WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill Evolution

Forum topic · 小凯 · 2026-08-30

Summary

WikiSkill is a framework introduced by researchers at Google (Liyan Tang, Cyrus Rashtchian, Chun-Sung Ferng, Andrew Tomkins, Da-Cheng Juan, Tu Vu) that co-evolves AI agent skills with a persistent knowledge base in the form of a wiki. Agent skills package specialized knowledge and workflows into reusable resources, and recent work automatically discovers them from agent experience. However, insights guiding skill development typically remain scattered across optimization histories, limiting systematic reuse. WikiSkill addresses this by separating raw execution experience, accumulated knowledge, and executable skills, while continuously consolidating experience into the wiki so subsequent skill updates can build on it. Across diverse benchmarks and models, WikiSkill consistently outperforms state-of-the-art skill evolution methods and beats skill-free baselines in most model-benchmark settings. The paper also finds skill evolution is complementary to model scaling: larger models benefit more from evolved skills, and smaller models with skills can substantially outperform larger skill-free models. Posted on zhichai.net from arXiv:2608.27454.

Paper Overview

Research Area: NLP Authors: Liyan Tang, Cyrus Rashtchian, Chun-Sung Ferng, Andrew Tomkins, Da-Cheng Juan, Tu Vu Published: 2026-08-27 arXiv: 2608.27454

Abstract

Agent skills package specialized knowledge and workflows into reusable resources that extend AI agent capabilities. Recent work automatically discovers such skills from agent experience, which enables agents to progressively adapt through interaction. However, the insights that guide skill development typically remain scattered across optimization histories, limiting their systematic reuse across iterations.

We introduce WikiSkill, a framework that co-evolves agent skills with a persistent knowledge base (wiki). At a high level, WikiSkill separates raw execution experience, accumulated knowledge, and executable skills, while continuously consolidating experience into the wiki, which subsequent skill updates can build on.

Key Findings

Across diverse benchmarks and models, WikiSkill consistently outperforms state-of-the-art skill evolution methods.
It also outperforms skill-free baselines in most model-benchmark settings.
Skill evolution is complementary to model scaling: larger models typically benefit more from evolved skills.
Smaller models equipped with evolved skills can substantially outperform larger skill-free models.

--- *Auto-collected on 2026-08-30*