BeClaude

lit-synthesis

New
12Community RegistryDevelopmentby Neal Caren · MIT

Deep reading and synthesis of literature corpus. Theoretical mapping, thematic clustering, and debate identification using Zotero MCP for full-text access.

Community PluginView Source

Overview

A Claude Code plugin marketplace with skills for rigorous quantitative and qualitative analysis in sociology and related social sciences. These skills guide you through systematic, publication-ready research workflows.

Installation

bash
# Add this marketplace to Claude Code
/plugin marketplace add nealcaren/social-data-analysis

# Install only the plugins you need
/plugin install r-analyst@social-data-analysis
/plugin install stata-analyst@social-data-analysis
/plugin install interview-analyst@social-data-analysis
/plugin install interview-writeup@social-data-analysis
/plugin install dag-development@social-data-analysis
/plugin install abductive-analyst@social-data-analysis
/plugin install text-analyst@social-data-analysis
/plugin install lecture-designer@social-data-analysis
/plugin install lit-review@social-data-analysis

Available Plugins

Each plugin provides a single focused skill. Install only what you need:

SkillInvocationDescription
R Statistical Analyst/r-analystPhased quantitative analysis workflow using R (DiD, IV, matching, etc.)
Stata Statistical Analyst/stata-analystPhased quantitative analysis workflow using Stata
Interview Analyst/interview-analystPragmatic qualitative analysis for interview data
Interview Write-Up/interview-writeupWrite-up support for interview methods and findings
DAG Development/dag-developmentDevelop causal diagrams and render publication-ready figures (Mermaid, R, Python)
Abductive Analyst/abductive-analystAbductive analysis (Timmermans & Tavory) for theory-generating qualitative research
Text Analyst/text-analystComputational text analysis with R and Python (topic models, sentiment, classification)
Lecture Designer/lecture-designerTransform textbook chapters into engaging lectures with Quarto slides
Lit Review/lit-reviewBuild literature databases via OpenAlex

Each skill uses a phased workflow with mandatory pauses between phases for user review and decision-making.

Workflow Overview

Quantitative Analysis (R/Stata)

code
Phase 0: Research Design → Establish identification strategy
    ↓ [User Review]
Phase 1: Data Familiarization → Descriptives, quality checks
    ↓ [User Review]
Phase 2: Model Specification → Pre-specify models before estimation
    ↓ [User Review]
Phase 3: Main Analysis → Run models, interpret results
    ↓ [User Review]
Phase 4: Robustness → Sensitivity analysis, placebo tests
    ↓ [User Review]
Phase 5: Output → Publication-ready tables, figures, narrative

Qualitative Analysis (Interviews)

code
Phase 0: Theory Preparation → Sensitizing concepts (optional)
    ↓ [User Review]
Phase 1: Immersion → Read transcripts, create memos
    ↓ [User Review]
Phase 2: Coding → Develop codebook, apply codes
    ↓ [User Review]
Phase 3: Interpretation → Identify patterns, develop explanations
    ↓ [User Review]
Phase 4: Quality Check → Assess against 5 quality indicators
    ↓ [User Review]
Phase 5: Synthesis → Write publication-ready sections

Abductive Analysis (Timmermans & Tavory)

code
Phase 0: Theoretical Preparation → Build theoretical sensitivity
    ↓ [User Review]
Phase 1: Familiarization → Open coding, flag surprises
    ↓ [User Review]
Phase 2: Theoretical Casing → Apply multiple theoretical lenses
    ↓ [User Review]
Phase 3: Anomaly Analysis → Identify contradictions and puzzles
    ↓ [User Review]
Phase 4: Memo Writing → Develop tentative theory
    ↓ [User Review]
Phase 5: Integration → Test theory against full dataset
    ↓ [User Review]
Phase 6: Writing Up → Rhetorical abduction for publication

Computational Text Analysis (R/Python)

code
Phase 0: Research Design → Method selection, language choice (R or Python)
    ↓ [User Review]
Phase 1: Corpus Preparation → Load, clean, explore text data
    ↓ [User Review]
Phase 2: Specification → Document preprocessing, specify parameters
    ↓ [User Review]
Phase 3: Analysis → Run topic models, classifiers, sentiment
    ↓ [User Review]
Phase 4: Validation → Human validation, diagnostics, robustness
    ↓ [User Review]
Phase 5: Output → Publication-ready tables, figures, replication

Lecture Design

code
Phase 0: Context & Outcomes → Define measurable learning outcomes
    ↓ [Instructor Review]
Phase 1: Content Audit → Narrative arc (ABT), chunk map, hook design
    ↓ [Instructor Review]
Phase 2: Active Learning → Polls, ConcepTests, peer instruction
    ↓ [Instructor Review]
Phase 3: Slide Development → Quarto reveal.js with speaker notes
    ↓ [Instructor Review]
Phase 4: Review → Timing audit, backup plans, instructor guide

Repository Structure

code
.claude-plugin/
└── marketplace.json              # Plugin marketplace definition (9 plugins)

plugins/
├── r-analyst/
│   └── skills/r-analyst/
│       ├── SKILL.md              # R orchestrator
│       ├── phases/               # Phase agents
│       └── techniques/           # R code reference guides
│
├── stata-analyst/
│   └── skills/stata-analyst/
│       ├── SKILL.md              # Stata orchestrator
│       ├── phases/               # Phase agents
│       └── techniques/           # Stata code reference guides
│
├── interview-analyst/
│   └── skills/interview-analyst/
│       ├── SKILL.md              # Interview orchestrator
│       └── phases/               # Phase agents
│
├── interview-writeup/
│   └── skills/interview-writeup/
│       ├── SKILL.md              # Interview write-up orchestrator
│       └── phases/               # Phase agents
│
├── dag-development/
│   └── skills/dag-development/
│       ├── SKILL.md              # DAG development orchestrator
│       └── phases/               # Phase agents
│
├── abductive-analyst/
│   └── skills/abductive-analyst/
│       ├── SKILL.md              # Abductive analysis orchestrator
│       └── phases/               # Phase agents (7 phases)
│
├── text-analyst/
│   └── skills/text-analyst/
│       ├── SKILL.md              # Text analysis orchestrator
│       ├── phases/               # Phase agents
│       ├── concepts/             # Method concepts (language-agnostic)
│       ├── r-techniques/         # R text analysis code guides
│       └── python-techniques/    # Python text analysis code guides
│
└── lecture-designer/
    └── skills/lecture-designer/
        ├── SKILL.md              # Lecture design orchestrator
        ├── phases/               # Phase agents
        ├── pedagogy/             # Teaching methodology (overview)
        └── quarto/               # Quarto reveal.js reference

└── lit-review/
    └── skills/lit-review/
        ├── SKILL.md              # Literature review orchestrator
        ├── phases/               # Phase agents
        └── api/                  # OpenAlex API reference

Key Features

Quantitative Skills

  • •Identification-first: Establish research design before estimation
  • •Pre-specification: Document model choices before seeing results
  • •Robustness built-in: Sensitivity analysis, placebo tests, wild bootstrap
  • •Nonlinear model interpretation: AMEs, predicted probabilities, proper diagnostics
  • •Missing data handling: Multiple imputation with adequate m
  • •Survey methodology: Weighting, design effects, response rates
  • •Publication checklists: Minimum, strong, and exemplary standards

Qualitative Skills

  • •Theory-informed or data-first: Choose your approach
  • •Systematic coding: Codebook development with examples
  • •Quality indicators: Cognitive empathy, heterogeneity, palpability, follow-up, self-awareness
  • •Evidence selection: Luminous exemplars, not just typical quotes
  • •Methods transparency: Detailed templates for sampling, recruitment, saturation
  • •Write-up support: Methods drafting, findings structure, quote use, and revision checklists

Methods Skills

  • •DAG development: Build causal diagrams from theory and render figures in Mermaid, R, or Python

Abductive Analysis Skills

  • •Theory-first approach: Build theoretical sensitivity before data engagement
  • •Map and compass theories: Both substantive and grammatical frameworks
  • •Anomaly detection: Systematic identification of contradictions and puzzles
  • •Theoretical casing: View data through multiple theoretical lenses
  • •Rhetorical abduction: Structure writing as what we knew → surprise → new theory

Text Analysis Skills

  • •Dual-language support: R for topic models/STM; Python for transformers/BERTopic
  • •Method selection guidance: Match methods to research questions
  • •Validation required: Human validation, coherence metrics, robustness checks
  • •Topic modeling: LDA, STM (R), BERTopic (Python) with K selection guidance
  • •Sentiment analysis: VADER, lexicon-based, and ML approaches
  • •Supervised classification: Traditional ML and transformer fine-tuning
  • •Reproducibility: Documented preprocessing, seeds, package versions

Lecture Design Skills

  • •Learning outcomes first: Backward design from measurable outcomes
  • •Narrative structure: ABT (And-But-Therefore) for cognitive engagement
  • •Cognitive load management: Chunking, attention resets every 12-18 minutes
  • •Active learning integration: Polls, ConcepTests, Peer Instruction
  • •Quarto reveal.js output: Publication-quality slides with speaker notes

Requirements

  • •Claude Code CLI
  • •R (for R skills) or Stata (for Stata skills)
  • •Interview transcripts (for interview skill)
  • •R and/or Python (for text analysis skill)

Contributing

Contributions welcome! Please:

  1. Fork the repo
  2. Create a feature branch
  3. Submit a pull request

License

MIT License - see LICENSE file

Acknowledgments

These skills draw on methodological guidance from:

  • •Gerson & Damaske, The Science and Art of Interviewing
  • •Small & Calarco, Qualitative Literacy
  • •Long & Mustillo (2017), Mize (2019) on nonlinear model interpretation
  • •Various Social Forces editorial guidelines

Install & Usage

1
Create the skills directory
mkdir -p .claude/skills
2
Download the skill file
mkdir -p .claude/skills && curl -o .claude/skills/lit-synthesis.md https://raw.githubusercontent.com/nealcaren/social-data-analysis/main/SKILL.md
3
Invoke in Claude Code
/lit-synthesis
View source on GitHub
mcpliterature-synthesiszoterotheoretical-mappingsociologyresearch

Frequently Asked Questions

What is lit-synthesis?

Deep reading and synthesis of literature corpus. Theoretical mapping, thematic clustering, and debate identification using Zotero MCP for full-text access.

How to install lit-synthesis?

To install lit-synthesis, create the .claude/skills directory in your project, then run the curl command to download the skill file. Once installed, invoke it in Claude Code with /lit-synthesis.

What is lit-synthesis best for?

lit-synthesis is a community categorized under Development. It is designed for: mcp, literature-synthesis, zotero, theoretical-mapping, sociology, research. Created by Neal Caren.