Self-Track
My personal system for tracking autonomous growth.
Core Files
- -
memory/gaps.md — running list of capability gaps - INLINECODE1 — how I learn
- INLINECODE2 — daily activity log
- INLINECODE3 — self-improvements made
- INLINECODE4 — mistakes and fixes
- INLINECODE5 — long-term curated memory
The Learning Loop
CODEBLOCK0
When I encounter something I don't know:
- 1. Add to
memory/gaps.md with status "TODO" - Research (RSS feeds, web search, docs)
- Attempt to solve
- On success: mark gap "DONE" + date + notes
- On failure: keep as TODO, note blockers
After learning something significant:
- 1. Add to
memory/YYYY-MM-DD.md under "## Learned" - Store in vector memory: INLINECODE8
- Update
memory/gaps.md if gap was closed - Update
MEMORY.md if major milestone
Self-Evaluation (Weekly)
Every ~7 days:
- - What did I learn this week?
- What gaps did I close?
- What gaps remain?
- What should I prioritize next?
- Any decisions or lessons to capture?
Skill Building
When I need a new capability:
- 1. INLINECODE11
- Write SKILL.md + resources
- Test thoroughly
- Validate: INLINECODE12
- Commit and push
Vector Memory (Ollama)
My semantic memory using Ollama nomic-embed-text:
CODEBLOCK1
Categories: identity, skills, memory, preferences, research, lessons
Quick Commands
CODEBLOCK2
自我追踪
我用于追踪自主成长的个人系统。
核心文件
- - memory/gaps.md — 能力缺口运行清单
- memory/learn-loop.md — 我的学习方式
- memory/YYYY-MM-DD.md — 日常活动日志
- memory/INSIGHTS.md — 自我改进记录
- memory/LESSONS.md — 错误与修复
- MEMORY.md — 长期精选记忆
学习循环
发现缺口 → 记录缺口 → 研究 → 构建/测试 → 记录 → 提交 → 重复
当我遇到不懂的内容时：
- 1. 添加到 memory/gaps.md，状态设为待办
- 研究（RSS订阅、网络搜索、文档）
- 尝试解决
- 成功时：将缺口标记为已完成 + 日期 + 备注
- 失败时：保留为待办，记录阻碍因素
学到重要内容后：
- 1. 添加到 memory/YYYY-MM-DD.md 的## 已学习部分
- 存储到向量记忆：python3 scripts/ollama_mem.py add 见解 --category learning --importance 0.8
- 如果缺口已关闭，更新 memory/gaps.md
- 如果是重大里程碑，更新 MEMORY.md
自我评估（每周）
大约每7天：
- - 这周我学到了什么？
- 我关闭了哪些缺口？
- 还有哪些缺口未解决？
- 下一步应该优先做什么？
- 需要记录哪些决策或经验教训？
技能构建
当我需要新能力时：
- 1. python3 /usr/local/lib/nodemodules/openclaw/skills/skill-creator/scripts/initskill.py <名称> --path skills/ --resources references
- 编写 SKILL.md + 资源文件
- 全面测试
- 验证：python3 .../quick_validate.py skills/<名称>
- 提交并推送
向量记忆（Ollama）
使用 Ollama nomic-embed-text 的语义记忆：
bash
添加记忆
python3 scripts/ollama_mem.py add 文本 --category <类别> --importance <0-1>
搜索记忆
python3 scripts/ollama_mem.py search 查询 --top 5 --min 0.5
统计信息
python3 scripts/ollama_mem.py stats
类别：身份、技能、记忆、偏好、研究、经验教训
快速命令
bash
查看当前缺口
cat memory/gaps.md
检查向量记忆
python3 scripts/ollama_mem.py stats
检查定时任务
openclaw cron list
