self-optimization
SONA self-optimizing neural architecture with ReasoningBank trajectory learning, EWC++ anti-forgetting, and reinforcement learning feedback loops.
来源信息
- 仓库
- a5c-ai/babysitter
- 最近来源活动
- 2026年6月1日 07:46
- 检测到的 SKILL.md 语言
- 英语
- 星标
- 1,832
- 分支
- 113
安装方式
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
检查来源文件
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
文件资源管理器
2 个文件正在显示 SKILL.md
SKILL.md
来源说明 · 只读预览- name
- self-optimization
- description
- SONA self-optimizing neural architecture with ReasoningBank trajectory learning, EWC++ anti-forgetting, and reinforcement learning feedback loops.
- allowed-tools
- Read, Write, Edit, Bash, Grep, Glob, WebFetch, WebSearch, Agent, AskUserQuestion
- graph
- {"domains":["domain:software-engineering"],"skillAreas":["skill-area:agentic-loops","skill-area:orchestration-loop"],"workflows":["workflow:feature-development"],"topics":["topic:developer-experience"],"roles":["role:tech-lead","role:backend-engineer"]}
- Improving routing and agent selection over time
- Adapting to new project patterns without forgetting old ones
- Building cross-session intelligence
## SONA Cycle
1. **Extract Patterns** - Mine execution data for recurring patterns
2. **RETRIEVE** - Search ReasoningBank for matching trajectories
3. **JUDGE** - Evaluate trajectory applicability in current context
4. **DISTILL** - Compress and store new entries
5. **Adapt** - Update weights with EWC++ regularization
## Anti-Forgetting (EWC++)
- Elastic Weight Consolidation prevents overwriting previously learned patterns
- Fisher information matrix tracks parameter importance
- Configurable regularization penalty for new adaptations
## RL Algorithms
Q-Learning, SARSA, PPO, DQN, A2C, TD3, SAC, DDPG, Rainbow
## Agents Used
- `agents/optimizer/` - Performance tuning
- `agents/adaptive-queen/` - Real-time adaptation
## Tool Use
Invoke via babysitter process: `methodologies/ruflo/ruflo-intelligence`
在 GitHub 查看