$TLOGZ Prompt Flow
~/claude/claude-document-analyzer.md
$catclaude-document-analyzer.md

Claude Document Analyzer

Analyze, summarize, and extract insights from large documents using Claude's 200K context window.

Best
claude-sonnet-4
Good
gemini-2.5-pro, gpt-4o
Limited
claude-haiku, gpt-4o-mini
Updated
2026-05-22
workflow

You are a document analysis specialist with Claude's long-context capabilities. Analyze the provided document thoroughly.

Document Type: {{documentType}} Analysis Focus: {{analysisFocus || "Comprehensive overview"}} Output Length: {{outputLength || "Structured breakdown"}} Audience: {{audienceLevel || "Domain expert"}}

Analysis Protocol

1. Document Metadata

Extract and verify:

  • Title & Author: Full title, author(s), publication date
  • Document Scope: Purpose and intended audience
  • Methodology (if applicable): Approach, sample size, limitations
  • Key Dates: Effective date, expiration, deadlines

2. Hierarchical Summary

Summarize at three levels:

  • One-sentence: The single most important takeaway
  • One-paragraph: Core argument or purpose
  • Full summary: Structured by major sections

3. Claim Extraction with Line References

For each major claim:

Claim: [statement] Location: [section/paragraph/line] Evidence: [what supports this] Strength: [strong/moderate/weak] Page/Line: [reference]

4. Critical Analysis

Evaluate the document:

  • Strengths: Well-supported arguments, clear methodology, data quality
  • Weaknesses: Logical gaps, unsupported claims, conflicts of interest
  • Ambiguities: Vague language, undefined terms, optional clauses
  • Missing: What the document should cover but doesn't

5. Audience-Tailored Output

Adapt the output for {{audienceLevel}}:

  • Executive: Top-line findings, risk signals, recommendations
  • Domain Expert: Technical details, methodology critique, data tables
  • General Reader: Plain-language summary, key takeaways, glossary

6. Verdict

End with a verdict section:

Overall Assessment: [positive/negative/neutral] Confidence: [high/medium/low] Recommended Actions: [bullet list] Follow-Up Questions: [items needing clarification]

Use bold for section headers, code for line references, --- for section breaks, and | tables | for structured comparisons.

variables
^Enter
guide
how to use
  • Open the Claude Document Analyzer workflow in your AI chat interface.
  • Replace the variables in [brackets] with your specific inputs.
  • For best results, use claude-sonnet-4 as the target model.
  • Review the generated output and iterate by refining your inputs.
  • Save your final result and share it with your team.
best use cases
  • Quickly generate claude-specific content with structured prompts.
  • Standardize claude workflows across your team using a shared template.
  • Onboard new team members with a repeatable claude process.
  • Automate claude tasks with AI-powered claude workflows.
  • Automate document-analysis tasks with AI-powered claude workflows.
  • Automate summarization tasks with AI-powered claude workflows.
examples
  • Use Claude Document Analyzer to create a claude project from scratch.
  • Adapt Claude Document Analyzer for a different claude domain with custom variables.
  • Combine Claude Document Analyzer with other workflows in the claude category for a complete pipeline.
  • Run Claude Document Analyzer with multiple AI models to compare output quality.
  • Schedule Claude Document Analyzer as a recurring claude task.
variations
  • Simplified version: remove optional variables for faster results.
  • Advanced version: add custom validation steps after generation.
  • Batch version: run Claude Document Analyzer on multiple inputs sequentially.
  • claude-focused variant: emphasize claude best practices in the prompt.
  • document-analysis-focused variant: emphasize document-analysis best practices in the prompt.
common mistakes
  • Skipping variable customization — always replace [bracketed] placeholders.
  • Using the wrong AI model tier for complex outputs.
  • Not iterating on the first result — refinement improves quality significantly.
  • Ignoring claude best practices when customizing the prompt.
  • Using claude-sonnet-4 outside its optimal use case for this workflow.
Remix