and get output with anti-hallucination checks.
See Plans* These prompts are drafting and analysis aids for licensed attorneys. They are not legal advice and do not replace professional judgment. AI output can contain errors, including invented citations and incorrect deadlines: verify every fact, figure, authority and deadline against primary sources before relying on it. Rules vary by jurisdiction, and you remain responsible for the final work product.