Home
hi, I am neuroflow
— the agentic operating system for neuroscience research. From hypothesis to manuscript, every phase orchestrated by AI.
🌱 Did I fail you? :') I'm in pre-release and open to the community — 🐛 open an issue or 🔀 improve me with a PR
✨ why neuroflow?
.neuroflow/ — Claude reads it at the start of every session so you never repeat yourself.
Who is this for? Neuroscientists working with EEG, iEEG, fMRI, eye tracking, ECG…
Research pipelines¶
neuroflow adapts to your research — click a pipeline to see how the commands connect.
Full research cycle Full cycle
From first hypothesis to peer-reviewed manuscript — the complete end-to-end journey. Ideate, run an experiment, collect and preprocess data, analyze, write, and get it reviewed.
All context is stored in .neuroflow/ — shared memory that carries every decision forward.
Data analysis study Analysis
You already have data collected. Inventory it, run preprocessing, extract results, and write them up — without touching the experiment design phase.
ERPs, time-frequency, connectivity, decoding, GLM — automatically reported with pre-registration compliance checks.
Research tool build Engineering
Building a new analysis algorithm, preprocessing library, or experiment software? Ideate the design, scaffold the implementation, validate it — and optionally write it up as a methods paper.
The paper step is optional — publish a methods article or keep the tool as internal infrastructure.
Computational modelling Modelling
Build and run a brain simulation model. Ideate the architecture, construct the model, then optimize and run it — with empirical data analysis running in parallel and feeding into the optimization step.
Data analysis runs in parallel with model construction and merges before optimization — the paper step is optional.
Literature review & ideation Ideation
Start with a broad question, search PubMed and bioRxiv, synthesise the field, and sharpen a novel hypothesis — then write it up as a review article and optionally convert it into a grant proposal.
The scholar agent searches PubMed, bioRxiv, and Semantic Scholar in real time to ground every idea in the current literature.
Quick install¶
Then open any project folder and run:
How it works¶
.neuroflow/
├── project_config.md ← active phase, research question, modality, tools
├── flow.md ← index of all subfolders
├── reasoning/ ← structured per-phase decision logs (JSON)
├── sessions/ ← daily session logs (git-ignored)
├── ideation/ ← research questions, literature reviews, proposals
├── grant-proposal/ ← grant application drafts
├── experiment/ ← paradigm scripts, recording setup
├── tool-build/ ← tool specs and implementation notes
├── tool-validate/ ← validation plans and test results
├── data/ ← data inventory and intake reports
├── data-preprocess/ ← preprocessing configs and QC reports
├── data-analyze/ ← analysis plans and result summaries
├── paper/ ← manuscript drafts and critic logs
├── notes/ ← session and meeting notes
└── write-report/ ← generated project reports
Every command reads .neuroflow/ at the start and writes its output there. This shared memory means Claude always knows what phase you're in, what you've decided, and what's been done — across sessions and across commands.