메뉴
维护 .learnings/ 经验库，把过多或反复出现的学习记录、错误日志、规则失效问题聚类诊断，追溯并修改对应 skill、模板、hook、校验脚本或项目规则；修复并验证后再归档或移除已解决记录；同时检查 Codex .agents/skills 与 Claude Code .claude/skills 的共享功能同步。用户提到 learnings 太多、错误反复犯、清理经验库、维护自我学习、压缩错误日志、从错误中修技能、同步 Codex 和 Claude Code 技能时触发。
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when writing an OpenCLI adapter for a new site or adding a new command to an existing site. Guides end-to-end from first recon through field decoding, adapter coding, and verify. Replaces opencli-oneshot / opencli-explorer. For ad-hoc browser driving (no adapter), see opencli-browser instead; for a top-level orientation to opencli, see opencli-usage.
Automatically fix broken OpenCLI adapters when commands fail. Load this skill when an opencli command fails — it guides you through diagnosing the failure via OPENCLI_DIAGNOSTIC, patching the adapter, retrying, and filing an upstream GitHub issue after a verified fix. Works with any AI agent.
Use when an agent needs to drive a real Chrome window via opencli — inspect a page, fill forms, click through logged-in flows, or extract data ad-hoc. Covers the selector-first target contract, compound form fields, stale-ref handling, network capture, and the agent-native envelopes the CLI returns. Not for writing adapters — see opencli-adapter-author for that.
Use at the start of any OpenCLI session — this is the top-level map of what `opencli` can do, how to discover adapters, what flags and output formats are universal, and which specialized skill to load next. Point here when an agent asks "what can opencli do?" or "how do I find the right command?".
This skill should be used when the user asks to sync study progress with MemOS, especially `/sync 拉取` to read the latest cross-device context or `/sync 上传` to save today's learning state from local notes.
整理考研数学/专业课章节笔记，生成结构化总结文件。**触发词**："章节总结"、"整理章节"、"汇总这一章"、"章节笔记"、"数学���记总结"、"整理数学笔记"、"做一个总结"、"总结一下这一章"、"帮我总结"、"写个XX的章节总结"。功能包括：提取重要定义定理公式、保留个人理解、按章节结构组织、生成📝章节总结.md文件。**支持科目**：考研数学、数字电子技术、模拟电子技术等。
| name | maintain-learnings |
| description | 维护 .learnings/ 经验库，把过多或反复出现的学习记录、错误日志、规则失效问题聚类诊断，追溯并修改对应 skill、模板、hook、校验脚本或项目规则；修复并验证后再归档或移除已解决记录；同时检查 Codex .agents/skills 与 Claude Code .claude/skills 的共享功能同步。用户提到 learnings 太多、错误反复犯、清理经验库、维护自我学习、压缩错误日志、从错误中修技能、同步 Codex 和 Claude Code 技能时触发。 |
让 .learnings/ 保持小而有用：它是发现流程缺陷的雷达，不是永久堆放错误的仓库。重复错误必须优先修源头，验证有效后再清理活跃记录。
先读取：
.learnings/RULES.md.learnings/LEARNINGS.md.learnings/ERRORS.md然后运行审计脚本。Codex 示例：
python3 .agents/skills/maintain-learnings/scripts/audit_learnings.py --root . --skills-dir .agents/skills --rules-file AGENTS.md --hooks-path .codex/hooks
Claude Code 示例：
python3 .claude/skills/maintain-learnings/scripts/audit_learnings.py --root . --skills-dir .claude/skills --rules-file CLAUDE.md --hooks-path .claude/hooks
优先处理：
RULES.md 但仍在 ERRORS.md / LEARNINGS.md 中复发。根据审计报告读取对应源文件：
<skills-dir>/<skill>/SKILL.md<skills-dir>/<skill>/references/AGENTS.md / CLAUDE.md / 项目自定义规则文件.codex/hooks/ / .claude/hooks/ / 项目自定义 hookscripts/修复必须落到可执行机制之一：
SKILL.md 中加入明确步骤、硬性约束或验证 checklist。不要只把“下次注意”追加到 .learnings/。如果没有源头修改，不能清理对应错误记录。
至少验证两件事：
python3 -c 'from pathlib import Path; p=Path("<skills-dir>/<skill>/SKILL.md"); t=p.read_text(encoding="utf-8"); assert t.startswith("---\n") and "\n---" in t[4:]; assert "name:" in t and "description:" in t; print("skill metadata ok")'
如果项目同时使用 Codex 和 Claude Code，修改任何共享 skill 后都运行：
python3 .agents/skills/maintain-learnings/scripts/sync_platform_skills.py --root . --skill <skill>
如果另一侧缺失，先 dry-run：
python3 .agents/skills/maintain-learnings/scripts/sync_platform_skills.py --root . --from-platform agents --to-platform claude --skill <skill>
确认后应用：
python3 .agents/skills/maintain-learnings/scripts/sync_platform_skills.py --root . --from-platform agents --to-platform claude --skill <skill> --apply
保留平台专属内容：
agents/openai.yaml）只留在 .agents/。只处理已经验证修复的记录：
.learnings/archive/YYYY-MM-DD-maintenance.md 追加归档块，包含原记录摘要、修复路径、验证方式、处理结果。.learnings/LEARNINGS.md 或 .learnings/ERRORS.md 移除对应详细记录。.learnings/RULES.md 中的简短铁律。.learnings/。