NeuroClaw

Use this model doc whenever the user wants to run BrainGNN for fMRI phenotype prediction, including graph construction, training, and evaluation. This document focuses on model-level usage and delegates upstream preprocessing to fmri-skill (and optionally hcpya-skill for HCP data).

Use this skill when users need to build, populate, or extend a domain-specific knowledge graph from literature and structured databases. Triggers include: 'build knowledge graph', 'extract claims from papers', 'ingest data into graph', 'batch extract claims', 'knowledge graph construction', 'populate graph from PubMed', 'extract structured claims', 'ingest atlas data', or any request involving knowledge graph population from scientific literature or biomedical databases. Covers both structured data ingestion (Phase 1) and LLM-based claim extraction from papers (Phase 2).

Use this skill whenever the user wants to formalize a network architecture and derive theoretical components from a research idea. Triggers include: 'method design', 'design method', 'network architecture', 'formula derivation', 'method-design', 'theoretical framework', 'derive equations', 'compare alternatives', 'architecture tournament', 'rank designs', 'design space exploration', or any request to transform IDEA.md into a detailed METHOD.md. This skill is the **mandatory interface-layer method formalizer** in NeuroClaw: it reads IDEA.md, optionally drafts N alternative architectures and ranks them via a Bradley-Terry pairwise tournament (Tournament Mode), designs concrete network structures (layers, modules, connections), performs mathematical derivations (equations, loss functions, proofs), and always outputs a structured METHOD.md.

Use this skill whenever the user wants to remove site/scanner/batch effects from neuroimaging features before running downstream models, run mega-analysis across multiple datasets, or evaluate models with leave-site-out / site-stratified protocols. Triggers include: 'harmonize', 'ComBat', 'CovBat', 'site effect', 'scanner effect', 'batch effect', 'leave-site-out', 'mega-analysis', 'multi-site', 'cross-site', 'neuroHarmonize'. This is a horizontal cross-cutting layer between dataset skills and model skills.

Use this skill whenever the user wants to run phenotype-prediction models, browse model cards, map model inputs/outputs, or choose an execution route for fMRI/sMRI based models. This is a model-entry orchestration skill: it routes requests to model-specific docs and delegates preprocessing to modality skills.

Use this model doc whenever the user wants to run Com-BrainTF (Community-aware Brain Transformer) for fMRI phenotype prediction. Com-BrainTF uses dense FC matrices with a two-level Transformer (per-community local + global) and DEC pooling. NeuroClaw auto-derives community partitions from atlas naming conventions (Yeo 7-net for Schaefer, lobe-based for AAL).

Use this model doc whenever the user wants to run IBGNN (Interpretable Brain Graph Neural Network) for fMRI phenotype prediction. IBGNN is a PyG-based GNN with a learnable MLP message function over [x_i, x_j, edge_attr], designed for connectome-based brain disorder analysis with post-hoc edge-mask explainer support.

Use this model doc whenever the user wants to run LG-GNN (Local-to-Global GNN) for fMRI phenotype prediction. LG-GNN is a PyG-based GNN with SABP (Self-Attention Brain Pooling) and mutual-information regularization. NeuroClaw adapts the original population-graph version to single-subject brain graphs.

Use this model doc whenever the user wants to run BrainNetworkTransformer for fMRI phenotype prediction, including data loading, training, and evaluation. BNT uses dense FC matrices (no PyG dependency) with DEC pooling + interpretable transformer encoder.

Use this skill whenever the user wants to run the NeuroSTORM multi-model fMRI platform: preprocessing, pretraining (MAE or contrastive), fine-tuning, inference, or benchmarking. It covers 8 built-in models — NeuroSTORM, SwiFT, BrainGNN, BrainNetworkTransformer (BNT), LG-GNN, Com-BrainTF, IBGNN, BrainNetCNN — across 3 input modalities (voxel 4D, ROI time series 2D, functional connectivity 2D). Triggers include: 'fMRI', 'NeuroSTORM', 'SwiFT', 'BrainGNN', 'BNT', 'BrainNetCNN', 'LG-GNN', 'Com-BrainTF', 'IBGNN', 'fMRI preprocessing', 'fMRI foundation model', 'ROI time series', 'functional connectivity', 'brain graph', 'HCP', 'ABCD', 'UKB', 'ADHD200', 'COBRE', 'UCLA', 'NSD', 'BOLD5000', 'disease diagnosis from fMRI', 'pretrain fMRI model', 'fine-tune fMRI', or any request involving .nii/.nii.gz fMRI volume files.

Use this skill when users need to search academic papers, download research documents, extract citations, or gather scholarly information. Triggers include: requests to "find papers on", "search research about", "download academic articles", "get citations for", or any request involving academic databases like arXiv, PubMed, Semantic Scholar, or Google Scholar. Also use for literature reviews, bibliography generation, and research discovery.

Use this skill after a conversation or task is completed when the user wants a clean, beautiful HTML chat log. It keeps only direct NeuroClaw <-> User dialogue, filters out tool calls / internal traces / SKILL.md reading notes, and renders distinct colored message cards for each side.

Use this skill whenever the user wants to automatically organize raw neuroimaging data (DICOM, NIfTI, EEG, etc.) into a valid BIDS (Brain Imaging Data Structure) dataset. Triggers include: 'organize to BIDS', 'BIDS organizer', 'convert to BIDS', 'BIDS conversion', 'bidsify', 'create BIDS dataset', 'raw data to BIDS', or any request to structure data according to BIDS specification.

Use this skill whenever the user wants to visualize neuroimaging analysis results, including 3D brain connectivity networks, atlas-based regional activation summaries, or FreeSurfer cortical surface meshes with anatomical colors. Triggers include: 'brain visualization', 'visualize connectome', '3D brain network', 'zALFF visualization', 'brain activation map', 'FreeSurfer PLY export', 'surface mesh rendering', or any request to turn neuroimaging outputs into interpretable figures or 3D models.

Use this skill whenever the user wants an end-to-end workflow for the Cam-CAN (Cambridge Centre for Ageing and Neuroscience) dataset, including BIDS validation, multimodal processing of sMRI, rs-fMRI, task-fMRI, and MEG, phenotype extraction, and QC integration. Triggers include: 'Cam-CAN', 'CamCAN', 'process Cam-CAN data', 'Cam-CAN MEG', 'Cam-CAN fMRI', or any request to run the Cam-CAN multimodal pipeline.

Use this skill whenever any NeuroClaw skill, sub-agent, or model needs to execute shell commands safely (e.g. source environment scripts, run recon-all, git operations, conda commands, ls, cat logs, etc.). Triggers include: 'run shell', 'execute command', 'shell command', 'tmux claw', 'run in claw session', 'safe shell execution', or any request that requires running terminal commands. This skill is the mandatory gatekeeper for all shell execution in NeuroClaw: it ALWAYS routes commands through the dedicated tmux session `claw`, never touches other sessions, and returns captured output to the calling agent.

Use this skill whenever the user wants to create, activate, list, export, update, clone, remove, or otherwise manage conda environments, or when a deep-learning / model skill requires a clean, isolated conda environment (e.g. 'create conda env for torch 2.3 cuda', 'export current env to yml', 'list all my conda envs', 'update packages in neuroclaw-dl', 'remove old env', 'clone env for reproducibility', 'install pytorch in new env'). Triggers include: 'conda create', 'conda env', 'make new environment', 'export yml', 'activate env', 'conda list envs', 'update conda env', 'clean environment', 'reproduce env', 'conda remove'. This skill is the mandatory gatekeeper for conda operations: it ALWAYS plans first, shows commands + risks + best practices, and waits for explicit user confirmation before executing anything.

Use this skill whenever the user wants to perform advanced functional connectivity (ROI-to-ROI, seed-to-voxel, ICA) or effective connectivity (PPI, gPPI, DCM) analysis using the CONN Toolbox. Triggers include: 'conn', 'CONN toolbox', 'functional connectivity', 'effective connectivity', 'ROI-to-ROI', 'seed-to-voxel', 'PPI', 'gPPI', 'DCM', 'psychophysiological interaction', or any request for connectivity analysis after preprocessing.

Use this skill whenever the user wants to convert DICOM files or folders to NIfTI format (.nii or .nii.gz), extract neuroimaging volumes from clinical DICOM series (MRI, CT, PET, etc.), prepare raw DICOM data for research processing pipelines, anonymize while converting, or batch-convert multiple series/studies. Triggers include: 'DICOM to NIfTI', 'dcm to nii', 'convert dicom to nii.gz', 'dcm2niix', 'extract nii from dicom', 'batch dicom to nifti', 'prepare dicom for freesurfer/fsl/spm', 'anonymized nifti conversion', or any request to transform clinical DICOM data into analysis-ready NIfTI format while preserving orientation, voxel spacing, slice timing (when available), and important metadata in the JSON sidecar.

Use this skill whenever a NeuroClaw skill, model, or sub-agent reports a missing dependency (e.g. ImportError, ModuleNotFoundError, command not found), or when the user explicitly requests to install, setup, configure, or fix any library, package, compiler, CUDA toolkit, conda environment, system tool, or git-based repository. Triggers include: 'install', 'setup', 'missing dependency', 'fix import error', 'install torch cuda', 'conda create environment', 'pip install from git', 'install nnU-Net', 'setup gcc nvcc', 'prepare environment for deep learning', 'handle dep error', or any phrase indicating the need to prepare or install software components. This skill is the **mandatory gatekeeper**: it ALWAYS plans first, never installs anything without explicit user confirmation.

Use this skill whenever any NeuroClaw diffusion MRI / DWI modality skill needs to execute concrete DIPY operations: load DWI (NIfTI+bvals+bvecs), optional masking, DTI fitting, compute FA/MD/AD/RD, and extract ROI statistics. This is the dedicated base/tool skill that contains all specific DIPY code and usage patterns. Never called directly by the user.

Use this skill whenever the user wants to pull, run, build, compose, list, prune, manage, or otherwise handle Docker containers, images, volumes, networks, or Docker Compose projects, or when a NeuroClaw skill (e.g. wmh-segmentation, freesurfer-processor in container mode) requires a clean, isolated, GPU-enabled Docker environment (e.g. 'pull mars-wmh image', 'run container with GPU', 'docker compose up', 'prune unused images', 'build custom dockerfile', 'manage nvidia docker'). Triggers include: 'docker run', 'docker pull', 'docker compose', 'docker build', 'docker env', 'manage container', 'nvidia docker', 'pull image', 'docker prune', 'containerize'. This skill is the **mandatory gatekeeper for all Docker operations** in NeuroClaw: it ALWAYS plans first, shows commands + risks + best practices, and waits for explicit user confirmation before executing anything. All actual Docker execution is routed through `claw-shell`.

Use this skill whenever the user wants to preprocess diffusion MRI / DWI data, compute diffusion metrics (FA/MD/AD/RD, etc.), extract ROI-wise diffusion features, or run tractography/connectome-related workflows. Triggers include: 'DWI', 'DTI', 'diffusion MRI', 'FA', 'MD', 'AD', 'RD', 'eddy', 'topup', 'QSIPrep', 'tractography', 'connectome', 'TBSS', 'white matter microstructure'. This is the NeuroClaw modality-layer interface: it plans WHAT to do and delegates execution to tool skills.

Use this skill whenever the user wants to load, preprocess, epoch, filter, or extract features from EEG data (resting-state, task-based, BCI, clinical, motor imagery, emotion, epilepsy, fatigue, etc.). Triggers include: 'eeg', 'EEG preprocessing', 'EEG feature extraction', 'band power', 'downsample to frequency bands', 'motor imagery BCI', 'emotion EEG', 'epilepsy detection', or any request involving .set/.edf/.bdf/.fif/.bids files.

Use this skill whenever the user wants to execute experiments based on a finalized method and record results. Triggers include: 'run experiment', 'experiment controller', 'implement experiment', 'run model', 'execute training', 'experiment-controller', 'record results', 'ablation study', or any request to turn METHOD.md into concrete runs and output to EXPERIMENT.md. This skill is the **mandatory interface-layer experiment executor** in NeuroClaw: it searches literature/GitHub for matching experimental setups and codebases, proposes one scheme + repo after user discussion, uses git skills to download and setup, runs the experiment(s), and iteratively appends every result + observation to EXPERIMENT.md.

Use this skill whenever the user wants to perform fMRI preprocessing, first-level analysis, ROI extraction, functional connectivity, effective connectivity, or atlas-based alignment to MNI152 space using either fMRIPrep, HCP-style pipelines, or CONN Toolbox. Triggers include: 'fmri', 'fMRI analysis', 'functional connectivity', 'effective connectivity', 'ROI extraction', 'seed-based correlation', 'PPI', 'DCM', 'atlas alignment', 'MNI152', 'HCP pipeline', 'CONN toolbox', or any request involving BOLD data.

Use this skill whenever the user wants to process structural MRI data (T1w, T2w, FLAIR, etc.) with FreeSurfer, especially for cortical/subcortical segmentation, surface reconstruction, parcellation, cortical thickness, volume statistics, or full recon-all pipeline. Triggers include: 'freesurfer', 'recon-all', 'segment MRI', 'FreeSurfer processing', 'cortical segmentation', 'subcortical segmentation', 'run recon-all', 'freesurfer T1', 'process brain MRI with freesurfer', 'aseg aparc', or any request to run FreeSurfer on NIfTI MRI data for research analysis.

Essential Git commands and workflows for version control, branching, and collaboration.

Advanced git operations beyond add/commit/push. Use when rebasing, bisecting bugs, using worktrees for parallel development, recovering with reflog, managing subtrees/submodules, resolving merge conflicts, cherry-picking across branches, or working with monorepos.

Core harness library providing standardized self-verification, checkpoint management, drift detection, and audit logging utilities for all NeuroClaw skills. This is NOT directly called by users; instead, it is imported as a Python module by other skills for harness-compliant execution, validation, and reproducibility. Use this as a foundation/plugin SDK when building or enhancing other skills. Triggers: none (library import only). This skill provides: HarnessController class, VerificationRunner, CheckpointManager, DriftDetector, AuditLogger, DependencyManifest, and related utilities.

Use this skill whenever the user wants to perform high-quality, HCP-style preprocessing of multimodal MRI data (structural, functional, diffusion) using the official HCP Pipelines. Triggers include: 'HCP pipeline', 'HCP preprocessing', 'hcp-fmri', 'hcp-dwi', 'hcp-structural', 'MSMAll', 'ICA-FIX', 'bedpostx', 'probtrackx', or any request to run the Human Connectome Project preprocessing pipelines.

Use this skill whenever any NeuroClaw modality skill (especially eeg-skill) needs to execute concrete MNE-Python operations for EEG loading, preprocessing, filtering, artifact removal, epoching, frequency-band analysis, or feature extraction. This is the dedicated base/tool skill that contains all specific MNE-Python code and usage patterns.

Multi search engine integration with 17 engines (8 CN + 9 Global). Supports advanced search operators, time filters, site search, privacy engines, and WolframAlpha knowledge queries. No API keys required.

Use this skill whenever NeuroClaw needs concrete nibabel operations for neuroimaging files: loading and validating NIfTI images, inspecting shapes and affine matrices, saving derived images, converting voxel coordinates to MNI/world coordinates, or reading FreeSurfer geometry and annotation files. Triggers include: 'nibabel', 'inspect NIfTI', 'read affine', 'save nifti', 'voxel to MNI', 'atlas coordinates', 'read FreeSurfer surface', 'read annot', or any request focused on low-level neuroimaging I/O rather than full preprocessing.

Use this skill whenever the user wants an end-to-end workflow for the Neuroimaging in Frontotemporal Dementia (NIFD) dataset, including BIDS validation, multimodal processing of sMRI, rs-fMRI, and dMRI, phenotype extraction, and QC integration. Triggers include: 'NIFD', 'frontotemporal dementia', 'FTD', 'bvFTD', 'PPA', 'process NIFD data', or any request to run the NIFD multimodal pipeline.

Use this skill whenever the user wants to convert NIfTI files (.nii or .nii.gz) to DICOM format, create DICOM series from processed neuroimaging results, write segmentation/registration/analysis outputs back to DICOM for PACS compatibility or clinical viewer comparison, or transfer metadata from reference DICOM files. Triggers include: mentions of 'NIfTI to DICOM', 'nii to dcm', 'convert nii.gz to DICOM', 'dicomify segmentation', 'nii2dcm', 'bring results back to DICOM', 'create DICOM from NIfTI', 'nii to dicom series', or any request to take post-processed neuroimaging results (segmentation, registration, bias field correction, synthesis, etc.) and store/view them alongside original patient DICOM data. Also use when modality-specific metadata (especially MR, SVR) or preservation of patient/study information from a reference DICOM is needed. Do NOT use for the reverse conversion (DICOM to NIfTI), non-medical imaging file conversions, or any clinical diagnostic or treatment-related workflows.

Use this skill whenever any NeuroClaw fMRI modality skill needs to execute concrete Nilearn operations: ROI/atlas time-series extraction, confounds handling (fMRIPrep), seed-based connectivity maps, ROI-to-ROI connectivity matrices, and optional GLM/decoding utilities. This is the dedicated base/tool skill that contains Nilearn usage patterns and lightweight wrappers. Never called directly by the user.

Use this skill whenever the user wants to synchronize NeuroClaw-generated LaTeX manuscripts with Overleaf, read/write .tex files, download/upload projects, create/rename/archive projects, compare versions, or manage project structure. Triggers include: 'sync to Overleaf', 'upload paper to Overleaf', 'Overleaf project', 'LaTeX sync', 'push draft', 'download Overleaf', 'create Overleaf project', 'tex file to Overleaf', or any request involving paper_draft.tex / collaboration. This skill is the **mandatory interface-layer LaTeX collaborator** in NeuroClaw: it strictly enforces pull-first workflow with diff reporting and per-operation user authorization for any write/create/delete action, uses pyoverleaf (cookie-based), preserves Overleaf version history, integrates directly after paper-writing, and never performs unauthorized modifications.

Use this skill whenever the user wants to generate a full academic paper draft from existing research materials. Triggers include: 'write paper', 'generate manuscript', 'draft paper', 'paper-writing', 'hierarchical drafting', 'manuscript composer', 'create LaTeX paper', 'write research paper from IDEA METHOD EXPERIMENT', or any request to transform IDEA.md + METHOD.md + EXPERIMENT.md into a typeset-ready manuscript. This skill is the **mandatory interface-layer writer** in NeuroClaw: it strictly follows the hierarchical manuscript drafting and iterative refinement process (section 4.4 + provided flowchart), never generates the full paper in one shot, saves every intermediate step as a separate file, and produces either clean plain-text or LaTeX output.

Use this skill whenever the user wants to run QSIPrep (BIDS App) for diffusion MRI (DWI) preprocessing with best-practice workflows (topup/eddy, denoising/unringing options, susceptibility/motion correction, coregistration/normalization, QC reports) on BIDS datasets. This skill is the NeuroClaw interface-layer wrapper for QSIPrep: it checks installation (Docker/Singularity/conda), generates an execution plan with exact commands and resource estimates, waits for explicit confirmation, then routes all execution through claw-shell.