Academic Reading Workflow

Purpose

Execute systematic reading of academic papers, books, and complex texts using Blue's (OSP) 3-phase methodology: summary-first reading, active annotation with searchable keyword system, and evidence-based writing.

When to Use This Skill

Use this skill when:

• ✅ Reading academic papers or dense books requiring deep understanding
• ✅ Building searchable knowledge base from readings
• ✅ Need to retain and find information later ("command-F in real life")
• ✅ Preparing to write evidence-based essays/analyses with citations

Do NOT use for:

• ❌ Quick skimming (<30 min)
• ❌ Casual reading without note-taking
• ❌ Fiction/entertainment reading
• ❌ Already familiar material (just creating citations)

Decision Tree: See references/decision-tree.md

Quick Reference

Step	Agent	Deliverable	Duration	Quality Gate
0	researcher	Master keyword list (if multi-source project)	5-10 min	Keyword vocabulary defined
1	researcher	Reading roadmap with critical sections identified	15-30 min	Clear thesis + sections
2	researcher	20-50 searchable annotations with keyword tags	1-4 hours	≥20 notes, ≥5 keywords
3	analyst	Validated annotation set + keyword index	15-30 min	Searchable, <30% quote-paraphrases

Optional: Use evidence-based-writing skill separately when ready to write (not part of this workflow)

---

Agent Coordination Protocol

Sequential Execution

Each step passes deliverables to next step. Do NOT proceed if Quality Gate fails.

Agent Roles

• researcher: Roadmap creation, reading, annotation (Steps 0, 1, 2)
• analyst: Validation, quality checks, keyword standardization (Step 3)

Annotation Storage Format

All annotations stored as Markdown with YAML frontmatter:

---
source: "[Title] - [Author] ([Year])"
page: [number]
keywords: [keyword1, keyword2, keyword3]
date_annotated: [YYYY-MM-DD]
project: [research-topic-slug]
annotation_id: [unique-id]
---

**Summary**: [Your paraphrase in own words]

**Quote** (if applicable): "[Exact text]" (p. [X])

**Why This Matters**: [Connection to research question]

**Links**: See also [Page Y], Conflicts with [Source B]

Memory MCP Tags

Store with: WHO=[agent], WHEN=[timestamp], PROJECT=[topic], WHY=annotation, SOURCE=[title], PAGE=[number]

---

Blue's Core Principles

This workflow embeds Blue's (OSP) methodology:

Principle	Implementation
"Read the Roadmap Before You Get Lost"	Step 1: Summary-first, create plan BEFORE deep reading
"Annotation is Command-F in Real Life"	Step 2: Keyword tagging for searchable notes
"Paraphrase > Highlighting"	Step 2: Force genuine paraphrase, not quote-rewording
"Write Like You Speak"	(Evidence-based-writing skill): Natural draft, polish later
"Thesis Comes LAST"	(Evidence-based-writing skill): Let thesis emerge from notes
"Every Claim Needs Source"	(Evidence-based-writing skill): All assertions cited with pages

See references/blue-methodology.md for full explanation.

---

Step-by-Step Workflow

STEP 0: Initialize Master Keyword List (Multi-Source Projects)

Agent: researcher Goal: Define consistent keyword vocabulary across all sources in project

When to Use:

• ✅ Reading 3+ sources for same research project
• ✅ Building cross-source knowledge base
• ❌ Skip if reading single source

Procedure:

1. List main topics/concepts in your research project
2. Define standard keywords for each:
   • Use domain-standard terms when possible
   • Be specific (#methodology, not #method)
   • Use consistent formatting (#snake-case)
3. Create master keyword list:

# MASTER KEYWORD LIST: [Project Name]

## Core Concepts
- #[concept-1] - Definition/scope
- #[concept-2] - Definition/scope

## Methodology
- #methodology - Research methods discussed
- #data-collection - Data gathering approaches
- #analysis - Analytical techniques

## Key Themes
- #[theme-1]
- #[theme-2]

## Authors/Schools
- #[author-name] - When referencing this scholar
- #[school-of-thought] - When discussing this approach

4. Store in Memory MCP with PROJECT tag
5. Use SAME keywords across ALL sources in this project

Deliverable: Master keyword list (10-20 keywords)

Quality Gate 0: Keyword vocabulary defined, specific (not vague)

---

STEP 1: Summary-First Reading (Roadmap Phase)

Agent: researcher Objective: Create reading roadmap BEFORE deep dive to avoid getting lost

Procedure:

A. Read Summary Materials FIRST (priority order):

1. Abstract (for papers)
2. Introduction + Conclusion chapters (for books)
3. Table of Contents
4. Section headers
5. Wikipedia article on topic (if exists)
6. Existing reviews/summaries

B. Create Reading Roadmap:

# READING ROADMAP: [Source Title]

## Main Argument/Thesis
[1-2 sentences - or "No clear thesis, exploratory paper" if applicable]

## Key Questions (if no thesis)
- Question 1
- Question 2

## Critical Sections (Read Carefully + Annotate Heavily)
1. [Chapter/Section] (pages X-Y) - **Why critical**: [Reason]
2. [Chapter/Section] (pages A-B) - **Why critical**: [Reason]

## Supplementary Sections (Skim for Key Points)
1. [Chapter/Section] (pages M-N) - **Extract**: [What to get]

## Skip Sections
1. [Chapter/Section] - **Why skipping**: [Reason]

## Reading Focus Question
[What am I trying to learn from this source?]

## Estimated Reading Time
- Critical sections: [X] hours
- Supplementary: [Y] hours
- **Total**: [Z] hours

C. Handle Edge Cases:

• No clear thesis: List "key questions" paper explores instead
• Too technical/unfamiliar: Add "Define unfamiliar terms" sub-step
  • Create glossary section for domain-specific terms
  • Define before attempting to paraphrase

Deliverable: Reading roadmap with sections categorized

Quality Gate 1:

• GO: Main argument OR key questions identified, critical sections listed with page ranges
• NO-GO: Vague roadmap → Re-read intro/conclusion for clarity

---

STEP 2: Deep Reading + Active Annotation

Agent: researcher Objective: Read with searchable margin notes using keyword tagging system

Procedure:

A. Read Actively (Not Passively)

• Follow reading plan from Step 1
• Read with annotation tools ready
• Pause after each paragraph: "What's the point?"
• Don't just highlight—ANNOTATE with your thoughts

B. Create Searchable Annotations

Use this template for EVERY annotation:

---
source: "[Title] - [Author] ([Year])"
page: [X]
keywords: [keyword1, keyword2, keyword3]
date_annotated: [YYYY-MM-DD]
project: [project-slug]
annotation_id: [source-slug]-p[page]
---

## ✅ SUMMARY [REQUIRED - Min 1 sentence]
[Your paraphrase in YOUR words - force yourself to rephrase, not just reword]

## ⚠️ QUOTE [OPTIONAL - Only if exact wording matters]
"[Exact text]" (p. [X])

## ✅ KEYWORDS [REQUIRED - Min 2, use master list if multi-source]
#keyword1 #keyword2 #keyword3

## ⚠️ WHY THIS MATTERS [OPTIONAL BUT RECOMMENDED]
**Research Question**: How does this address my focus?
**Argument Structure**: Is this claim/evidence/counter-evidence?
**Cross-Reference**: Links to [Page Y], [Source B, p. Z]

## ⚠️ DEFINE TERMS [IF UNFAMILIAR DOMAIN]
- **Term 1**: [Definition]
- **Term 2**: [Definition]

Example: See examples/annotation-example.md

C. Annotation Principles (Blue's Rules)

✅ DO:

• Force paraphrase in YOUR words (if you can't paraphrase, you don't understand—re-read)
• Tag with 2-5 keywords for searchability
• Include page numbers for ALL quotes and claims
• Link related passages ("See also page 42", "Conflicts with Source B, p. 15")
• Write for future you (enough context to understand 6 months later)

❌ DON'T (Anti-Patterns):

• ❌ "Important!" (too vague, not searchable)
• ❌ Copy-paste with slight rewording (not genuine paraphrase)
• ❌ Keywords like "#page15" (not conceptually searchable)
• ❌ Highlighting without notes (doesn't create understanding)

Example: See examples/good-vs-bad-paraphrase.md

D. Annotation Levels by Section

Critical Sections (from Step 1):

• Annotate EVERY major claim
• Extract 2-5 direct quotes
• Create 5-10 annotations per section
• Heavy keyword tagging

Supplementary Sections:

• Annotate key points only
• 1-2 quotes if notable
• 2-3 annotations per section
• Light keyword tagging

E. Handle Special Cases

Long Books (100+ pages):

• Create "Summary Note" every 50 pages
• Recap main themes so far
• Prevent annotation overflow

Unfamiliar Domain:

• Define ALL technical terms inline
• Build glossary section as you read
• Only paraphrase AFTER understanding terms

No Clear Thesis (exploratory papers):

• Focus on "key questions" from Step 1
• Tag with question numbers: #question1, #question2

F. Store Annotations

Store each annotation in Memory MCP with tags:

npx claude-flow@alpha memory store \
  --key "annotations/[project]/[source-slug]/p[page]" \
  --value "[Markdown annotation with YAML frontmatter]" \
  --tags "WHO=researcher,WHEN=[timestamp],PROJECT=[topic],WHY=annotation,SOURCE=[title],PAGE=[page],KEYWORDS=[keyword1,keyword2]"

Deliverable: 20-50 searchable annotations (depending on source length)

Quality Gate 2:

• GO: ≥20 annotations for full paper/chapter, ≥5 keywords used, page numbers present
• NO-GO: <20 annotations → Extend reading time, annotate more thoroughly

---

STEP 3: Annotation Quality Check

Agent: analyst Objective: Validate annotations are searchable, useful, and complete

Procedure:

A. Completeness Check

For each annotation, verify:

• ✅ Has ≥2 keyword tags?
• ✅ Has page number?
• ✅ Has genuine paraphrase (own words)?
• ⚠️ If quote claimed, has exact text in "quotes"?
• ⚠️ If cross-reference claimed, has specific page/source link?

Flag incomplete → return to Step 2

B. Keyword Consistency Check

1. Extract all keywords used
2. Check for duplicates/synonyms:
   • #method vs #methodology → Standardize to one
   • #keypoint → Too vague, make specific
3. If multi-source project:
   • Compare to master keyword list (Step 0)
   • Flag deviations → update to match master list
4. Create keyword index:

# KEYWORD INDEX: [Source Title]

## Keywords Used (Alphabetical)
- #argument (pages 15, 42, 88) - [3 uses]
- #bias (pages 23, 67) - [2 uses]
- #evidence (pages 15, 28, 35, 49, 72) - [5 uses]
- #key-claim (pages 12, 34, 56) - [3 uses]
- #limitation (pages 89, 91) - [2 uses]
- #methodology (pages 8, 10, 45) - [3 uses]

**Total keywords**: 6
**Total annotations**: 32
**Average annotations per keyword**: 5.3

C. Paraphrase Quality Check

Sample 5-10 annotations randomly. For each:

• Is it in reader's OWN words (not just slightly reworded)?
• Does it capture essence without source?
• Understandable 6 months later?

If >30% are quote-paraphrases → return to Step 2, force genuine paraphrasing

D. Searchability Test

1. Pick 3 concepts from the source
2. Search annotations using keywords
3. Can you find ALL relevant passages quickly?
4. Are notes useful on their own?

Example:

• Search #methodology → Should find all passages about research methods
• Search #byzantine-trade → Should find all passages about Byzantine commerce

Quality Gate 3:

• GO: ≥20 annotations, ≥5 keywords, <30% quote-paraphrases, searchability works
• NO-GO: Return to Step 2 for improved annotation depth or keyword consistency

Deliverable:

• Validated annotation set
• Keyword index
• Quality assessment report

---

Success Metrics

Quantitative

• ✅ Reading roadmap created (Step 1)
• ✅ ≥20 annotations for full paper/chapter
• ✅ ≥5 consistent keywords used
• ✅ ≥2 keywords per annotation
• ✅ Page numbers for ALL quotes and claims
• ✅ <30% quote-paraphrases (genuine paraphrasing)
• ✅ Keyword index searchable

Qualitative

• ✅ Can find passages using keyword search
• ✅ Paraphrases understandable without source
• ✅ Annotations useful 6 months later
• ✅ Links between passages documented
• ✅ If multi-source: keywords consistent across all sources

---

Error Handling

Failure Mode	Gate	Resolution
Vague roadmap	1	Re-read abstract/intro, clarify argument OR list key questions
Getting lost while reading	1	Return to roadmap, refocus on critical sections
<20 annotations	2	Extend reading, annotate more thoroughly
<5 keywords	2	Review notes, add specific keywords
>30% quote-paraphrases	3	Force genuine paraphrasing, re-read if needed
Keyword inconsistency	3	Standardize terms, update master list
Can't find via keywords	3	Add more keywords, improve tagging
Unfamiliar domain	2	Define terms inline before paraphrasing
No clear thesis	1	Identify "key questions" instead
Annotation overflow (100+)	2	Create summary notes every 50 pages
Keyword drift (multi-source)	3	Update master keyword list, standardize

---

Integration

Before This Skill:

• Use general-research-workflow Steps 2-3 to find and classify sources first
• Prioritize which sources to read deeply using credibility/priority scores

During This Skill:

• Can annotate multiple sources in parallel
• Use SAME keyword vocabulary across all sources (Step 0 master list)
• Annotations feed into general-research-workflow Step 5 (note-taking)

After This Skill:

• Use evidence-based-writing skill when ready to write essay (separate invocation)
• Export keyword index to build personal knowledge base
• Search annotations across ALL sources using shared keywords

Companion Skill:

• evidence-based-writing (Step 4 from original SOP, now separate skill)
  • Use when: Ready to write essay/analysis based on annotations
  • Input: Validated annotations from Step 3
  • Output: Draft with citations, relativist language, evidence-based claims

---

Process Visualization

See academic-reading-process.dot for complete workflow diagram showing all steps, gates, and decision points.

---

Storage Format Specification

Annotation File Structure:

annotations/
  [project-slug]/
    [source-slug]/
      p001.md (annotation for page 1)
      p015.md (annotation for page 15)
      p042.md (annotation for page 42)
      keyword-index.md
      summary.md (if book >100 pages)

YAML Frontmatter Format:

---
source: "Byzantium and Renaissance - Wilson 1992"
page: 45
keywords: [greek-migration, manuscripts, bessarion]
date_annotated: 2025-01-06
project: byzantine-renaissance-italy
annotation_id: wilson1992-p45
type: annotation
---

Memory MCP Storage:

Key: annotations/[project]/[source]/p[page]
Value: [Full Markdown with YAML frontmatter]
Tags: WHO=researcher,WHEN=[ISO8601],PROJECT=[slug],WHY=annotation,SOURCE=[title],PAGE=[page],KEYWORDS=[csv]

---

Blue's Annotation Principles: "Read the Roadmap, Command-F in Real Life, Paraphrase > Highlighting, Write for Future You"