Harness
A Harness denotes the orchestration framework or engineering layer that wraps a large-language-model (LLM) to control execution flow, manage context, and integrate external tools. Performance differentiation is now dominated by harness design rather than base model selection.
Core Insights
- Orchestration Priority: Stanford research indicates that LLM performance variance is primarily determined by orchestration code quality, superseding architectural improvements. Orchestration Over Architecture: Harness Engineering for Optimal LLM Performance
- Engineering Shift: Development focus moves from weight training to structural control, implementing robust [[entities/
Coding Agent Architectures
The design of the harness significantly dictates the utility of coding agents, creating a spectrum between minimalism and comprehensive integration:
- Bare Core vs. Batteries Included: Recent analysis highlights a critical divergence in harness design for coding agents. The “Bare Core” approach (exemplified by Pi) focuses on minimal overhead and direct model interaction, whereas “Batteries Included” frameworks (exemplified by Claude Code) provide extensive pre-configured tooling, context management, and workflow automation out-of-the-box.
- Performance Impact: The choice between these architectures impacts developer experience and model efficacy. A well-designed harness can mitigate model limitations, while a poor one can bottleneck even the most capable LLM.
- Detailed Analysis: For a comparative breakdown of these specific implementations, see AI Coding Agent Harnesses: Bare Core (Pi) vs. Batteries Included (Claude Code).