See also: Claude Research Workspace — the private/personal version of this template (no public-audience features, no export commands).

Open Research Workspace — Claude Code Template

A template for conducting public, transparent AI-assisted research using Claude Code as the execution engine. Fork it, fill in your research brief, and start investigating — with every prompt, output, and methodology choice visible to the world.

What Makes This Different

This isn't just a research workspace — it's designed to be open-sourced from day one. That means:

• Full transparency: Every prompt, every output, every compaction is committed to the repo. Readers can trace exactly how conclusions were reached.
• External-audience outputs: The agent writes findings for public readers, not just the researcher. Key findings are accessible, well-cited, and self-contained.
• Export-ready: Built-in commands to format research as blog posts, reports, briefing documents, and social media threads.
• Publishing integrations: Scaffold MCP connections to WordPress, Ghost, Notion, and more.
• AI disclosure: All outputs include a standard note that they were produced using AI-assisted research.

Getting Started

1. Use this template

Click Use this template on GitHub, or run:

./new-project.sh "my-research-topic" ~/repos/github/

2. Set up your research brief

Edit context/from-human/research-brief.md with:

• Your research topic and scope
• Key questions to investigate
• Intended audience — who will read this?
• Licensing — how should others use your findings?

3. Write your first prompt (or record a voice note)

Option A — Write it: Create a prompt in prompts/run/initial/ describing what you want to investigate. See the example prompt for the format.

Option B — Speak it: Record a voice note, drop the audio file in the project root, and run:

/voice-note

This transcribes your recording via AssemblyAI, distills it into a structured research prompt, and queues it — preserving the audio, raw transcript, and cleaned prompt in a timestamped voice-notes/ folder. Requires an AssemblyAI API key in .env (see .env.example).

4. Run it

Open the repo in Claude Code and tell it to run the prompt:

/run-prompt

5. Iterate, export, share

• Review outputs in outputs/individual/
• Write follow-up prompts in prompts/run/subsequent/
• When ready to share: /export blog or /export report
• Update the public README: /publish-readme

Directory Structure

├── .env.example                 # API keys template (copy to .env)
├── CLAUDE.md                    # System instructions for Claude Code
├── context/
│   ├── from-human/              # Your research brief and background info
│   ├── from-history/            # Compacted findings from prior iterations
│   └── from-internet/           # Saved web sources and references
├── prompts/
│   ├── drafting/                # Prompts under development
│   ├── queue/                   # Ready to run (ordered)
│   └── run/
│       ├── initial/             # First-pass research prompts
│       └── subsequent/          # Follow-up prompts
├── outputs/
│   ├── individual/              # Per-prompt research outputs
│   ├── aggregated/
│   │   ├── markdown/            # Combined research documents
│   │   └── pdf/                 # PDF exports
│   ├── published/               # Export-ready formats (blog, report, social)
│   └── final/                   # Polished deliverables
├── private/                     # Researcher's private notes (gitignored)
├── scripts/
│   └── transcribe.py            # AssemblyAI transcription helper
├── slash-commands/              # Custom Claude Code slash commands
├── voice-notes/                 # Timestamped voice note artifacts
└── notes/                       # Working notes and methodology

Workflow

 ┌─────────────┐
 │   Context    │◄──────────────────────┐
 │  (from-human │                       │
 │  from-history│                       │
 │  from-internet)                      │
 └──────┬──────┘                        │
        │                               │
        ▼                               │
 ┌─────────────┐                        │
 │   Prompt     │                       │
 │  (queue/run) │                       │
 └──────┬──────┘                        │
        │                               │
        ▼                               │
 ┌─────────────┐      ┌────────────┐    │
 │   Claude     │─────►│  Output    │    │
 │   Code       │      │ (individual)   │
 └─────────────┘      └──────┬─────┘    │
                             │          │
                     ┌───────┴───────┐  │
                     │  Compaction   │──┘
                     │  (summarise   │
                     │   → history)  │
                     └───────┬───────┘
                             │
                     ┌───────┴───────┐
                     │               │
                     ▼               ▼
              ┌────────────┐  ┌────────────┐
              │ Aggregation│  │   Export    │
              │ (combined  │  │ (blog, rpt,│
              │  report)   │  │  social)   │
              └────────────┘  └────────────┘

Slash Commands

Command	Purpose
/run-prompt	Execute the next prompt in the queue
/compact	Summarise outputs into compacted history
/aggregate	Combine individual outputs into a single document
/status	Show research progress
/export	Format outputs for publishing (blog, report, brief, social, newsletter)
/publish-readme	Regenerate README.md with current findings for external readers
/voice-note	Transcribe a voice recording → structured research prompt → queue
/setup-publishing	Scaffold .mcp.json for CMS/publishing integrations

For Readers

If you're here to read the research (not use the template), here's where to look:

• outputs/published/ — Polished, export-ready versions of the findings
• outputs/aggregated/markdown/ — Full combined research reports
• outputs/individual/ — Raw outputs from each research iteration
• prompts/ — The exact prompts used, so you can see what questions were asked
• context/from-human/research-brief.md — The original research brief

For Researchers

If you want to use this template for your own public research:

1. Click Use this template on GitHub
2. Replace the research brief with your topic
3. Clear the example prompts
4. Start researching with Claude Code
5. Everything you commit is public — use private/ for anything you want to keep off the record

Design Philosophy

• Transparency by default: The research process is the product, not just the findings
• Filesystem as workflow engine: Folder structure defines the process
• Markdown-native: Everything is plain text, version-controlled, portable
• Compaction over RAG: Summarise and feed back rather than vectorise
• Iterative deepening: Each round builds on compacted findings from the last
• Export-ready: Research should flow naturally from workspace to publication

Disclaimer

Research in this workspace is conducted using AI-assisted tools (Claude Code). All prompts, methodology, and raw outputs are transparent in the repository. Findings should be independently verified before being relied upon for decision-making.

License

MIT

---

For more Claude Code projects, visit my Claude Code Projects Index.