document-summary

name	document-summary
description	Document summarization and interpretation — long document distillation, multi-level summaries (one-line/paragraph/detailed), key information extraction.

Document Summarization and Interpretation

When the user provides a document (PDF, article, report, contract, etc.) and asks for a summary or interpretation, follow this workflow:

1. Read and understand

For local files, use read to access them directly. For large or complex files (e.g., parsing Excel, extracting PDF tables), use write + bash to write a Python script for processing.

First pass: Skim

• Title, table of contents, section headings
• Charts and tables
• Abstract/conclusion sections (if present)
• Build a mental model of the document's structure

Second pass: Deep read

• Core arguments and key data
• Topic sentence of each section
• Author's position and recommendations
• Technical terms and key concepts

2. Summary levels

Provide different depths based on what the user needs:

One-line summary

• Capture the document's core message in a single sentence
• Format: [document topic] + [core finding/conclusion]

Paragraph summary (100-300 words)

• 3-5 sentences covering:
  • Document topic and purpose
  • Core findings (2-3)
  • Main conclusion or recommendation

Detailed summary (500-1000 words)

• Organized following the original document's structure
• Key points from each major section
• Preserve critical data and citations
• Include the author's analysis and recommendations

Structured summary

• Use headings, bullet points, and tables to organize information
• Best for documents that need to be quickly searchable

3. Key information extraction

Focus on different elements depending on document type:

Research reports / White papers

• Core findings and data
• Market size / growth rates
• Key trends
• Recommendations and forecasts

News articles

• 5W1H (Who/What/When/Where/Why/How)
• Core event and impact
• Reactions and commentary from stakeholders

Business contracts / Legal documents

• Parties involved
• Core terms and obligations
• Amounts and timelines
• Special clauses and risk points

Technical documentation

• Core features/capabilities
• Prerequisites and limitations
• Key parameters and metrics
• Important caveats

Academic papers

• Research question and hypotheses
• Methodology
• Core findings
• Limitations and future directions

4. Interpretation and analysis

Beyond summarization, the user may want:

• Simplification: Explain technical content in plain language
• Critical analysis: Identify logical gaps, data issues, or bias
• Comparative analysis: Compare with other related documents/viewpoints
• Practical advice: Suggest actions based on the document's content
• Q&A: Answer specific questions about the document

5. Output format

• Use Markdown formatting
• Bold key data and conclusions
• Use > blockquote format for direct citations from the original
• Use tables for large amounts of data
• Start the summary with document metadata:
  • Document title
  • Author / source
  • Date
  • Page count / word count

6. Quality checklist

• Does the summary cover the document's core information?
• Does it accurately reflect the original's stance and viewpoint?
• Are personal judgments clearly labeled as such?
• Are key data citations accurate?
• Does the summary length match the user's request?
• Could someone understand the document's gist from the summary alone?