How it works:

• Select the generator from the top menu of models.
• Enter the command and part of the old chat name.
• Examples: $$$script, recall:project #67, summa:n8n, 合計 999, ---черновик дипломной работы
• The plugin will find the file, clear it of AI "thoughts" and produce a compressed summary.
• Config: You can specify any custom prefixes in the config, with separator (e.g., summa:;$$$;リコール:;recall:;---)
• The plugin doesn't work out of the box. You'll need to at least enter a custom command in the config.
• LLM model for summary should already be loaded. Otherwise, the console will return an error.

There's no automatic model loading yet, even though the field exists in the config.
The generator uses any LLM model that is already loaded.

PS: for Windows only yet, path to .lmstudio/conversations/ hardcoded
Who knows how it is in the Linux/MacOS you could fork and fix it.
or I'll fix it later.

Known issues:

Since the chat name (in LM Studio) is generated automatically, the new chat will have the same name as the previous one, since the chat starts with the same words. You'll have to rename it manually. For your convenience. Sometimes it shows token:0. This isn't an error; this value is stored within the chat. The chat files aren't perfect. Ideally, the size of the annotation should not be fixed, but rather depend on the size of the source text. (7-10%-25) Summarizing System prompt should be in file.md - I know it.