llm-toolbox
Public
Give local models tools for files, the live web, research, screenshots, audio, video, memory, and todos — without unrestricted shell access.
Install oihimekpen/llm-toolbox from the LM Studio Hub, then enable it on a chat.
| Family | Tools |
|---|---|
| System | ping, get_local_datetime |
| Files | workspace_info, list_dir, read_file, write_file, replace_text_in_file, delete_file |
| Web | web_search, fetch_url |
| Research | search_literature, literature_review_brief |
| Screenshots | take_screenshot, describe_screenshot, , , |
Give local models tools for files, the live web, research, screenshots, audio, video, memory, and todos — without unrestricted shell access.
Install oihimekpen/llm-toolbox from the LM Studio Hub, then enable it on a chat.
| Family | Tools |
|---|---|
| System | ping, get_local_datetime |
| Files | workspace_info, list_dir, read_file, write_file, replace_text_in_file, delete_file |
| Web | web_search, fetch_url |
| Research | search_literature, literature_review_brief |
| Screenshots | take_screenshot, describe_screenshot, , , |
screenshot_applist_screenshotsanalyze_image_frames| Audio | list_workspace_audio, get_audio_info, transcribe_audio, analyze_audio, extract_audio_clip |
| Video | get_video_info, extract_video_frames, analyze_video_with_vision |
| Memory | load_user_context, profile / facts / rules |
| Todos | read_todos, add_todo, complete_todo, remove_todo |
| Accuracy | accuracy_rules, groundedness_check, verify_claim, record_correction |
| Prompt prep | analyze_prompt, preprocess_prompt |
Each family except system can be turned off in the plugin settings for that chat.
whisper CLI if you have one installed. Video describe needs a vision-capable model loaded in LM Studio.~/.llm-toolbox/ (0700 / 0600).Media tools accept workspace files. You can pass an exact path to an audio/video/image file when that setting is on. The plugin does not scan Voice Memos, Desktop, or Downloads.
There is no shell, browser automation, Notes, or Calendar access.
npm install lms dev
npm test
MIT © 2026 Andrew Ihimekpen
screenshot_applist_screenshotsanalyze_image_frames| Audio | list_workspace_audio, get_audio_info, transcribe_audio, analyze_audio, extract_audio_clip |
| Video | get_video_info, extract_video_frames, analyze_video_with_vision |
| Memory | load_user_context, profile / facts / rules |
| Todos | read_todos, add_todo, complete_todo, remove_todo |
| Accuracy | accuracy_rules, groundedness_check, verify_claim, record_correction |
| Prompt prep | analyze_prompt, preprocess_prompt |
Each family except system can be turned off in the plugin settings for that chat.
whisper CLI if you have one installed. Video describe needs a vision-capable model loaded in LM Studio.~/.llm-toolbox/ (0700 / 0600).Media tools accept workspace files. You can pass an exact path to an audio/video/image file when that setting is on. The plugin does not scan Voice Memos, Desktop, or Downloads.
There is no shell, browser automation, Notes, or Calendar access.
npm install lms dev
npm test
MIT © 2026 Andrew Ihimekpen