Agent skills

Related

Organizations

East China Normal UniversityGoogleShanghai AI Lab

Techniques

Agent memoryCredit assignmentKnowledge distillationMulti-agent orchestrationOn-policy distillationPrompt optimization

Trends

• Agents improve themselves through skills, harnesses, and memory
• Agent reliability becomes a funded tooling and monitoring layer
• Agent post-training becomes predictable engineering

Items

1. Sep 11

   Skills work better as fresh-context subagents

   An empirical study finds that invoking reusable skill packages as fresh-context subagents beats loading skill instructions into the main context when packages expose clear input-output contracts, at the cost of extra coordination tokens.

2. Aug 28

   Google's WikiSkill co-evolves skills and knowledge

   An agent framework that co-evolves executable skills with a persistent wiki-style knowledge base, separating raw experience, distilled knowledge, and skills; evolved skills transfer across model families, so smaller models with skills outperform larger ones without them.

3. Aug 8

   OpenDataLab's SKILLER writes skills for small models

   A natural-language RL loop uses a strong model as actor and critic to generate executor-specific skills for small open models such as Qwen3.5-9B and 4B, adding up to 20.4 points and matching closed models on single-skill tasks.

4. Jul 25

   SkillRise curates evolving skills across agent tasks

   An agentic-RL framework in which one policy alternates between solving a task and curating a skill document passed to the next, with decoupled cross-task credit assignment adding 2.3-8.5 Pass@1 points on ALFWorld, WebShop, and ScienceWorld.

5. Jul 25

   DecoEvo co-evolves solver and rubric skills

   A text-space method that co-evolves a solver skill and a rubric-generator skill under decoupled objectives to avoid reward-easy rubric drift, beating prior text-space optimizers by 2.8-5.0% across five benchmarks.

6. Jul 17

   SEED self-evolves on-policy distillation

   Agents turn their own completed trajectories into natural-language hindsight skills, then distill the resulting probability shift into a dense token-level signal optimized jointly with outcome-based RL, improving performance and sample efficiency on text and vision tasks.

7. Jun 6

   Bayesian-Agent calibrates agent skill repair

   Treats reusable agent skills and SOPs as posteriors over whether a frozen model will succeed, driving inspectable patch, split, and retire actions; incremental repair lifts SOP-Bench from 80% to 95% and RealFin-Bench from 45% to 65%.

8. May 27

   ECNU's Skill0.5 for agentic RL

   The framework internalizes general skills through privileged distillation on hard tasks and applies task-specific skills on easy ones via a difficulty-aware selector, beating memory- and skill-based RL baselines on ALFWorld and WebShop in and out of distribution.