VanadisGithub/dsh-skill-evolution ↗★ 0
dsh-skill-evolution
DSH web plugin: Hermes-style skill self-evolution. Watches agent execution traces, crystallizes reusable skills from successful turns via LLM review, progressively improves them, and manages everything in a dedicated Settings section.
安装
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:VanadisGithub/dsh-skill-evolution说明文档
阅读完整 README ↗配置
| 键 | 默认 | 说明 |
|---|---|---|
minTraceSteps | 3 | 轨迹最小步数（噪音地板） |
minToolCalls | 5 | complex 信号阈值 |
minPatternOccurrences | 3 | repeated 信号阈值 |
minSuccessRate | 0.7 | repeated 信号成功率下限 |
signals | 全开 | {complex, recovered, repeated} 分别开关 |
autoRegister | true | 结晶后自动注册进技能目录 |
llmProvider / llmModel | deepseek / deepseek-chat | 评审用的模型（必须是部署中存在的 provider） |
maxEvolvedSkills | 20 | 进化技能容量上限 |
improvementEvery | 3 | 每 N 次同序列重跑触发一次改进评审 |
maxImprovementsPerProcess | 5 | 每进程每技能改进次数上限 |
maxEvidenceSteps | 30 | 送评审的最大步骤数 |