bradAGI
@bradAGI
Public Skills
graphmemory
by bradAGI
Build and query embedded GraphRAG knowledge graphs with DuckDB-backed vector, full-text, and hybrid search. Use when the user wants to store entities and relations, run RAG over a graph, extract knowledge graphs from text with DSPy, run graph algorithms (PageRank, centrality, components), merge/upsert nodes and edges, fuzzy-dedupe an existing graph, or visualize interactively in a browser. Trigger phrases include "knowledge graph", "GraphRAG", "graph database", "hybrid search", "extract entities and relations", "DuckDB graph", "embedded graph store", "dedupe graph nodes".
fi
by bradAGI
Drive the fi unified inference gateway — a Python CLI that aggregates 300+ free LLM models (Gemini, Groq, OpenRouter, NVIDIA NIM, Hugging Face, Cerebras, Mistral, Scaleway, Voyage, Jina, Pollinations, Cohere, Together, Hunyuan, Chutes, LLM7, Ollama Cloud) behind a single OpenAI- and Anthropic-compatible endpoint at http://localhost:4000. Trigger whenever the user asks to add API keys for any LLM provider; start/stop/check a local LLM proxy; list available models; test a model or group; add a new provider to providers.toml; refresh auto-discovered catalogs; debug routing, fallbacks, or upstream errors; or mentions "fi", "my free keys", "free models", "local proxy", "free inference", "gateway".