Fine-Tuning Is Often the Wrong First Move: Introducing Harneloop
An open-source framework for artifact-aware, evidence-gated agent harness evolution
Jul 23, 20264 min read3
Command Palette
Search for a command to run...
Latest articles
What We Learned From Analyzing 28,000 Production AI System Prompts
80% start with "Act as an expert." 62% contain contradictory constraints. 14% emotionally threaten the AI. Here's the data.
Mar 24, 20262 min read8
AI Format Wars: Does the Shape of Your Prompt Matter? (1,080 Evals Later)
We tested 5 frontier models across 1,080 evaluations. JSON won. Short prompts won. Everything you thought you knew about prompt engineering is wrong.
Mar 22, 20265 min read36