When Skill Evolution Means Removing Instructions

Описание и применение

Как эволюционировать навыки агента, уменьшая инструкции и используя рабочий процесс, разделяющий опыт, знания и исполняемые навыки, чтобы повысить устойчивость к изменениям модели.

Задача

How to evolve agent skills without overloading them with instructions and ensuring they remain effective across model changes.

Результат

The author’s smaller implement-execplan skill achieved similar quality with ~18% fewer tokens, and the implement-approved-plan skill demonstrated a deterministic, multi-role workflow that reduced reliance on flagship models.

Рабочий процесс

• 1. Experiment manually with the model to understand the problem.
• 2. Design a workflow that works reliably without a skill.
• 3. Capture deterministic knowledge into mechanisms (e.g., state ledger, role-based chats).
• 4. Consolidate only the necessary instructions into a skill.
• 5. Validate the skill against quality gates and real pull requests.
• 6. Iterate: remove redundant instructions, adapt to model changes, and keep uncertain experience outside the skill until proven useful.

Доказательства

In the first valid comparison, both versions passed the same quality gates, while the smaller version used about 18% fewer candidate input + output tokens.

Источник

The skill itself coordinates a multi-chat workflow inside ChatGPT Desktop.

Источник