Claude Think Tank

A template repository for structured policy research using Claude Code as an AI-assisted research environment.

Overview

This repository provides a structured framework for conducting deep policy research using Claude Code. It's designed to help researchers, advocates, and policy analysts organize their research workflow, leverage AI assistance effectively, and build comprehensive policy analysis over time.

What is this?

Claude Think Tank is a research workspace template that treats a code repository as a structured research environment. It combines:

• Organized file structure for context, prompts, and outputs
• Specialized AI agents for research tasks
• Slash commands for common research workflows
• Clear processes for iterative research and policy development

This approach works particularly well for:

• Policy research and analysis
• Comparative policy studies
• Advocacy and lobbying preparation
• Long-term research projects
• Evidence-based policy development

Key Features

• Context Management: Centralized storage for background information and sources
• Prompt Organization: Structured workflow from draft to execution to archive
• Output Management: Date-organized research outputs with clear provenance
• Policy Development: Dedicated space for translating research into actionable policy
• Specialized Agents: AI agents trained for specific research tasks
• Research Workflows: Slash commands for common research operations

Repository Structure

.
├── context/           # Background information and persistent context
│   ├── sources/       # Source materials and references
│   └── context.md     # Main context document
├── prompts/           # Research prompts and questions
│   ├── drafting/      # Prompts being developed
│   ├── to-run/        # Prompts ready to execute
│   └── run/           # Completed prompts (archive)
├── outputs/           # Generated research documents
│   └── DDMMYY/        # Date-organized folders (e.g., 071125)
├── policy/            # Policy recommendations and advocacy materials
├── data/              # Supporting data
│   ├── docs/          # Document sources
│   └── structured/    # Extracted structured data
├── archive/           # Historical materials
├── audio/             # Podcast and audio content
│   ├── scripts/       # Reformatted scripts for TTS
│   └── episodes/      # Generated audio files
├── exports/           # Export outputs
│   └── pdf/           # PDF compilations
└── .claude/           # Claude Code configuration
    ├── commands/      # Slash commands for workflows
    └── agents/        # Specialized research agents

Getting Started

1. Clone This Template

git clone https://github.com/danielrosehill/Claude-Think-Tank your-research-project
cd your-research-project

2. Define Your Policy Challenge

Recommended: Run the /define-policy-challenge slash command to interactively define your research focus:

# In Claude Code
/define-policy-challenge

This command will guide you through defining:

• Your specific policy challenge or research area
• Research objectives and scope
• Key stakeholders and timeline
• Geographic or jurisdictional boundaries

The agent will then automatically update your context files with this information.

3. Manual Customization (Alternative)

If you prefer to customize manually, edit these files:

1. context/context.md - Add background information about your research topic
2. CLAUDE.md - Update the instructions for Claude to reflect your research focus
3. README.md (this file) - Update with your project details
4. .claude/agents/ - Customize agent descriptions and focus areas

4. Start Researching

1. Add context materials to context/ and context/sources/
2. Create research questions in prompts/drafting/
3. Move ready prompts to prompts/to-run/
4. Use slash commands like /run-prompt to execute research
5. Review outputs in outputs/DDMMYY/
6. Create policy documents in policy/ when ready

Available Slash Commands

Setup

• /define-policy-challenge - Define your policy research focus (recommended first step)

Research Operations

• /new-prompt - Create a new research prompt
• /run-prompt - Execute a research prompt
• /run-batch-of-prompts - Process multiple prompts
• /organize-outputs - Maintain output organization
• /synthesize-research - Combine multiple outputs

Export & Publishing

• /export-pdf - Generate PDF compilations
• /generate-podcast - Create audio versions of research

Research Workflow

The "Gather and Distill" Approach

This template follows a research methodology:

1. Context: Gather background information and sources
2. Questions: Formulate research questions as prompts
3. Research: Execute prompts to generate initial analysis
4. Organization: Keep outputs organized by date and topic
5. Synthesis: Combine and refine research over time
6. Policy: Translate findings into actionable recommendations

Working with Claude

This repository is designed to work seamlessly with Claude Code. The AI agents and slash commands provide specialized assistance for:

• Executing research prompts with proper context
• Organizing and maintaining the research repository
• Synthesizing findings across multiple research outputs
• Generating policy documents grounded in research
• Creating derivative materials (PDFs, podcasts)

Customization Guide

For Policy Research

If you're researching a specific policy area:

1. Update context/context.md with your policy focus and objectives
2. Customize the policy-writer agent in .claude/agents/ for your domain
3. Add relevant sources to context/sources/
4. Create initial research questions in prompts/drafting/

For Different Research Types

While designed for policy research, this template can be adapted for:

• Comparative analysis across jurisdictions
• Literature reviews and meta-analysis
• Evidence synthesis for advocacy
• Long-form investigative research
• Academic research projects

Best Practices

1. Maintain Context: Keep context/context.md updated as your understanding evolves
2. Organize Promptly: Use /organize-outputs regularly to maintain structure
3. Iterative Refinement: Treat research as iterative - revisit and refine over time
4. Document Sources: Track source materials in context/sources/
5. Version Control: Commit regularly to preserve research history
6. Separate Research from Policy: Keep exploratory research (outputs) separate from policy recommendations (policy)

Philosophy

This template embodies a few key principles:

• Structure enables creativity: Clear organization frees mental energy for analysis
• AI as research partner: Claude Code can accelerate research while you provide validation and direction
• Transparency through documentation: Everything is tracked and traceable
• Long-term thinking: Build a research base that compounds over time
• Evidence-based advocacy: Ground policy recommendations in solid research

Example Use Cases

• Policy Advocacy: Research to support lobbying efforts
• Comparative Analysis: Study how different jurisdictions approach a policy problem
• Issue Education: Build comprehensive understanding of a complex policy area
• Legislative Preparation: Develop evidence base for proposed legislation
• Academic Research: Organize literature review and analysis

Contributing

This is a template repository. Feel free to:

• Fork it for your own research projects
• Adapt the structure to your needs
• Share improvements back to the template
• Use it as a model for other research types

License

MIT