Use Python 3.10 or newer in your existing analysis environment.
python -m pip install auto-eeg-analysis==0.3.3
aea-eeg init ./my-eeg-study --agent codex
cd my-eeg-study
Use --agent claude for Claude Code or --agent both for both clients. The
command creates a new project containing 22 skills, 10 recipes, templates and
tools, then registers project-local skill links. Existing nonempty folders are
refused. Start your separately installed coding client inside the new project.
The default package has no scientific runtime dependencies. Reuse an existing
MNE environment. To create a new Conda environment from the copied specification,
run conda env create -f environment.yml and conda activate aeais in the project.
Alternatively install the existing analysis dependency set with
python -m pip install 'auto-eeg-analysis[analysis]==0.3.3'.
Read AGENT_GUIDE.md and approve a study-specific plan before computation.
The package version is separate from retained validation and plugin versions.
aea-eeg --version reports the resource source commit.
The PyPI page contains release files. Version 0.3.3 preserves the archive source-version record. The first rc1 remains available. Initial package checks and fixes are distribution checks and add no scientific validation.
Download the AEA 0.1.3 toolbox and install the .mltbx file in MATLAB.
The candidate was built and checked with Windows MATLAB R2026a Update 2.
Run aea_setup with the folder containing your eeglab.m. Existing AEA plugin
files are retained. Native MATLAB requires adapter 0.1.3 or newer. Restart
EEGLAB after a new plugin installation. In Tools > AEA > Settings, use the
path returned by aea_library_root as the AEA checkout folder.
Choose backend matlab, leave Python blank, and select your installed and
signed-in Codex CLI or Claude Code executable. Export a dataset to a new study,
approve the analysis plan, then open the saved report from EEGLAB.
The native helper covers explicit ERP means from existing epochs. The broader
resource library primarily uses MNE-Python. See native MATLAB coverage
and the toolbox setup guide.
The File Exchange listing provides the GitHub source.
Use the linked GitHub release for the tested .mltbx installer.
| Entry | Published version or source | What it contains |
|---|---|---|
| GitHub | main, changing with commits |
Current source and documentation. Record the commit used. |
| GitHub stable release | v0.3.2, 3 October 2026 |
Earlier fixed release. Later package and adapter changes have separate versions. |
| PyPI | 0.3.3 |
Resource installer and project initialization. Bundled resources come from 62944027e4484b051614dc2f1bd397db97f5823c. |
| EEGLAB catalogue | Adapter 0.1.3 |
EEGLAB menu extension. Connect it to a matching AEA resource checkout and coding client. |
| FieldTrip directory | External extension listing, merged 7 November 2026 | AEA entry linking to its independently maintained source and support. |
| MATLAB toolbox | 0.1.3 preview |
.mltbx containing the adapter and resource library, with installation helpers. |
| File Exchange | GitHub main source |
Source download and a link to the tested .mltbx. Its listing version can advance separately from the binary preview. |
| Zenodo | Snapshot 2026-11-06, published 7 November 2026 |
Frozen resources and validation records for the 600-session study and retained ERP evaluations. |
| Website | GitHub Pages from docs/ |
Current usage guidance and links to these releases and archives. |
FieldTrip accepted the directory entry in PR #968. The MNE directory request remains open as of 7 November 2026.
Software, adapter and archive versions identify different objects. The Zenodo record does not include later package builds or runs completed after its frozen snapshot. For the numerical and behavioural configurations actually checked, see validation coverage.
Use CITATION.cff for the software and record the actual release or source commit. Cite 10.5281/zenodo.24012489 when using that fixed resource and validation archive. The all-version DOI identifies the archive series. Recipe source papers and original datasets retain their own citation requirements.
Maintain source and documentation in GitHub. The website deploys from docs/,
and the current File Exchange connection follows the default branch. PyPI
packages, EEGLAB catalogue updates and .mltbx assets require their own release
steps. Build matching packages from the recorded source and update the links
and this version table when publishing them.
Keep published archives fixed. Add a new Zenodo version for a subsequent archival snapshot rather than describing the existing record as current source.