ScrapChat updates local LLM workbench with task chaining

ScrapChat's approach to context window management via pipeline steps is a clever workaround for local, resource-constrained models. "Taskmaster" explicitly resets the context window between chained steps, which is vital when running models like Qwen 27B/30B locally without massive VRAM. Deep financial integration with E*TRADE and Python-based calculation guarantees math accuracy, sidestepping common LLM hallucination issues in trading. The extensible architecture and lack of a complex frontend build step make it highly approachable for developers who want to self-host without Javascript fatigue.