Banger paper from BAAI.

If you are building research agents, this one is worth your time.

(bookmark it)

They find that adding skills scores 134.3% higher on MLE-bench, 34.4% higher on PaperBench, 9.2% higher on FrontierCS and 14.0% higher on PassNet.

More details on the approach:

The agent has a strong backbone and a harness for planning, execution, memory and verification, and it still does not know how to make a given method actually work.

That know-how lives in repositories and papers, written for human readers and far too large to load during a task.

DisCo distills it. Task-agnostic distillation condenses 1,000 widely used ML repositories into the AREX-Skill Library, over 5,000 verified skills organized into 20 areas and 178 capability families. Task-oriented distillation writes the skills a concrete task calls for.

Paper: https://arxiv.org/abs/2609.02749

Chat with Paper: https://academy.dair.ai/papers/repo-to-skill-distilling-github-repositories-into-ai4ai-skills-2609.02749