Next tasks

1. Field synchronization (optional)

Problem: When selecting a model from the dropdown, the value is NOT copied to the Custom Model ID field.

Possible solutions:

• Leave as is (user can use either list or manual field)
• Add "Copy to Custom Field" button next to dropdown
• Research LM Studio SDK for onUpdate hooks

Status: Not critical, works as-is

2. Empty cache handling

Problem: If models-cache.json doesn't exist or is outdated, the model list is empty on first launch.

Possible solutions:

• Show "Loading models..." indicator in UI
• Block model selection until cache loads
• Add "Refresh Models" button to interface

Status: Needs work

3. Only Free Models filter

Problem: The onlyFreeModels flag currently doesn't affect the UI list — only free models are always shown.

Possible solutions:

• Recreate schema with full/filtered list when flag is toggled
• Or load all models and filter on the fly

Status: Needs work

4. API error handling

Problem: There's no explicit user notification when model loading fails.

Possible solutions:

• Add error output to chat log
• Show fallback model with warning
• Add "Last Error" field to config

Status: Partially implemented in logs

5. Model display customization

Problem: generateDisplayName() may incorrectly format some model names.

Possible solutions:

• Add display format settings
• Show full model name in tooltip
• Add search/filter in dropdown

Status: Works acceptably

Known LM Studio SDK limitations

• No dynamic schema update — can't change options after schema creation
• No onUpdate hooks for config fields (or documentation is incomplete)
• Validation at getPluginConfig() — value must be in options for type "select"

Development commands

# Build TypeScript
npm run build

# Run in development mode (install in LM Studio)
npm run dev

# Push updated version
npm run push

Install plugin locally, without pushing to Hub

lms dev --install -y

File structure

D:\LMS-src\o!-endpoint\
├── src/                      # TypeScript source
│   ├── api.ts               # API calls (fetchModels)
│   ├── config.ts            # Default config schema
│   ├── file-cache.ts        # Model caching
│   ├── generator.ts         # Response generation (uses string schema)
│   ├── index.ts             # Entry point, cache loading
│   ├── models.ts            # Model loading logic
│   └── schema.ts            # Schema creation with options (select)
├── dist/                     # Compiled .js files
├── .lmstudio/
│   ├── entry.ts             # Entry point for LM Studio
│   └── dev.js               # Bundle from lms dev
├── models-cache.json         # Model cache (created on startup)
└── package.json

Contacts / Resources

• LM Studio SDK: @lmstudio/sdk v1.4.0
• OpenRouter API: https://openrouter.ai/api/v1
• LM Studio documentation: (needs to be found)