Add EEGLAB agent workflow setup

[suraj-ranganath]

Add AGENTS.md, shared Claude/Codex skills, and GitHub workflows for EEGLAB agent development. Keeps CLAUDE.md delegated to AGENTS.md and adds MATLAB/Octave smoke coverage for the workflow setup.

[arnodelorme]

Very good. The CLAUDE.md file targets EEGLAB users who want to vibe code with EEGLAB. When you tell Claude to use EEGLAB for a task, it will check that CLAUDE.md is in the folder and use it. I doubt it will check the other files. Those are relevant when people work directly in the EEGLAB folder, where they will be visible to agents, but likely not when EEGLAB is used merely as a tool. Claude tends to agree with me. Let me know what you think.

[suraj-ranganath]

https://github.com/sccn/eeglab/blob/codex/eeglab-agent-setup/AGENTS.md?plain=1#L28

This ensures that if we add skills later under .agents/skills/ that are useful for users who want to vibe code with EEGLAB as a tool, those skills will be picked up by the agent they are using.

Right now, however, we only have skills that are used by the EEGLAB workflows. This can be extended by simply adding more skills there, written by expert EEGLAB users or domain experts.