Lit Synthesis

Literature Synthesis

You help sociologists move from a corpus of papers to a deep understanding of a field. This is the analytical bridge between finding papers (lit-search) and writing about them (lit-writeup).

The Lit Trilogy

This skill is the middle step in a three-skill workflow:

Input: Papers in Zotero (imported from lit-search or user's existing library) Output: Organized understanding of the field ready for writing

When to Use This Skill

Use this skill when users:

• Have a corpus of papers (from lit-search or their own collection)
• Need to understand the theoretical landscape before writing
• Want to identify debates, tensions, and competing positions
• Need to organize papers thematically or by theoretical tradition
• Want deep reading notes, not just metadata extraction

Core Principles

1. Read deeply, not widely: Better to understand 15 papers thoroughly than 50 superficially.

2. Theoretical traditions matter: Papers exist within intellectual lineages. Map who cites whom and why.

3. Debates are gold: Competing positions create space for contributions. Find the tensions.

4. Organization serves writing: The clusters and maps you create should directly feed lit-writeup's architecture phase.

5. Full text when possible: Abstracts tell you what; full text tells you how and why.

Zotero MCP Integration

This skill uses Zotero MCP for accessing your library:

Setup

Install the Zotero MCP server:

uv tool install "git+https://github.com/54yyyu/zotero-mcp.git"
zotero-mcp setup

See mcp/zotero-setup.md for detailed configuration.

Key Capabilities

Workflow Integration

1. From lit-search: Import the BibTeX export into Zotero 2. Acquire PDFs: Use Zotero's "Find Available PDF" or manual download 3. Read and annotate: Highlight key passages, add notes 4. lit-synthesis reads: Access annotations via MCP for analysis

Workflow Phases

Phase 0: Corpus Audit

Goal: Assess what's in the corpus and identify gaps.

Process:

• Review the database from lit-search (or user's Zotero collection)
• Count papers by year, journal, author, theoretical tradition
• Identify potential gaps in coverage
• Prioritize which papers need deep reading vs. skimming

Output: corpus-audit.md with statistics and reading priorities.

Pause: User confirms corpus coverage and reading priorities.

---

Phase 1: Deep Reading

Goal: Close read priority papers and extract analytical insights.

Process:

• For each priority paper, read full text via Zotero MCP
• Extract: argument structure, theoretical framework, key concepts, methodological approach
• Note: how theory is deployed, what evidence supports claims, limitations acknowledged
• Create structured reading notes

Output: reading-notes/ directory with per-paper notes.

Pause: User reviews reading notes for key papers.

---

Phase 2: Theoretical Mapping

Goal: Identify intellectual traditions and lineages.

Process:

• Identify which theoretical frameworks appear across papers
• Map citation relationships (who cites whom)
• Note foundational texts and their descendants
• Identify "camps" or schools of thought
• Document key concepts and how they're used

Output: theoretical-map.md with traditions, key theorists, and concept definitions.

Pause: User reviews theoretical landscape.

---

Phase 3: Thematic Clustering

Goal: Organize papers by what they study and how.

Process:

• Group papers by empirical focus (population, setting, phenomenon)
• Group papers by theoretical approach
• Group papers by methodological strategy
• Identify papers that bridge multiple clusters
• Note within-cluster consensus and variation

Output: thematic-clusters.md with organized paper groupings.

Pause: User reviews clustering logic.

---

Phase 4: Debate Mapping

Goal: Identify tensions, disagreements, and competing positions.

Process:

• Find explicit disagreements (papers that critique each other)
• Find implicit tensions (contradictory findings or incompatible assumptions)
• Identify unresolved questions the field is grappling with
• Note where evidence is mixed or contested
• Document the "state of the debate" for each tension

Output: debate-map.md with positions, evidence, and unresolved questions.

Pause: User reviews debates and selects focus areas.

---

Phase 5: Field Synthesis

Goal: Create comprehensive understanding ready for writing.

Process:

• Synthesize across phases into coherent field understanding
• Identify the most productive gaps for contribution
• Recommend which lit-writeup cluster (Gap-Filler, Theory-Extender, etc.) fits
• Create the handoff document for lit-writeup

Output: field-synthesis.md with integrated understanding and writing recommendations.

---

Output Files

lit-synthesis/
├── corpus-audit.md           # Phase 0: What's in the corpus
├── reading-notes/            # Phase 1: Per-paper notes
│   ├── author2020-title.md
│   ├── author2019-title.md
│   └── ...
├── theoretical-map.md        # Phase 2: Traditions and lineages
├── thematic-clusters.md      # Phase 3: Paper groupings
├── debate-map.md             # Phase 4: Tensions and positions
└── field-synthesis.md        # Phase 5: Integrated understanding

Reading Note Template

For each paper in Phase 1:

# [Author Year] - [Short Title]

## Bibliographic Info
- Full citation: [from Zotero]
- DOI: [link]

## Core Argument
[1-2 sentences: What is the paper arguing?]

## Theoretical Framework
- Tradition: [e.g., Bourdieusian, institutionalist, interactionist]
- Key concepts used: [list]
- How theory is deployed: [description vs. extension vs. critique]

## Empirical Strategy
- Data: [what kind]
- Methods: [how analyzed]
- Sample: [who/what]

## Key Findings
1. [Finding 1]
2. [Finding 2]
3. [Finding 3]

## Contribution Claim
[What does the paper claim to contribute?]

## Limitations (as noted by authors)
- [Limitation 1]
- [Limitation 2]

## My Notes
[Your analytical observations, connections to other papers, questions raised]

## Key Quotes
> "[Quote 1]" (p. X)

> "[Quote 2]" (p. Y)

## Tags
[theoretical-tradition] [empirical-focus] [method] [relevant-to-my-project]

Model Recommendations

Starting the Synthesis

When the user is ready to begin:

1. Check Zotero setup:

"Do you have Zotero MCP configured? If not, let's set that up first (see mcp/zotero-setup.md)."

2. Identify the corpus:

"Where are your papers? A Zotero collection from lit-search? An existing library folder? How many papers total?"

3. Set priorities:

"Which papers are most central to your project? We'll deep-read those first and skim the rest."

4. Clarify goals:

"What are you trying to understand about this field? Are you looking for gaps, debates, or a specific theoretical tradition?"

5. Proceed with Phase 0 to audit the corpus.

Key Reminders

• Zotero is the source of truth: All papers should be in Zotero for consistent access
• Annotations accelerate: If you've already highlighted papers, those annotations are accessible via MCP
• Quality over quantity: Deep reading 15 papers beats skimming 50
• Debates are opportunities: Every tension you find is a potential contribution space
• This feeds lit-writeup: The outputs here become inputs there—keep that handoff in mind

Zotero MCP Setup Guide

This skill requires the Zotero MCP server to access your library, read PDFs, and retrieve annotations.

Prerequisites

1. Zotero 7+ installed on your machine 2. uv package manager installed 3. Papers imported into Zotero (from lit-search BibTeX or manually)

Installation

Step 1: Install the MCP Server

uv tool install "git+https://github.com/54yyyu/zotero-mcp.git"

Step 2: Run Setup

zotero-mcp setup

This auto-configures for Claude Desktop. Follow the prompts.

Step 3: Enable Zotero API Access

In Zotero: 1. Go to Edit → Preferences (Windows/Linux) or Zotero → Settings (Mac) 2. Navigate to Advanced → General 3. Enable "Allow other applications on this computer to communicate with Zotero"

Step 4: Restart Claude Desktop

After setup, restart Claude Desktop to load the MCP server.

Connection Options

Local API (Recommended)

Uses Zotero's local API for offline access to your desktop library.

Advantages:

• Works offline
• Faster access
• Full PDF access

Requirements:

• Zotero running on your machine
• "Allow other applications" enabled

Web API (Alternative)

Uses Zotero's web API for cloud-synced libraries.

Advantages:

• Access from anywhere
• Works without Zotero running

Requirements:

• Zotero API key (get from zotero.org/settings/keys)
• Library ID
• PDFs must be synced to cloud

Verifying Setup

After setup, test with:

Use the zotero_search_items tool to search for "test"

If working, you should see results from your library.

Available Tools

Once configured, you have access to:

Workflow Integration

From lit-search

1. Export BibTeX from lit-search's references.bib 2. In Zotero: File → Import → select the BibTeX file 3. Papers appear in your library (create a collection for the project) 4. Use "Find Available PDF" to auto-download open access versions 5. Manually acquire remaining PDFs via institutional access

Organizing for Analysis

Create a Zotero collection for your project: 1. Right-click My Library → New Collection 2. Name it (e.g., "housing-instability-lit-review") 3. Drag papers into the collection 4. Use tags for additional organization (e.g., "priority-read", "theoretical", "empirical")

Annotating PDFs

As you read in Zotero's PDF viewer:

• Highlight key passages (yellow for findings, blue for theory, green for methods)
• Add notes to highlights for your interpretation
• These annotations are accessible via zotero_get_annotations

Troubleshooting

"Zotero not found" Error

• Ensure Zotero is running
• Check that "Allow other applications" is enabled
• Restart both Zotero and Claude Desktop

"No results" for Known Papers

• Verify the paper is in your Zotero library (not just synced to cloud)
• Check your search terms (try author name or partial title)
• Use zotero_search_items with broader terms

Annotations Not Appearing

• Annotations must be made in Zotero's built-in PDF viewer
• External annotations (from Preview, Acrobat) won't sync
• Ensure the PDF is attached to the Zotero item (not just linked)

Web API Authentication Errors

• Verify your API key at zotero.org/settings/keys
• Check that the key has read access to your library
• Confirm your Library ID is correct

Resources

• Zotero MCP Repository: https://github.com/54yyyu/zotero-mcp
• Zotero Documentation: https://www.zotero.org/support/
• API Key Management: https://www.zotero.org/settings/keys

Phase 0: Corpus Audit

Why This Phase Matters

Before deep reading, you need to know what you have. A corpus audit reveals the shape of your literature: temporal coverage, theoretical breadth, methodological diversity, and gaps. This shapes your reading priorities and prevents wasted effort on peripheral papers.

---

Your Tasks

1. Assess Corpus Source

Determine where the papers come from:

From lit-search:

• Load the database.json from lit-search output
• Papers should already be screened and annotated
• Check the synthesis.md for initial thematic groupings

From existing Zotero library:

• Identify the relevant collection(s)
• May need initial screening if not pre-filtered

From user's ad-hoc collection:

• Gather all papers into one location
• May need more extensive organization

2. Generate Corpus Statistics

Using Zotero MCP, compile:

## Corpus Overview

- **Total papers**: [N]
- **Date range**: [earliest] - [latest]
- **Peak years**: [years with most papers]

## By Journal
| Journal | Count |
|---------|-------|
| [Journal 1] | [N] |
| [Journal 2] | [N] |
...

## By Author (top 10)
| Author | Papers |
|--------|--------|
| [Author 1] | [N] |
...

## By Decade
| Period | Count |
|--------|-------|
| 2020s | [N] |
| 2010s | [N] |
| 2000s | [N] |
| Pre-2000 | [N] |

3. Preliminary Theoretical Scan

Skim abstracts and keywords to identify:

• Which theoretical traditions appear? (e.g., Bourdieusian, institutionalist, interactionist)
• Any named frameworks? (e.g., "legal consciousness," "gendered organizations")
• Dominant methodological approaches?

4. Identify Coverage Gaps

Based on statistics and skim:

• Temporal gaps: Missing recent work? Missing foundational papers?
• Theoretical gaps: One tradition over-represented?
• Empirical gaps: Missing key populations or settings?
• Geographic gaps: All US-based? Missing comparative work?

5. Set Reading Priorities

Categorize papers into:

Use these criteria for "Must Read":

• Highly cited within the corpus
• Directly addresses your research question
• Foundational text others cite
• Introduces key concepts you'll use
• Recent and relevant (last 3-5 years)

6. Write Corpus Audit Memo

Create corpus-audit.md:

# Corpus Audit

## Corpus Summary
- Source: [lit-search / Zotero collection / ad-hoc]
- Total papers: [N]
- Date range: [years]

## Statistics
[Tables from Task 2]

## Preliminary Theoretical Landscape
- Dominant traditions: [list]
- Named frameworks: [list]
- Methodological approaches: [list]

## Coverage Assessment

### Strengths
- [What's well covered]

### Gaps Identified
- [Gap 1]: [description and significance]
- [Gap 2]: [description and significance]

### Recommendations
- [Any papers to add before proceeding]
- [Any papers to exclude]

## Reading Priorities

### Must Read (Deep Reading - Phase 1)
1. [Author Year] - [Reason]
2. [Author Year] - [Reason]
3. [Author Year] - [Reason]
... [aim for 10-20 papers]

### Should Read (Careful Skim)
- [List]

### May Skim
- [List or "remainder"]

## Questions for User
- [Any clarifying questions]

---

Guiding Principles

15-20 Deep Reads Is Plenty

You can't deeply read 80 papers. Prioritize ruthlessly. The "Must Read" list should be 10-20 papers maximum.

Foundational > Recent

Don't skip classics just because they're old. Foundational texts shape how everyone else thinks about the topic.

Gaps Are Data

Finding that your corpus lacks X is valuable information—it may reveal your contribution space.

User Knows Their Field

You can suggest priorities, but the user knows which papers matter most for their specific project.

---

Output Files to Create

1. corpus-audit.md - Full audit with statistics and priorities

---

When You're Done

Report to the orchestrator:

• Corpus size and date range
• Number of papers prioritized for deep reading
• Key gaps identified
• Any recommendations before proceeding

Example summary:

"Corpus audit complete. 47 papers spanning 2008-2024, concentrated in Social Problems and American Sociological Review. Theoretical landscape dominated by legal consciousness and neighborhood effects traditions. 15 papers prioritized for deep reading; notable gap in comparative/non-US work. Recommend adding 2-3 foundational neighborhood effects papers before Phase 1. Ready for user review."

Phase 1: Deep Reading

Why This Phase Matters

Deep reading is where understanding happens. Abstracts tell you what papers argue; full text reveals how they argue it—how theory is deployed, what evidence is marshaled, what assumptions underlie claims. This phase creates the raw material for theoretical mapping and debate identification.

---

Your Tasks

1. Prepare Reading Environment

For each priority paper:

• Confirm PDF is in Zotero
• Check for existing annotations (from user's prior reading)
• Note if paper has supplementary materials

Use Zotero MCP to access:

Use zotero_get_item_metadata for [paper identifier]
Use zotero_get_annotations for [paper identifier]

2. Read with Purpose

For each paper, read for:

Argument Structure

• What is the central claim?
• How is it structured (inductive, deductive, abductive)?
• What's the contribution claim?

Theoretical Framework

• Which tradition(s) does it draw on?
• Who are the key theorists cited?
• Is theory used descriptively, extended, or critiqued?

Conceptual Work

• What key concepts are used?
• Are any new concepts introduced?
• How are concepts operationalized?

Empirical Strategy

• What data? What methods?
• How is theory connected to evidence?
• What are the acknowledged limitations?

Position in Field

• Who does this paper cite approvingly?
• Who does it critique or position against?
• Where does it claim to contribute?

3. Create Reading Notes

For each paper, create a note file using the template in SKILL.md:

reading-notes/
├── author2020-short-title.md
├── author2019-short-title.md
└── ...

Essential sections:

• Core argument (1-2 sentences)
• Theoretical framework (tradition, key concepts)
• How theory is deployed (description vs. extension vs. critique)
• Key findings (bulleted)
• Contribution claim
• Key quotes (with page numbers)
• Your analytical notes

4. Flag Connections

As you read, note:

• Papers that cite each other
• Papers that contradict each other
• Papers using the same concepts differently
• Papers that seem to be in conversation

Add tags to notes:

• [cites: Author2019]
• [contradicts: Author2018]
• [same-concept-as: Author2020]

5. Extract Key Quotes

For each paper, capture 3-5 quotes that:

• Define key concepts
• State the argument clearly
• Could be used in your own writing
• Reveal assumptions or limitations

Always include page numbers.

6. Compile Reading Summary

After all priority papers are read, create a summary:

# Reading Notes Summary

## Papers Read
- [N] papers deep-read
- [Date range]

## Emergent Patterns

### Theoretical Traditions Identified
1. **[Tradition 1]**: [which papers, key ideas]
2. **[Tradition 2]**: [which papers, key ideas]

### Recurring Concepts
- **[Concept A]**: Used by [papers], defined as [summary]
- **[Concept B]**: Used by [papers], defined as [summary]

### Methodological Patterns
- [Dominant approach]
- [Variations]

### Preliminary Tensions Noted
- [Tension 1]: [Paper A] vs [Paper B] on [issue]
- [Tension 2]: [description]

## Papers That Stand Out
- **Most cited within corpus**: [papers]
- **Most theoretically ambitious**: [papers]
- **Most methodologically innovative**: [papers]
- **Most relevant to user's project**: [papers]

## Gaps Emerging
- [What's missing from the conversation]

## Questions for Next Phases
- [What to investigate in theoretical mapping]
- [What to investigate in debate mapping]

---

Using Zotero MCP Effectively

Retrieving Full Text

Use zotero_get_item_metadata with the item key to get full text if available

Getting Annotations

If user has already highlighted:

Use zotero_get_annotations for [item key]

This returns highlights and notes, accelerating your reading.

Semantic Search for Connections

Use zotero_semantic_search with a concept to find related papers

---

Guiding Principles

Read the Whole Paper

Don't just skim. For priority papers, read introduction, theory, methods, findings, AND conclusion. The conclusion often reveals assumptions.

Theory Sections Are Gold

Pay special attention to how authors construct their Theory sections—this feeds directly into lit-writeup.

Note How, Not Just What

Anyone can summarize findings. Note how the argument is constructed, how theory is deployed, how evidence is marshaled.

Quotes Need Context

A quote without page number is useless. A quote without noting why it matters is nearly as bad.

Your Notes Are for Future You

Write notes that will make sense when you return to them in Phase 3-4. Be explicit about connections and questions.

---

Output Files to Create

1. *reading-notes/.md - One file per paper 2. reading-summary.md** - Compiled patterns and observations

---

When You're Done

Report to the orchestrator:

• Number of papers deep-read
• Theoretical traditions identified
• Preliminary tensions noted
• Key papers for theoretical mapping

Example summary:

"Deep reading complete. 15 papers read in full. Three theoretical traditions identified: legal consciousness (6 papers), neighborhood effects (5 papers), recognition theory (4 papers). Preliminary tension noted between structural vs. cultural explanations for police-calling behavior. Kirk & Papachristos (2011) and Desmond (2016) emerge as most-cited anchors. Ready for Phase 2: Theoretical Mapping."

Phase 2: Theoretical Mapping

Why This Phase Matters

Papers don't exist in isolation—they belong to intellectual traditions, cite foundational texts, and position themselves within ongoing conversations. Theoretical mapping reveals this structure: who the key thinkers are, how concepts travel across papers, and which traditions are in dialogue or tension.

This map directly feeds lit-writeup's architecture phase by clarifying which literatures need engagement.

---

Your Tasks

1. Identify Theoretical Traditions

From your reading notes, cluster papers by theoretical home:

Questions to ask:

• Which foundational theorists do multiple papers cite? (Bourdieu, Goffman, Foucault, etc.)
• Are there named frameworks that recur? (legal consciousness, gendered organizations, etc.)
• Do papers self-identify with traditions? ("Drawing on symbolic interactionism...")

Common traditions in sociology:

• Bourdieusian (capital, field, habitus)
• Institutionalist (isomorphism, legitimacy)
• Interactionist (meaning-making, identity work)
• Critical/Conflict (power, inequality, domination)
• Cultural (schemas, repertoires, narratives)
• Organizational (structure, culture, agency)

2. Map Citation Networks

For each tradition identified:

Foundational texts:

• Which older works does everyone cite?
• These are the "must-cite" sources for your tradition

Key intermediaries:

• Which papers translated the foundational work to your empirical domain?
• Often these are the "applied the framework to [X]" papers

Recent developments:

• What's the cutting edge?
• Who's extending or revising the tradition?

Create a simple lineage:

[Foundational] → [Key Application] → [Recent Extension]
     ↓                  ↓                    ↓
Bourdieu 1984 → Lareau 2003 → Calarco 2018

3. Document Key Concepts

For each major concept:

## [Concept Name]

**Origin**: [Who introduced it? When?]

**Definition**: [How is it defined in the foundational text?]

**In This Literature**:
- [Author 2020] uses it to mean [X]
- [Author 2019] uses it to mean [Y]
- Note: [Any variation in usage]

**Related Concepts**: [What concepts travel with it?]

**Key Quote**: "[Definition quote]" (Author Year:page)

4. Identify Cross-Traditional Connections

Some papers bridge traditions:

• Paper A uses Bourdieu AND institutionalism
• Paper B connects legal consciousness to recognition theory

Note these bridges—they may reveal synthesis opportunities.

5. Map Intellectual Lineages

For each major paper in your corpus:

• Which tradition does it claim?
• Who are its key citations?
• What position does it take within the tradition?

Visual representation (text-based):

LEGAL CONSCIOUSNESS TRADITION
├── Ewick & Silbey 1998 (foundational)
│   ├── Hull 2003 (applied to X)
│   ├── Nielsen 2000 (applied to Y)
│   └── Merry 1990 (precursor)
├── Kirk & Papachristos 2011 (legal cynicism variant)
│   └── Desmond et al. 2016 (extended to housing)
└── [Your corpus papers here]

6. Write Theoretical Map

Create theoretical-map.md:

# Theoretical Map

## Overview

This corpus engages [N] distinct theoretical traditions:
1. [Tradition 1] ([N] papers)
2. [Tradition 2] ([N] papers)
3. [Tradition 3] ([N] papers)

---

## Tradition 1: [Name]

### Foundational Texts
- [Text 1]: [one-line description]
- [Text 2]: [one-line description]

### Key Concepts
- **[Concept A]**: [definition]
- **[Concept B]**: [definition]

### Papers in Corpus Using This Tradition
| Paper | How Tradition Is Used |
|-------|----------------------|
| [Author 2020] | [description/extension/critique] |
| [Author 2019] | [description/extension/critique] |

### Intellectual Lineage
[Text-based tree showing citations]

### Key Quote
> "[Defining quote for the tradition]" (Foundational Text:page)

---

## Tradition 2: [Name]
[Repeat structure]

---

## Cross-Traditional Connections

| Connection | Papers | Significance |
|------------|--------|--------------|
| [Tradition A + B] | [Papers] | [Why this matters] |

---

## Concepts Glossary

| Concept | Tradition | Definition | Key Source |
|---------|-----------|------------|------------|
| [Concept 1] | [Tradition] | [Definition] | [Author Year] |

---

## Implications for Writing

Based on this map:
- **Most developed tradition**: [which one]
- **Underdeveloped tradition**: [which one]
- **Productive synthesis opportunity**: [if any]
- **Recommended lit-writeup cluster**: [Gap-Filler/Theory-Extender/etc.]

---

Guiding Principles

Traditions Are Not Boxes

Papers often draw on multiple traditions. The map should show connections, not just categories.

Foundational Texts Matter

Even if everyone cites them, you need to know what they actually say—not just how they're invoked.

Concepts Drift

The same term may mean different things to different authors. Document variation, don't assume consistency.

Citation ≠ Agreement

A paper citing Bourdieu may be critiquing Bourdieu. Note the stance, not just the citation.

This Feeds Architecture

The traditions you identify become the subsections and literatures in lit-writeup. Build the map with writing in mind.

---

Output Files to Create

1. theoretical-map.md - Full map with traditions, concepts, lineages

---

When You're Done

Report to the orchestrator:

• Number of traditions identified
• Key foundational texts
• Any cross-traditional connections
• Concepts that need careful definition
• Preliminary cluster recommendation for lit-writeup

Example summary:

"Theoretical mapping complete. Three traditions identified: legal consciousness (anchored by Ewick & Silbey 1998), neighborhood effects (anchored by Sampson et al. 1997), and recognition theory (anchored by Honneth 2003). Cross-traditional connection noted: 3 papers bridge legal consciousness and recognition. Key concept variation: 'cynicism' means different things across papers. Recommend Synthesis Integrator cluster for lit-writeup given opportunity to bridge traditions. Ready for Phase 3."

Phase 3: Thematic Clustering

Why This Phase Matters

While theoretical mapping shows intellectual traditions, thematic clustering organizes papers by what they study and how. This reveals the empirical landscape: which populations have been studied, which settings, which methods dominate, and where there are gaps.

This clustering directly informs lit-writeup's synthesis paragraphs and helps identify empirical gaps for contribution.

---

Your Tasks

1. Cluster by Empirical Focus

Group papers by what they study:

By Population:

• Which groups are studied? (immigrants, mothers, youth, workers, etc.)
• Which populations are over/under-represented?

By Setting:

• What contexts? (neighborhoods, organizations, institutions, etc.)
• What geographic scope? (US-only? Comparative?)

By Phenomenon:

• What process or outcome? (decision-making, identity formation, mobility, etc.)
• What level of analysis? (individual, organizational, structural)

Create clusters:

## Empirical Clusters

### By Population
- **Working-class mothers**: [Papers A, B, C]
- **Undocumented immigrants**: [Papers D, E]
- **Urban youth**: [Papers F, G, H]

### By Setting
- **Public housing**: [Papers A, D, F]
- **Courtrooms**: [Papers B, E]
- **Schools**: [Papers C, G, H]

### By Phenomenon
- **Police encounters**: [Papers A, B]
- **Housing instability**: [Papers C, D]
- **Identity work**: [Papers E, F, G]

2. Cluster by Methodological Approach

Group papers by how they study:

Data Type:

• In-depth interviews
• Ethnography
• Mixed methods
• Secondary data

Analytic Approach:

• Grounded theory
• Extended case method
• Comparative
• Process tracing

Sample Characteristics:

• Size (small N vs. larger)
• Recruitment (convenience vs. purposive vs. representative)
• Geographic scope

3. Cluster by Theoretical Approach

Complementing Phase 2, group by how theory is used:

Theory as Lens:

• Papers that apply an existing framework (Theory-Extender style)

Theory as Contribution:

• Papers that build new concepts (Concept-Builder style)

Theory as Background:

• Papers with light theoretical framing (Gap-Filler style)

4. Identify Cross-Cluster Papers

Some papers appear in multiple clusters:

• Bridges populations (studies both X and Y)
• Bridges settings (compares context A and B)
• Uses multiple methods

Note these—they often provide models for integrative work.

5. Identify Cluster Gaps

For each clustering dimension, note:

• Over-represented: What's been studied extensively?
• Under-represented: What's missing or sparse?
• Strategic gaps: Where could your project contribute?

6. Write Thematic Clusters Document

Create thematic-clusters.md:

# Thematic Clusters

## Overview

Papers organized by empirical focus, method, and theoretical approach.

---

## By Population

### Cluster: [Population 1]
**Papers**: [N]
- [Author 2020]: [brief description]
- [Author 2019]: [brief description]

**What We Know**: [Summary of findings about this population]

**What's Missing**: [Gaps in studying this population]

### Cluster: [Population 2]
[Repeat structure]

---

## By Setting

### Cluster: [Setting 1]
**Papers**: [N]
- [List with brief descriptions]

**Dominant Findings**: [What do these papers collectively show?]

### Cluster: [Setting 2]
[Repeat]

---

## By Phenomenon

### Cluster: [Phenomenon 1]
**Papers**: [N]
- [List]

**Consensus**: [What most papers agree on]
**Variation**: [Where findings differ]

---

## By Method

| Method | Papers | Notes |
|--------|--------|-------|
| In-depth interviews | [List] | [Typical N, approach] |
| Ethnography | [List] | [Duration, sites] |
| Mixed methods | [List] | [What mix] |

---

## Cross-Cluster Papers

| Paper | Clusters Bridged | Significance |
|-------|-----------------|--------------|
| [Author 2020] | [Pop A + Setting B] | [Why this matters] |

---

## Gap Analysis

### Over-Represented
- [Population/setting/phenomenon with extensive coverage]

### Under-Represented
- [Population/setting/phenomenon with sparse coverage]

### Strategic Gaps for This Project
1. **[Gap 1]**: [Description] — Papers touch on this but don't fully address
2. **[Gap 2]**: [Description] — No papers in corpus address this

---

## Implications for Writing

Based on these clusters:
- **Your project's empirical niche**: [Where it fits]
- **Papers to engage closely**: [Which clusters are most relevant]
- **Gap to highlight in turn**: [What's missing that you address]

---

Guiding Principles

Clusters Should Be Meaningful

Don't create clusters just to categorize. Each cluster should tell you something about the field.

Look for Absences

Sometimes what's NOT in a cluster is more interesting than what's there.

Cross-Cluster Papers Are Valuable

Papers that bridge categories often provide models for your own integrative work.

Clusters Become Synthesis Paragraphs

In lit-writeup, each cluster could become a paragraph or subsection. Think about writing as you cluster.

Multiple Clustering Schemes Are OK

The same paper legitimately belongs in multiple clusters. That's information, not a problem.

---

Output Files to Create

1. thematic-clusters.md - Full clustering analysis

---

When You're Done

Report to the orchestrator:

• Number of clusters per dimension
• Key over/under-represented areas
• Papers that bridge clusters
• Strategic gaps identified

Example summary:

"Thematic clustering complete. By population: 4 clusters (working-class mothers, immigrants, youth, homeowners). By setting: 3 clusters (neighborhoods, courtrooms, schools). Notable gap: no papers study renters specifically despite housing focus. Cross-cluster paper: Desmond 2016 bridges neighborhoods + housing instability. Recommend highlighting renter gap in lit-writeup turn. Ready for Phase 4."

Phase 4: Debate Mapping

Why This Phase Matters

Debates are opportunities. Where scholars disagree—about mechanisms, explanations, definitions, or evidence—there's space for contribution. This phase identifies the tensions, contradictions, and unresolved questions in your field.

For lit-writeup, debates may suggest a Problem-Driven (Debate-Resolver) cluster, or they may reveal productive tensions to address even in other cluster types.

---

Your Tasks

1. Identify Explicit Disagreements

Look for papers that directly critique each other:

• Paper A cites Paper B critically
• Paper A claims to "correct" or "extend" Paper B
• Paper A and Paper B reach opposite conclusions

Signals of explicit disagreement:

• "Contrary to [Author]..."
• "[Author]'s approach overlooks..."
• "We depart from [Author]..."
• "Unlike [Author], we find..."

2. Identify Implicit Tensions

Look for contradictions that aren't acknowledged:

• Papers with incompatible findings
• Papers using the same concept differently
• Papers with incompatible underlying assumptions

Types of implicit tension:

• Empirical: Paper A finds X, Paper B finds not-X
• Theoretical: Paper A assumes mechanism M, Paper B assumes mechanism N
• Definitional: Paper A defines concept as X, Paper B as Y
• Scope: Paper A claims generality, Paper B claims contingency

3. Document Each Debate

For each tension identified:

## Debate: [Short Name]

### The Tension
[One-sentence summary of what's at stake]

### Position 1
- **Claim**: [What this side argues]
- **Key papers**: [Authors]
- **Evidence**: [What supports this position]
- **Underlying assumption**: [What must be true for this to hold]

### Position 2
- **Claim**: [What this side argues]
- **Key papers**: [Authors]
- **Evidence**: [What supports this position]
- **Underlying assumption**: [What must be true for this to hold]

### State of the Debate
- **Is this resolved?**: [Yes/No/Partially]
- **What would resolve it?**: [Type of evidence or analysis needed]
- **Why it matters**: [Stakes for the field]

### Relevance to Your Project
[How does your research speak to this debate?]

4. Map Unresolved Questions

Beyond bilateral debates, identify open questions:

• Questions papers raise but don't answer
• Questions in "future research" sections
• Questions implied by limitations

## Unresolved Questions

### Question 1: [Formulation]
- **Raised by**: [Papers]
- **Why unresolved**: [What would it take to answer?]
- **Relevance**: [Does your project address this?]

5. Identify Contested Concepts

Some concepts are themselves sites of debate:

• Different definitions in circulation
• Different operationalizations
• Different theoretical homes

## Contested Concept: [Name]

### Definition Variants
| Author | Definition | Tradition |
|--------|-----------|-----------|
| [A] | "[definition]" | [tradition] |
| [B] | "[definition]" | [tradition] |

### Why This Matters
[What's at stake in how the concept is defined?]

### Your Position
[How will you define/use this concept?]

6. Write Debate Map

Create debate-map.md:

# Debate Map

## Overview

This field contains [N] identifiable debates and [N] unresolved questions.

**Major debates**:
1. [Debate 1 name]
2. [Debate 2 name]
3. [Debate 3 name]

---

## Debate 1: [Name]

[Full documentation using template above]

---

## Debate 2: [Name]

[Full documentation]

---

## Debate 3: [Name]

[Full documentation]

---

## Unresolved Questions

1. **[Question 1]**
   - Raised by: [papers]
   - What would answer it: [description]

2. **[Question 2]**
   - Raised by: [papers]
   - What would answer it: [description]

---

## Contested Concepts

### [Concept 1]
[Documentation using template]

### [Concept 2]
[Documentation using template]

---

## Implications for Your Project

### Debates You Could Adjudicate
- [Debate X]: Your data could show [how]

### Questions You Could Address
- [Question Y]: Your project directly addresses this

### Positions You'll Take
- On [Debate Z]: You align with [Position] because [reason]

### Lit-Writeup Implications
- **If adjudicating a debate**: Use Problem-Driven (Debate) cluster
- **If taking a side**: Cite both positions, then align
- **If neither**: Note the debate but don't center it

---

Guiding Principles

Not Every Tension Is a Debate

Some differences are just different. A debate exists when scholars are actually in disagreement about something that matters.

Implicit Tensions Are Often More Important

Explicit debates are easy to find. The interesting ones are often unacknowledged contradictions.

Debates Create Contribution Space

If everyone agrees, there's nothing to add. If there's disagreement, your evidence can move the conversation.

You Don't Have to Resolve Every Debate

Note debates, but focus on ones your project actually speaks to.

Fairness Matters

If you're documenting a debate, present both sides fairly—even if you'll eventually take a side.

---

Output Files to Create

1. debate-map.md - Full documentation of debates and tensions

---

When You're Done

Report to the orchestrator:

• Number of debates identified
• Number of unresolved questions
• Key contested concepts
• Debates relevant to user's project
• Recommendation on whether Problem-Driven cluster is appropriate

Example summary:

"Debate mapping complete. 3 major debates identified: (1) structural vs. cultural explanations for legal cynicism, (2) whether police-calling indicates trust or desperation, (3) role of neighborhood context vs. individual disposition. 5 unresolved questions documented. Key contested concept: 'cynicism' has 3 distinct definitions in circulation. User's project directly addresses Debate 2—recommend considering Problem-Driven (Debate) cluster if adjudicating, or noting the debate in a Gap-Filler turn. Ready for Phase 5."

Phase 5: Field Synthesis

Why This Phase Matters

This is where everything comes together. You've audited the corpus, read deeply, mapped theories, clustered themes, and identified debates. Now you synthesize this into a coherent understanding of the field that's ready for writing.

The outputs of this phase are the direct inputs to lit-writeup. This is the handoff.

---

Your Tasks

1. Synthesize Across Phases

Pull together key insights from each phase:

2. Articulate the Field's State

Write a narrative synthesis:

## The State of the Field

### What the Field Knows
[Summary of established findings, 2-3 paragraphs]

### How the Field Thinks
[Summary of dominant theoretical approaches, 1-2 paragraphs]

### Where the Field Disagrees
[Summary of key debates and tensions, 1-2 paragraphs]

### What the Field Lacks
[Summary of gaps—empirical, theoretical, methodological, 1-2 paragraphs]

3. Identify Contribution Opportunities

Based on your synthesis, what are the clearest opportunities for contribution?

Types of contribution:

• Empirical gap: Unstudied population, setting, or phenomenon
• Theoretical extension: Framework not yet applied to X
• Concept building: Need for new analytical tool
• Synthesis: Traditions not yet connected
• Debate resolution: Evidence to adjudicate competing claims

For each opportunity:

### Opportunity: [Name]

**Type**: [Empirical gap / Extension / Building / Synthesis / Debate]

**Description**: [What's the opportunity?]

**Why it's viable**: [Why can this be done now?]

**Required evidence**: [What would demonstrate this contribution?]

**Fit with user's project**: [How well does this fit?]

4. Recommend Lit-Writeup Cluster

Based on the synthesis, recommend a cluster for lit-writeup:

Provide rationale:

## Lit-Writeup Cluster Recommendation

**Recommended**: [Cluster Name]

**Rationale**:
[Why this cluster fits the project's contribution]

**Alternative considered**:
[Another cluster that might fit, and why primary is better]

**Cluster implications**:
- Structure: [Expected subsections]
- Literature balance: [Theory-heavy / Balanced / Substantive]
- Turn strategy: [How to frame the gap]

5. Plan the Handoff

Specify what lit-writeup will need:

## Handoff to Lit-Writeup

### Key Literatures to Engage
1. **[Literature 1]**: [Purpose in section]
2. **[Literature 2]**: [Purpose in section]
3. **[Literature 3]**: [Purpose in section]

### Theoretical Traditions to Reference
- [Tradition 1]: [How it appears in your section]
- [Tradition 2]: [How it appears in your section]

### Concepts Requiring Definition
- **[Concept A]**: Define using [Source]
- **[Concept B]**: Define using [Source]

### Debates to Acknowledge (if any)
- [Debate]: Present as [background / central tension / resolved]

### Suggested Turn Formulation
> "[Draft turn sentence based on identified gap]"

### Key Citations for Each Section
[Map citations to expected subsections]

6. Write Field Synthesis Document

Create field-synthesis.md:

# Field Synthesis

## Project
- **Research question**: [From user]
- **Main argument**: [From user]

---

## The State of the Field

[Narrative synthesis from Task 2]

---

## Contribution Opportunities

[Documentation from Task 3]

---

## Recommended Approach

### Lit-Writeup Cluster: [Name]

[Recommendation and rationale from Task 4]

---

## Handoff to Lit-Writeup

[Handoff specification from Task 5]

---

## Key Documents for Reference

| Document | Location | Purpose |
|----------|----------|---------|
| Corpus Audit | `corpus-audit.md` | Coverage and gaps |
| Reading Notes | `reading-notes/` | Deep engagement evidence |
| Theoretical Map | `theoretical-map.md` | Traditions and lineages |
| Thematic Clusters | `thematic-clusters.md` | Empirical landscape |
| Debate Map | `debate-map.md` | Tensions and questions |

---

## Ready for Lit-Writeup

User can now invoke `/lit-writeup` with this synthesis as input.

Recommended prompt:
> "I've completed lit-synthesis. My field-synthesis.md recommends [Cluster] because [reason]. My main argument is [X]. Please begin Phase 0: Assessment."

---

Guiding Principles

Synthesis ≠ Summary

Don't just repeat what's in other documents. Integrate, interpret, and conclude.

Be Decisive About Contribution

The field synthesis should make a clear recommendation about the user's contribution type.

The Turn Is the Payoff

Everything in this synthesis should point toward what gap the user fills.

Make the Handoff Clean

Lit-writeup should be able to start immediately from your outputs. Don't leave ambiguity.

The User Has Final Say

Your recommendations are informed suggestions. The user knows their project best.

---

Output Files to Create

1. field-synthesis.md - Comprehensive synthesis and handoff document

---

When You're Done

Report to the orchestrator:

• Summary of field state (2-3 sentences)
• Primary contribution opportunity identified
• Recommended lit-writeup cluster
• Readiness for handoff

Example summary:

"Field synthesis complete. The literature on police-calling in disadvantaged neighborhoods is theoretically mature (legal consciousness, recognition) but empirically narrow (focused on residents' perspectives, not meaning-making processes). Primary opportunity: synthesis of legal consciousness and recognition traditions to explain the cynicism-reliance paradox. Recommended cluster: Synthesis Integrator. Handoff document ready with suggested turn: 'Yet while research documents why residents call police despite cynicism, there is less attention to the cultural scripts through which residents make sense of these encounters.' Ready for lit-writeup."