Quickstart¶
Get from zero to a running neuroflow project in under five minutes.
Step 1 — Install¶
Step 2 — Open your project¶
Navigate to your project folder (or any empty folder to start fresh) and open your AI assistant:
Step 3 — Run /neuroflow¶
Type this in Claude Code:
If this is a new project, neuroflow will:
- Scan your folder for signals (BIDS data, scripts, manuscript files, etc.)
- Ask you a few short questions about your project
- Create
.neuroflow/— the shared project memory
If you already have a .neuroflow/ folder, it will show your current phase and status instead.
Example session
You: /neuroflow:neuroflow
Claude: I found a `paradigm/` folder with PsychoPy scripts and a `scripts/` folder
with Python preprocessing code. This looks like an EEG study in progress.
What are you working on? (one or two sentences)
You: An auditory oddball study on attention in healthy adults. I need to preprocess
the EEG and run ERP analysis.
Claude: Got it. Setting up neuroflow project memory...
✅ Created .neuroflow/
✅ Active phase: data-preprocess
Next step: run /neuroflow:data-preprocess
Step 4 — Set up integrations (optional)¶
neuroflow can search PubMed and use Miro for visual collaboration. Run the setup wizard:
This stores credentials in .neuroflow/integrations.json (git-ignored).
Step 5 — Follow the pipeline¶
Run the command for your current phase:
Brainstorm a research question, search the literature, or write a project proposal.
Build and run an EEG/fMRI preprocessing pipeline with filtering, ICA, and QC.
Key concepts¶
| Concept | What it is |
|---|---|
| Project memory | .neuroflow/ folder — every command reads and writes here |
| Commands | /neuroflow:<name> — the main way you interact with neuroflow |
| Skills | Domain knowledge Claude loads automatically for each phase |
| Agents | Autonomous subprocesses for focused tasks (e.g. literature search) |