Claude Document Analysis: Legal, Research & Technical Docs

Master Claude for document analysis and summarization. Prompts for legal documents, research papers, transcripts, and technical documentation — leveraging Claude's long context and nuanced analysis.

January 14, 2026
ClaudeDocument AnalysisSummarizationLegalResearch

Claude's combination of 200K context, nuanced comprehension, and honest uncertainty makes it the strongest model for document analysis. It can read an entire contract, research paper, or technical specification and provide analysis that accounts for cross-document relationships and subtle implications — not just surface-level summaries.

The prompts below are domain-specific and leverage Claude's particular strengths: thoroughness without fabrication, identification of what's missing (not just what's present), and the ability to say "this document doesn't specify" rather than guessing.

Contract Review

Analyze this contract as a legal reviewer. Focus on:

1. UNUSUAL CLAUSES: Identify any clauses that deviate from standard industry practice.
   For each, explain what's unusual and the practical risk.

2. MISSING PROTECTIONS: What standard protections are absent?
   - Limitation of liability
   - Indemnification
   - Termination rights
   - Data protection / confidentiality
   - Dispute resolution

3. AMBIGUOUS LANGUAGE: Flag any language that could be interpreted multiple ways.
   Quote the ambiguous passage and explain the possible interpretations.

4. OBLIGATION SUMMARY: Create a table:
   | Party | Obligation | Deadline/Trigger | Penalty for Non-Compliance |
   |---|---|---|---|

5. RISK ASSESSMENT: Rate each section Low/Medium/High risk with 1-sentence justification.

6. RECOMMENDED CHANGES: For each issue found, propose specific revised language.

If the contract references external documents not provided, flag what's missing
and explain why it matters.

Multi-Document Comparison

Compare these [N] contracts/documents for consistency.

For each key provision (liability, termination, payment, IP, confidentiality),
create a comparison table:

| Provision | Doc 1 | Doc 2 | Doc 3 | Inconsistent? |
|---|---|---|---|---|

For inconsistencies:
- Explain the practical implication of each difference
- Recommend which version is most favorable to [party]
- Flag any conflicts that would make the documents incompatible if signed together

Research Paper Analysis

Literature Review Synthesis

Analyze these [N] papers on [topic]. Synthesize:

1. CONSENSUS FINDINGS: What do all/most papers agree on?
2. CONTRADICTIONS: Where do papers disagree? For each disagreement:
   - What's the specific point of disagreement?
   - What evidence does each side cite?
   - Is it a factual dispute or a methodological difference?
3. METHODOLOGY COMPARISON: Create a table comparing:
   | Paper | Method | Sample Size | Key Limitation (stated by authors) | Limitation (not stated but apparent) |
4. GAPS: What questions does none of these papers address?
5. EVOLUTION: If papers span multiple years, trace how understanding evolved.

For each finding, cite the specific paper and section.

Critical Review

Critically review this research paper:

1. THESIS: State the paper's central claim in one sentence.

2. EVIDENCE ASSESSMENT:
   - Strongest evidence supporting the thesis
   - Weakest evidence — what makes it weak?
   - Missing evidence: what SHOULD the authors have measured/included?

3. METHODOLOGY CRITIQUE:
   - Is the sample size adequate? (Compare to typical sizes in this field)
   - Are there confounding variables the authors didn't control for?
   - Would the methodology detect a false result? (False positive risk)

4. INTERPRETATION:
   - Do the RESULTS actually support the CONCLUSIONS?
   - Are there alternative explanations for the findings that the authors don't discuss?

5. IMPACT ASSESSMENT:
   - If true, how significant is this finding to the field?
   - What would it take to overturn this finding?

Technical Documentation

Specification Review

Review this technical specification for completeness and clarity:

1. MISSING SPECIFICATIONS: What should be specified but isn't?
   Check for:
   - Error handling (what happens when things go wrong?)
   - Edge cases (empty input, maximum values, concurrent operations)
   - Performance requirements (latency, throughput targets)
   - Security considerations
   - Backward compatibility

2. AMBIGUITIES: Flag any specification that could be implemented in multiple ways.
   "The system shall be fast" → Ambiguous. Suggest: "The system shall respond
   within 200ms for the 95th percentile."

3. CONTRADICTIONS: "Section 3 requires X but Section 7 implies not-X"
   (Claude's long context makes this cross-section contradiction detection powerful.)

4. TESTABILITY: For each requirement, state whether it's testable.
   Untestable: "The system shall be user-friendly"
   Testable: "New users shall complete onboarding in under 5 minutes"

5. IMPLEMENTATION CONCERNS: Flag any requirement that would be unusually
   expensive, complex, or risky to implement.

Transcript Analysis

Meeting/Interview Analysis

Analyze this transcript for:

1. KEY DECISIONS: List every decision made, who made it, and the stated rationale.

2. ACTION ITEMS: Extract all action items with:
   - Task
   - Owner (if specified)
   - Deadline (if specified)
   - Dependencies mentioned

3. DISAGREEMENTS: Identify points of disagreement:
   - Topic
   - Positions taken
   - Was it resolved? If so, how?

4. UNSPOKEN TENSIONS: Are there moments where someone seems to hold back,
   change the subject abruptly, or use hedging language? Flag these with timestamps.

5. FOLLOW-UP QUESTIONS: What questions were raised but left unanswered?

6. SENTIMENT SUMMARY: Overall tone — constructive, confrontational, uncertain, decisive?

Note:

Pro Move: For recurring document types (monthly reports, same-format contracts), save your analysis prompt as a template. Run it every time. This gives you consistent, comparable analysis across documents — and Claude's compliance with structured output formats makes this reliable.

Note:

Limitation awareness: Claude cannot verify factual claims in documents — it can only analyze what's present. A contract might reference regulations that don't exist; Claude won't know. Always verify legal, financial, and medical document analysis with domain experts.

  • Data Extraction & Structuring — Turn analyzed documents into structured JSON, CSV, or tables with Claude's strong output formatting compliance.
  • Academic Research Assistant — Specialized document analysis patterns for academic papers, literature reviews, and methodology design.