Claire's best Open WebUI Functions
| Function | Features |
|---|---|
| token-saver-reloaded | - Only sends the last N messages to the provider to save LLM costs and time - Injects real timestamps to current and historical messages so the AI knows exactly what time it is - Did it trim too much? Re-discover forgotten messages via Semantic (PyTorch) or BM25 search - Strips image attachments from older messages to save multimodal token bandwidth - Regex that removes filler words and corporate jargon from user prompts without breaking code blocks |
| concise-output | - Injects system-level instructions that push the LLM toward concise, structured output - Sets a soft target word count and enforces clean bullet points - Suppresses preambles (e.g., "As an AI") and filler phrases |
Requirements: Open WebUI v0.8.4 or higher.
-
Navigate to Admin Panel → Functions → Import From Link.
-
Paste the preferred raw URL:
- Token Saver:
token-saver-reloaded.py - Concise Output:
concise-output.py
- Token Saver:
-
Enable the filter.
MIT License © 2026 fractuscontext