A curated collection of high-utility, experimental tools for OpenWebUI. These tools are built to bridge the gap between LLM reasoning and real-world execution, focusing on automation, knowledge retrieval, and creative workflows.
Consensus-driven reasoning. Orchestrates a deliberation process among multiple models (e.g., GPT-4, Gemini, Claude). It features a 3-stage pipeline: individual responses, anonymous peer-ranking, and chairperson synthesis to provide high-confidence answers.
- Key Feature: Peer evaluation eliminates single-model bias.
Semantic music curation. Translates the mood and context of your chat into a curated Spotify playlist. Tell it how you feel or what you're working on, and it will build a matching soundtrack using semantic analysis.
- Key Feature: Automatic playlist generation based on conversational context.
The universal connector. Bridges OpenWebUI to the N8N automation ecosystem. Trigger complex workflows, interact with thousands of apps, and receive results directly in your chat interface.
- Key Feature: Session-aware executions that maintain context between your chat and N8N.
Long-term semantic memory. Provides agents with access to persistent knowledge bases. It handles embedding generation and semantic search to ground responses in your own data.
- Key Feature: Seamless integration with Pinecone for production-grade retrieval.
Last-mile productivity. Automates document creation by filling Google Doc templates via Google Apps Script. Perfect for generating invoices, proposals, or reports directly from a chat.
- Key Feature: Dynamic placeholder replacement with LLM-generated content.
My background is in Data Science and A/B testing, which drives a test-everything, ship-fast mentality. Each tool here is a specific exploration into "LLM Agency"—the ability for models to move beyond text and into action.
- Curiosity First: Solve a personal friction point or explore an API limit.
- Action & Shipping: Move from "what if" to "it works" quickly.
- Modular Design: Tools are self-contained and easy to port into any OpenWebUI instance.
Each tool is designed for easy installation:
- Pick a Tool: Navigate to the specific tool's directory (e.g.,
/llm_council_tool). - Copy the Code: Open the main Python file (usually
tool_name.py) and copy the content. - Install in OpenWebUI:
- Go to Workspace > Tools > + Create Tool.
- Paste the code and save.
- Configure Valves: Enter any required API keys (OpenAI, Spotify, etc.) in the tool's Valves configuration in the UI.
These tools are built through chronic experimentation. If you find a bug or have an idea for a new "vibe", feel free to open an issue or a PR.
Built with ❤️ by matheusbuniotto