steipete/oracle: Multi-Model Context Bundler — Run Queries Against GPT-5 Pro, Gemini 3.1, and Claude Simultaneously

GitHub·medium signal

Oracle bundles prompts and source files into rich context packages, then sends them to multiple LLMs simultaneously (GPT-5.4 Pro, Gemini 3.1 Pro, Claude Sonnet 4.5) with aggregated cost reporting. Features multi-model comparison runs, conversation follow-ups via session IDs, browser automation for keyless usage, and file safety guards preventing accidental context bloat. At 1.8K stars, it's a practical tool for builders who want to compare model responses on real codebase questions.