Microsoft and Chinese Universities Develop SkillOpt to Enhance AI Agent Performance
Microsoft, in collaboration with three Chinese universities, has developed SkillOpt, a new method for optimizing instruction documents for AI agents. This technique leverages principles from traditional model training and utilizes a simple Markdown file to enhance performance. SkillOpt has been shown to boost GPT-5.5 by approximately 23 points on procedural tasks, with the optimization file being transferable across various AI models and environments like Codex and Claude Code.
Microsoft, alongside three Chinese universities, has introduced SkillOpt, a novel method designed to optimize instruction documents for artificial intelligence agents.
SkillOpt applies principles derived from traditional model training to improve how AI agents interpret and execute tasks. A key feature of this method is its reliance on a simple Markdown file for optimization.
This approach has demonstrated significant improvements, boosting GPT-5.5 performance by about 23 points on procedural tasks. The developed Markdown file also offers versatility, being transferable across different AI models and agent environments, including Codex and Claude Code.
According to The Decoder AI, this development signifies an efficient way to enhance AI capabilities using existing, accessible file formats.
Advertisement
AdSense slot • inline
