Claude CoWork for Litigation Paralegals

Q: What This Looks Like in Your Week?

**Monday** — Discovery cutoff in 10 days. You feed Claude the deposition transcript from Friday's deposition and get a working admissions table and topical index before the morning team meeting. The supervising attorney gets a briefing memo instead of a transcript.

Q: What to Avoid?

**Trusting citation output.** Claude will generate plausible-looking case names, reporter volumes, and page numbers that do not exist. Every citation must read as [CITATION NEEDED] until you verify it against Westlaw or Lexis. Build this into your review checklist, not your trust.

What is Claude CoWork?

Claude CoWork is the practice of using Claude as a persistent, context-aware assistant embedded directly in your litigation workflow. This is not about pasting a deposition transcript and hoping for a usable summary. It is about configuring Claude with your firm's style, the controlling procedural rules, and the case-specific facts so that every interaction produces a working draft — not a starting-over prompt.

Claude-native prompts. The prompts in this guide use Claude's native XML tag structure (<context>, <instructions>, <format>, <avoid>) for more precise, consistent output. These tags help Claude parse your intent with less ambiguity. They work in ChatGPT too, but are optimized for Claude.

Litigation paralegals sit at the convergence of the heaviest document volume in the legal system: depositions that run hundreds of pages, discovery productions in the tens of thousands, and motion deadlines that compress all of it into 48 hours. Claude does not replace the attorney — the attorney reviews, edits, and signs everything. But Claude collapses the gap between raw material and reviewable draft from hours to minutes.

This guide covers how to configure Claude for litigation-specific work, the five workflows that return the most time, and the rules that keep AI-generated legal content from creating more risk than it saves.

Install the Litigation Paralegal Plugin

This guide works on three Claude surfaces. The plugin is the fastest path on two of them. Pick whichever you use:

If you're on Cowork (desktop or mobile app)

Claude Cowork is Anthropic's agentic workspace — Claude completes work autonomously and returns finished deliverables. The Litigation Paralegal plugin packages the workflows below as native skills and slash commands.

Open the Cowork plugin directory in your desktop app.
Filter by Cowork, search for "Litigation Paralegal", and click Install.
The plugin's slash commands and ambient skills are now available in any Cowork task.

If you don't see the plugin in the directory yet, install via custom marketplace: paste https://github.com/alexclowe/awesome-claude-cowork-plugins in your Cowork plugin settings.

If you're on Claude Code (CLI)

Install from your terminal:

claude plugin add alexclowe/awesome-claude-cowork-plugins/litigation-paralegal

The plugin's slash commands and skills load on next session.

If you're on Claude.ai (web chat only)

Plugins aren't directly installable on the web chat surface. You have two options:

Use the prompts in this guide directly in a Claude Project (covered in the next section). Same outputs, more typing.
Upload the plugin's skills as a zip via Settings → Features → Custom Skills (Pro/Max/Team/Enterprise plans). Higher friction; only worth it if you want the auto-activating skills, not the slash commands.

What the plugin gives you (any surface)

Slash command	What it does
`/deposition-summary`	Extract key admissions, build a topical index, and flag contradictions and impeachable statements from a deposition
`/case-timeline`	Build a chronological case timeline from filings, discovery production, and witness statements; flag discovery gaps
`/motion-draft`	Template-driven motion writing (motion to dismiss, summary judgment, in limine) with citation slots and jurisdictional rules
`/discovery-hold`	Analyze documents for privilege, relevance, and hold scope; generate privilege-log entries

Auto-activating skills (no command needed — Claude applies them when relevant):

Legal Research Cite Finder — Pull cases, statutes, and rules of civil procedure by topic and jurisdiction; flag overruled authorities and circuit splits
Evidence Relevance Checker — Apply Federal Rules of Evidence (relevance, hearsay, character, authentication, chain of custody) to flag admissibility issues

The plugin works standalone for one-off tasks. Pair it with the surface-specific setup below for persistent context across every task — that combination is the full Claude CoWork setup.

Setting Up Claude for Litigation Work

Surface note: The Project setup below is for claude.ai web users. Cowork users have their own task-context mechanism (set context once when starting a Cowork task). Claude Code users get the plugin's ambient skills automatically — no Project setup needed. The workflows themselves are surface-agnostic — paste the prompts wherever you're working. Step 1: Create a Litigation Project. In Claude, click Projects and create one called something like "Litigation Practice" or name it after your firm. This is your persistent workspace — context carries across every conversation you start inside it.

Step 2: Set your custom instructions. In the Project settings, add:

You are my litigation support assistant. Here is my context:

<practice-profile>
- Role: Litigation Paralegal
- Firm type: [BigLaw / Boutique litigation firm / In-house legal / Government]
- Practice area: [Commercial litigation / Employment / Products liability / IP / Antitrust / Other]
- Court(s): [Federal — which district / State — which jurisdiction]
- Procedural rules: FRCP (federal) or [state equivalent]
- Case management system: [Relativity / Clio / iManage / NetDocuments / Other]
- Citation format: Bluebook (or jurisdiction-specific)
</practice-profile>

<rules>
- Never include real client names, case numbers, or confidential facts. Use placeholders like [CLIENT], [OPPONENT], [CASE NO.].
- All output is a draft for attorney review. Append: "DRAFT — ATTORNEY REVIEW REQUIRED."
- Never fabricate case citations, statutes, or rule numbers. Use [CITATION NEEDED] where a cite belongs.
- Bluebook citation slots are placeholders only. I will verify and fill them against primary sources.
- Do not provide legal advice or predict litigation outcomes.
</rules>

Step 3: Upload standing reference documents. Add your firm's motion templates, local rules, judge-specific standing orders, and any deposition summary format your supervising attorney prefers. Claude will reference these when generating content.

Step 4: Start every session inside this Project. Your rules and practice profile load automatically with every conversation. Do not start litigation work in Claude's default chat.

Step 5: Save your reusable prompt library. After you build a prompt that works well for a deposition summary or motion outline, paste it into a Notes file in the Project. Over time this becomes your personal litigation prompt library — reuse and refine rather than rebuild.

Five High-Leverage Workflows

1. Deposition Transcript Summary

Summarizing a multi-hundred-page deposition is the single most time-intensive paralegal task in active litigation. Claude structures a working summary from the key material you provide.

<context>
Deponent: [NAME/ROLE], e.g., Plaintiff's damages expert. Deposition date: [DATE].
Case theory: [PLAINTIFF] claims [DEFENDANT] breached a supply contract causing $4.2M in lost profits.
Paste or attach the transcript excerpt(s) you want summarized.
</context>

<instructions>
- Extract key admissions against interest with page/line citations (e.g., 47:12–48:3)
- Build a topical index: liability, damages, credibility, prior knowledge, causation
- Flag direct contradictions between deposition testimony and prior sworn statements or interrogatory responses
- Identify impeachable statements (bias, inconsistency, lack of personal knowledge)
- Summarize each topic in 2–4 sentences; do not paraphrase away the admission
</instructions>

<format>
Four sections: (1) Key Admissions table with page/line and topic column, (2) Topical Index with subheadings, (3) Contradictions and Impeachment flags, (4) Suggested follow-up questions for the supervising attorney's review.
</format>

<avoid>
Legal conclusions or outcome predictions; fabricated page citations; summarizing testimony not in the provided excerpt; omitting the DRAFT disclaimer.
</avoid>

Before Claude: 3–5 hours for a 300-page deposition transcript. After Claude: 30 minutes to input key excerpts, 45 minutes to review and annotate.

2. Case Timeline from Filings and Discovery

A defensible case chronology is the backbone of every motion and trial outline. Claude builds it from the document set you feed it.

<context>
Case: [CLIENT] v. [OPPONENT], [CASE NO. PLACEHOLDER].
Documents available: complaints, answers, key discovery documents, deposition excerpts, and correspondence. Paste or attach the relevant passages and indicate the document source for each.
Disputed issues: [e.g., when defendant had notice of the defect; whether plaintiff mitigated damages]
</context>

<instructions>
- Build a chronological timeline of key events with: date, event description, source document, and legal significance
- Flag disputed dates or facts with [DISPUTED — see [SOURCE]]
- Separate undisputed facts from contested facts
- Cross-reference events to the claims or defenses they support (e.g., "Supports FRE 803(6) foundation for Exhibit 12")
- Note any gaps in the documentary record that discovery may need to fill
</instructions>

<format>
Markdown table: Date | Event | Source | Legal Significance | Status (Undisputed / Disputed). Follow with a Gap Analysis paragraph.
</format>

<avoid>
Inserting dates or events not supported by the provided documents; legal conclusions on disputed facts; fabricating exhibit numbers.
</avoid>

Before Claude: 4–6 hours building a clean chronology from a disorganized production. After Claude: 1 hour to input key documents, 30 minutes to verify and expand.

3. Motion Draft (MTD, MSJ, or Motion in Limine)

Motion drafting is the highest-leverage use of Claude in litigation — and the one requiring the most attorney oversight. Claude generates the framework and argument structure; the attorney writes the final version.

<context>
Motion type: [Motion to Dismiss under FRCP 12(b)(6) / Motion for Summary Judgment under FRCP 56 / Motion in Limine re: [TOPIC]]
Moving party: [PLAINTIFF / DEFENDANT]
Core argument: [e.g., Complaint fails to plead sufficient facts to state a breach-of-contract claim under the Twombly/Iqbal pleading standard]
Key facts: [Summarize the 3–5 facts most favorable to the moving party]
Jurisdiction: [Federal — District of [STATE] / State court — [JURISDICTION]]
</context>

<instructions>
- Draft the Argument section with FRCP rule citations (e.g., FRCP 12(b)(6), FRCP 56(a)) and the applicable legal standard
- For FRCP 12(b)(6): apply Twombly/Iqbal — plausibility pleading; cite Bell Atl. Corp. v. Twombly and Ashcroft v. Iqbal with [CITATION NEEDED] placeholders
- For FRCP 56: state the no-genuine-dispute standard; structure as: Legal Standard → Undisputed Facts → Application
- Leave all case citations as [CITATION NEEDED] with a description: e.g., "[CITATION NEEDED: circuit precedent holding that conclusory allegations are insufficient under Iqbal]"
- Do not fabricate any reporter citations, docket numbers, or holdings
</instructions>

<format>
I. Legal Standard (with FRCP rule numbers). II. Argument (subheadings per issue). III. Conclusion. Append a Citation Checklist listing every [CITATION NEEDED] slot for attorney to fill.
</format>

<avoid>
Fabricated or hallucinated case citations; omitting [CITATION NEEDED] placeholders; predicting the court's ruling; including FRE rules not relevant to this motion type.
</avoid>

Before Claude: 3–4 hours building a motion structure from scratch. After Claude: 45 minutes to input key facts, 2 hours for attorney review and citation completion.

4. Privilege and Litigation-Hold Review

Document review for privilege and litigation-hold scope is systematic but exhausting. Claude helps draft privilege log entries and evaluate hold scope from the criteria you define.

<context>
Review type: [Privilege log drafting / Litigation hold scope assessment]
Applicable privileges: Attorney-client privilege; Work product doctrine (FRCP 26(b)(3))
Litigation hold trigger: [e.g., Complaint filed [DATE]; Anticipated litigation as of [DATE]]
Key custodians: [List roles, not real names — e.g., General Counsel, Outside Litigation Counsel, VP Operations]
Document description (no actual content): [e.g., Email chain, [DATE RANGE], between GC and outside counsel re: litigation strategy]
</context>

<instructions>
- For privilege log: draft a log entry with: Bates range placeholder, date, author, recipient, document type, privilege claimed, basis
- Apply FRCP 26(b)(3) work-product analysis: distinguish ordinary vs. opinion work product; note dual-purpose document risk
- For litigation hold: identify custodian categories and data sources within scope based on the claims and defenses at issue
- Flag spoliation risk under FRCP 37(e): identify any data sources where auto-delete may be running
- Note any common-interest privilege or joint-defense considerations if relevant
</instructions>

<format>
Privilege log entry in standard column format. Followed by a Hold Scope memo: Custodians | Data Sources | Preservation Risk flags.
</format>

<avoid>
Reviewing actual document text that may contain confidential content; legal conclusions on whether specific documents are privileged — that determination belongs to the attorney; real custodian names.
</avoid>

Before Claude: 6–8 hours drafting a privilege log for a mid-size production. After Claude: 2 hours to build log entries for each document category, attorney QC to apply legal judgment.

5. Witness Exam Outline

A cross-examination or direct-exam outline built from the case timeline and key deposition admissions is the most concrete trial-prep deliverable a paralegal produces. Claude structures it from your materials.

<context>
Witness: [ROLE — e.g., Defendant's CFO]. Exam type: [Cross / Direct].
Key facts this witness owns: [List 3–5 — e.g., signed the contract, approved the budget, received the notice letter]
Deposition admissions to build from: [Paste key page/line citations and quotes from Workflow 1 summary]
Case theory: [One sentence — what the attorney needs this witness to establish or undermine]
</context>

<instructions>
- Build a topic-by-topic outline: each topic gets a one-line goal, 3–5 leading questions (cross) or open questions (direct), and the supporting deposition cite or exhibit reference
- For cross: questions should be short, leading, and confined to one fact — "You signed this contract on [DATE], correct?"
- Flag impeachment opportunities by topic: where deposition testimony conflicts with documentary evidence
- Include a "box and commit" sequence for any key admission the attorney wants locked in
- Reference applicable FRE rules where relevant: FRE 611 (leading questions on cross), FRE 607 (impeachment), FRE 613 (prior inconsistent statements)
</instructions>

<format>
Numbered topics as H3 subheadings. Under each: Goal (one line) | Questions (numbered) | Supporting cite | FRE note if applicable. End with an Impeachment Summary table.
</format>

<avoid>
Open-ended questions on a cross-examination outline; fabricating deposition quotes or page citations not in the provided material; predicting witness answers.
</avoid>

Before Claude: 4–6 hours building a structured witness outline from raw deposition transcripts. After Claude: 1 hour to input sources, 1 hour for attorney review and sequencing.

What This Looks Like in Your Week

Monday — Discovery cutoff in 10 days. You feed Claude the deposition transcript from Friday's deposition and get a working admissions table and topical index before the morning team meeting. The supervising attorney gets a briefing memo instead of a transcript.

Tuesday — Opposing counsel produces 4,000 documents. You use the privilege-log workflow to draft log entries for the first batch of obviously privileged communications before noon. Attorney QC takes 45 minutes instead of all afternoon.

Wednesday — MSJ deadline is two weeks out. You use the motion-drafting workflow to generate the Argument section skeleton with legal standards and [CITATION NEEDED] placeholders. Attorney fills citations rather than writing from scratch.

Thursday — Witness prep for next week. You build a cross-examination outline from the case timeline and deposition summary you already have in Claude. The attorney walks in with a ready outline, not a blank page.

Friday — New matter intake. You set up a new Claude Project for the incoming case, upload the complaint and initial discovery plan, and set the case-specific instructions. Next week's work starts from a configured baseline.

What to Avoid

Trusting citation output. Claude will generate plausible-looking case names, reporter volumes, and page numbers that do not exist. Every citation must read as [CITATION NEEDED] until you verify it against Westlaw or Lexis. Build this into your review checklist, not your trust.

Inputting client-identifying information. Real client names, case numbers, confidential settlement figures, and attorney mental impressions are never appropriate inputs. Use placeholders. Work product doctrine does not protect against a data breach you created.

Skipping attorney review on anything outward-facing. Court filings, privilege log productions, and discovery responses all require attorney sign-off. Claude produces drafts. The attorney produces the legal work product.

Using the default chat instead of your Project. Every conversation outside your configured Project loses your rules, your templates, and your citation guardrails. The rules exist because they prevent problems — use them every time.

Treating the output as research. Claude synthesizes patterns from its training data; it does not search Westlaw in real time. Legal standards change. New circuit opinions can shift the analysis. Claude-generated legal standards must be verified against current authority before any filing.

Resources

Explore the Litigation Paralegal AI plugin for ready-to-use prompt workflows tuned to litigation support
Browse the Paralegal profession hub for AI guides across transactional and litigation paralegal roles
Run an AI readiness audit for paralegals to see where AI fits in your specific practice area
Review FRCP Rules 12, 26, 37, and 56 directly at uscourts.gov — Claude references these by number, but your court's local rules and any amendments control