For Claude
Litigation prompt structures for Claude
Claude Projects can keep uploaded files and project instructions together across related chats. These prompts provide a scoped assignment, an output format, and source-location requirements; you still need to verify every result against the record.
Product-source review: . Capabilities and data terms vary by plan and can change. Confirm the current product documentation and your organization’s approval before using case material.
Claude for litigation — product fit and limits
Strengths
- Projects can organize chats, uploaded knowledge, and standing project instructions in one workspace.
- Available context length varies by model, plan, and access method; paid plans can support substantial record sets.
- Eligible API customers may have zero-data-retention arrangements, subject to Anthropic’s current terms and model-specific exceptions.
Watch-outs
- Claude is a general-purpose model, not a citator or matter-management system.
- A source-location instruction cannot prevent invented or inaccurate citations; compare each result with the original record.
- Consumer, team, enterprise, and API data terms differ. Confirm the exact account and model your firm approves.
This is an editorial product comparison, not an endorsement or a security assessment.
Using CasePrompts with Claude
Create a Project for the matter and upload the transcript(s) to project knowledge, then paste this as your first message. For ongoing work in the same matter, move the CONTEXT block into the project's custom instructions so every chat inherits it.
From the builder you can copy a finished prompt and open Claude. The prompt’s source-location instruction changes with the selected document type.
Tool interfaces change — adapt as needed, and check your firm's AI use policy before uploading case materials.
Why structure helps more than “summarize this”
Every prompt is built on the CITE method — Context, Instructions, Template, Evidence. It states who is doing the review, what source is in scope, the exact task, the requested output, and how to identify supporting locations. That structure can make an answer easier to inspect, but it cannot keep a model from inventing or misreading material. Check every material output against the source.
Prompts to try in Claude
All templates →Admissions & Key Testimony Extractor
Pull every admission, damaging concession, and 'I don't know' on your key issues out of a transcript — with verbatim Q&A and page:line cites.
Prior Statement Cross-Examination Builder
Align a witness's deposition against every prior statement, classify each conflict, and get a lock-in → confrontation cross sequence with dual cites.
AI Medical Chronology Prompt for Lawyers
Build a date-ordered medical chronology from the complete record, with source citations for every encounter and explicit flags for treatment gaps, prior history, causation language, and missing records.
AI Page-Line Deposition Summary Prompt
Turn a complete deposition transcript into a topic-based page-line summary with precise cites, verbatim excerpts, admissions, exhibits, and unanswered questions.
Record Interrogation Session Setup
The standing-rules prompt: paste this once at the start of a chat session and every answer after it comes cited, sourced, and honest about gaps.
Expert Report vs. Deposition Audit
Audit an expert's deposition against their report: new opinions, abandoned opinions, changed bases, and undisclosed reliance materials — exclusion-motion fuel.
CasePrompts is vendor-neutral and built local-first — the builder runs in your browser and never sees your case files. See how your data is handled → Know which tier of tool your firm has approved before uploading case materials to Claude.
Official product sources
These first-party pages supported the capability and data-term statements above when reviewed on .