Strands harness is a new open-source agent framework that achieves 28% lower token costs than competing solutions while maintaining equal or better accuracy across benchmarks. Available for Python and TypeScript, it runs locally or on cloud providers with built-in prompt caching, context management, and support for multiple model providers including Claude, GPT, and Deepseek.
An interactive tool demonstrates how AI coding agents like Claude Code work by replaying API requests and showing token consumption patterns. The article explains why sessions grow expensive through prompt caching mechanics and provides cost visualizations to help users optimize their usage patterns.
A multi-agent harness framework introduces context modes for subagents, allowing supervisors to either isolate subagents with fresh contexts or fork them with inherited conversation history. Forked subagents reuse the supervisor's context through prompt caching, reducing redundant work and costs, while isolated subagents provide context independence for specialized tasks like verification.