Plugins

Use a plugin

1. Open Settings → Plugins. Each plugin is one row with its name, a one-line description and an Enabled switch, grouped by where it comes from: Built-in, NeuroTerm official and My local.
2. Choose ⚙ on a row to open the plugin’s own page: its description, demo images, settings and Data & Privacy.
3. Where it appears lists every place the plugin adds something, and How to run it says how to open each one.

For example, NeuroFlash log search is a plugin. Its page lists, under Where it appears, Search logs… in the Terminal menu and in the selection menu, an Ask mode in the find bar and a NeuroFlash panel under the terminal. Its settings, such as the OpenRouter API key, are on the same page. See NeuroFlash log search.

Build one with your agent

Claude Code and Codex can build a plugin for you: a panel that counts errors per boot, a decoder for your own serial protocol or a marker for your board’s crash pattern. You do not need to know the plugin format.

1. Set up Terminal Use: the agent builds plugins with its plugin_ tools.
2. Start claude or codex in a NeuroTerm Local or WSL tab and ask for the plugin. While My local in Settings → Plugins is empty, it lists starter prompts with Copy and Start; Start hands the prompt to an agent in one of your tabs. For example: “Build a NeuroTerm plugin with a log panel that counts error lines per boot of my board. Start with plugin_guide.”
3. The agent reads plugin_guide, writes the project with plugin_scaffold, checks it with plugin_validate and runs it with plugin_apply. The folder is the plugin: there is no build step.
4. Apply shows a consent dialog in NeuroTerm, “Allow this Local plugin to run?”, with the plugin’s name, its folder and each permission it asks for. Choose Allow or Deny; only you can answer it.

The plugin then runs under My local, without a restart. You can also apply a plugin folder yourself: Apply local plugin… on the My local header picks a folder and shows the same consent dialog.

What a Local plugin may do

A Local plugin is made on your computer and is not signed by NeuroTerm. It runs sandboxed, with only the permissions you allowed. Apply asks again when the plugin asks for more permissions, when it is applied from a different folder, and when NeuroTerm stopped it for not responding.

A plugin that writes to your tabs asks in the tab, the same way an agent does. Its page has Allow its writes automatically; serial line changes are always asked.

Change or remove it

• Apply again reads the plugin’s source folder again and applies it.
• Remove plugin deletes its snapshots, consent and stored data; the source folder stays.
• Your own version copies a plugin whose licence allows changes to a folder under a new id, to edit and apply.
• Ask the agent to read the plugin’s log with plugin_logs, or to save a demo image of its view with plugin_capture.