Skip to content
neuroflow0.2.31 · docs

← neuroflow

Commands

Run commands with /neuroflow:<command> in any project folder. Always start with /neuroflow:neuroflow.

The research pipeline

neuroflow models a complete neuroscience research lifecycle. Commands pass context to each other through .neuroflow/ project memory, so each phase picks up where the last left off.

flowchart LR
    A["/ideation"] --> B["/experiment"]
    A --> C["/grant-proposal"]
    B --> D["/tool-build"]
    D --> E["/tool-validate"]
    E --> F["/data"]
    F --> G["/data-preprocess"]
    G --> H["/data-analyze"]
    H --> I["/paper"]
    I --> J["/review"]

Tip

You don't have to follow the pipeline in order. Jump to whichever phase you're in — neuroflow will read your project memory and orient itself.

Entry points

Command What it does
/neuroflow Main entry point — scans your project, sets up .neuroflow/, or shows current status
/setup Integration wizard — Google Workspace, optional Miro (added with claude mcp add, never a token in chat), and an Anthropic-compatible LLM gateway
/migrate After a plugin update, convert the project, your flowie and the team hive to the current format — plan first, writes after you agree

Research pipeline

Command Phase What it does
/ideation ideation Brainstorm, literature search, formalize a research question, write proposal
/preregistration preregistration Pre-register study design and analysis plan on OSF or AsPredicted; log deviations
/grant-proposal grant-proposal Full grant application — aims, significance, innovation, approach, budget
/finance finance Budget planning, expense tracking, financial reports, and grant compliance
/experiment experiment PsychoPy paradigm, recording setup, LSL/marker configuration
/tool-build tool-build Build lab tools — real-time systems, BCI, acquisition pipelines
/tool-validate tool-validate Testing pipeline for timing, markers, data output, edge cases
/data data Data inventory, BIDS validation, format conversion
/data-preprocess data-preprocess Filtering, ICA, epoching, artifact rejection, QC
/data-analyze data-analyze ERPs, time-frequency, connectivity, decoding, GLM
/paper paper Unified manuscript writing — draft section by section, write→critique loop
/review review Peer review a colleague's paper — structured referee report
/poster poster LaTeX conference poster from project memory — template (A0 portrait or landscape, A1 portrait, 90×120 cm, 48×36 in), QR code, poster-critic review loop
/notes notes Live note-taking — capture rough input, reformat into a clean document
/write-report write-report Generate a structured report from .neuroflow/ for any phase or the whole project

Brain simulation

Command What it does
/brain-build Assemble a computational brain model — neuron models, network topology, connectivity, simulation framework setup
/brain-optimize Run a parameter search or fit the model to experimental data
/brain-run Run the model as a simulation — configure run parameters, launch, and collect outputs

Memory & team

Command What it does
/flowie Personal research OS — a private GitHub repository for your research profile, cross-project task board, project registry and personal wiki
/hive Team layer — a shared GitHub org repository for research directions, explicit knowledge sharing and team-aware recommendations; never shares personal project data on its own
/tasks The 3-tier Kanban task board — view, add, move, complete and archive tasks at project, flowie (personal) or hive (team) level
/meeting Meetings — schedule, prepare agendas with project context, send calendar invites, take structured notes, and turn action items into tasks, at project, flowie or hive level
/wiki Project-level shared knowledge base — an LLM-maintained wiki at .neuroflow/wiki/, git-tracked and shared with collaborators (/flowie --wiki-* and /hive --wiki-* keep the personal and team wikis)

Utility commands

Command What it does
/ethics Ethics and IRB workflow — protocol submissions and amendments, versioned consent forms, approval status and expiry; the hard gate before any human data collection
/git Context-aware git utility — smart push/pull, commit message generation, branch management, and PR creation with shorthand aliases
/pipeline Define and run a multi-step research pipeline — one step per invocation from a saved plan; interactive by default, or --executor for brutal mode (no questions)
/autoresearch Open-ended improvement loop for any research artifact — one focused change per iteration, judged against the current best, kept or reverted, with a per-loop wiki as memory; stops at the caps you set or when you stop it
/output Export project memory or the whole project — pack as a zip archive or copy to a folder for sharing, archiving, or handoff
/slideshow Build a presentation from selected areas of the project — pick phases, figures and key findings, get a structured slide deck ready to export
/search Lightweight scoped search — use memory: to search .neuroflow/ or project: to search the codebase
/interview Interview preparation from either side — generate tailored questions, run practice Q&A, evaluate readiness
/phase Show the phase map and switch phase — a picker with the neuroflow mod
/dashboard The project at a glance — phase map, deadlines, integrity state, task board, autoresearch loop
/doctor Health check of the setup around a project — tools, backups, storage location, contracts, and whether the mod is live
/sentinel Full audit of .neuroflow/ — deterministic checks first (nf_check.py), then drift, broken references, preregistration vs progress
/quiz Neuroscience quiz — flashcards, pub quiz, or rapid-fire throw questions
/fails Log dissatisfaction — record plugin behavior, science quality, or UX issues; optionally opens a GitHub issue
/idk Personal support companion — decompress, break down overwhelming tasks, or just chat

How commands work

Every command follows the same lifecycle:

  1. Read skills — Claude loads the phase-specific skill (e.g. neuroflow:phase-ideation) for domain guidance
  2. Read project memory — Claude reads project_config.md and flow.md to orient itself
  3. Do the work — Claude interacts with you to complete the phase task
  4. Write outputs — Results go into the phase subfolder inside .neuroflow/
  5. Update memory — flow.md and project_config.md are updated; a session entry is appended

This means every command is context-aware — it knows your research question, your modality, your tools, and what's been done before.