← all repositories

stanford-iris-lab/meta-harness

A framework for automated search over task-specific model harnesses that control what LLMs store, retrieve, and display during execution.

1k stars Python AgentsLLMOps · Eval
meta-harness
Velocity · 7d
+19
★ / day
Trend
steady
star history

Meta-Harness provides a reusable framework for optimizing the scaffolding code that surrounds fixed base models in agent systems. It searches over harness configurations to find optimal combinations of memory, retrieval, and display mechanisms. The repository includes reference implementations for text classification harness optimization and Terminal-Bench 2.0 scaffold evolution, along with an onboarding process for applying the framework to new domains.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.