Yotta Learn — Development skill for Claude Code
元习是一个跨智能体的学习闭环工具.
How to install Yotta Learn
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open YottaMeta/yotta-learn and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Yotta Learn does
元习是一个跨智能体的学习闭环工具。它把命令失败、用户纠正、更好的做法与洞见,沉淀为结构化的 .learnings/ 条目,供后续会话与技能改进复用;支持 init / log / list / promote / review / stats / extract 等命令与 Pattern-Key 复发追踪,并附赠 OpenClaw / Claude Code / Codex 的 hook 模板。它让智能体「越用越懂你」,把每次踩坑与纠正都变成可检索、可回溯的长期资产。
Alternatives in Development
- Resume Work From A Handoff Document — You are tasked with resuming work from a handoff document through an interactive process 10k ★
- Promote Public Release — Build, refine, and publish cumulative public release notes, then promote the current alpha prerelease to stabl 1.6k ★
- Harness Improvement — Update Kandev harness files from session learnings or explicit cross-platform harness requests 697 ★
README
Language: English · 中文
yotta-learn · 元习
YottaMeta's cross-agent learning-loop skill: turns mistakes, corrections and insights into reusable .learnings/ entries for later sessions and skill improvement. Suited for command failures, user corrections, discovering a better practice, requesting a missing capability, external-interface failures, and stale knowledge.
Activates on command failure / user correction / a better approach / a missing capability / an external-interface failure / stale knowledge / a need to capture experience, or when the user says 记一笔 / learn / 沉淀 / self-improvement / learnings — judged by whether experience should be captured, not by keyword luck.
Python 3.8+ standard library, zero dependencies; Windows + Linux; init never overwrites existing .learnings/ data.
What it is
The most common waste for an AI agent is repea
Related Skills
Wp Dev Learn
Extract learnings from the current session and update wp-dev-skills skill files — captures corrections, new ru
Post Trade Review
Automatically run after every trade to log results, extract lessons, and update strategy stats. MUST run after
Self Improving Skill
Auto-learn writing rules from human edits. record original → record final → diff → extract rules → update SKIL
Harvest Memory
Promote durable learnings from the machine-local per-project memory store into the version-controlled repo mem
Aegis Distill
Knowledge distillation: compress learnings, promote patterns to resonance
Flux Agent
Manage flux agent registry — index/backfill/stats/prune/promote/record. Lifecycle for review agents with quali
Related Agents
Rayo Heartbeat Watcher
Check OpenClaw gateway health and alert if Rayo is dead or idle too long. Monitors ~/.openclaw/logs/gateway.lo
File Analyzer
Use this agent when you need to analyze and summarize file contents, particularly log files or other verbose o
Initialize Engagement At The Start
findings.sh init " " --client " " --type " " --scope " " findings.sh stats bash db/handoff.sh handoff_report.m