SkillZip: Evaluation-Free Skill Compression for Self-Evolving Agents by Discovering Reusable Structure

2026-08-11

AI Summary

This work addresses the problem of skill bloat in self-evolving agents, where redundant content accumulates during skill acquisition, leading to unwieldy skill representations, high maintenance overhead, and costly injection. To mitigate this, the paper introduces SkillZip—the first skill compression method that operates without task replay or performance evaluation. SkillZip employs a typed Minimum Description Length objective with coverage constraints to automatically identify and share reusable structures across skills, retaining only divergent components as exceptions. The approach integrates structured extraction, deterministic optimization, and an incremental Zip-on-Write mechanism to enable efficient local updates and residual modeling. Experimental results demonstrate that SkillZip significantly outperforms existing baselines in compression ratio, generalization capability, and computational efficiency, confirming its effectiveness and superiority.

Abstract

Self-evolving agents accumulate reusable skills by appending successful procedures and failure fixes. Over time, the same requirement is often restated in several branches, examples, and warnings, while common action sequences are copied rather than reused. The resulting skill becomes expensive to inject and difficult to maintain. Generic prompt compression is ill-suited to this setting because a skill is not a flat passage: its name and description define when it applies, its workflow controls execution, its tool and output contracts constrain validity, and rare exceptions may remain essential even when no sampled task activates them. Evaluation-guided compression can test these behaviors, but it introduces rollouts, cost, and dependence on the compression-time evaluation set. We present SkillZip, an evaluation-free method that compresses a skill by finding its shortest faithful structural explanation. The intuition is explain once, reference many: state a repeated rule once at the scope where it applies, factor a repeated action sequence into a shared procedure, and keep only the differences as explicit exceptions. We formalize this intuition as a typed minimum description-length objective over a skill contract and a residual, subject to a hard coverage constraint for every extracted trigger, workflow edge, tool requirement, obligation, and output field. The formulation provides simple sharing thresholds, preserves unique rare rules by construction, and supports efficient local updates. SkillZip has a one-shot mode with one structured extraction call and deterministic optimization, and a continual Zip-on-Write mode that integrates each self-evolution patch without replaying tasks or reparsing the full history. Through comprehensive experimental evaluations, we demonstrate the effectiveness and superiority of SkillZip in compression performance, generalizability, and cost overhead.

Problem

• skill compression
• self-evolving agents
• reusable structure
• evaluation-free
• structured representation

Innovation

• skill compression
• self-evolving agents
• structure discovery
• minimum description length
• evaluation-free