A curated collection of OpenClaw agent skills — performance optimization, self-improvement, Chinese model support, and community tools.
精心整理的 OpenClaw 智能体 Skills 集合 — 性能优化、自我进化、国产模型支持、社区工具。
| Skill | Name | 名称 | Description | 简介 |
|---|---|---|---|---|
openclaw-speed-optimizer |
Speed Optimizer | 响应速度优化 | Checks Context usage, analyzes Bootstrap files, validates config status | 检查 Context 使用率、分析 Bootstrap 文件、验证配置状态 |
openclaw-compute-guardian |
Compute Guardian | 计算资源守护者 | Monitors process resources, analyzes Cron token consumption | 监控进程资源、分析 Cron Token 消耗 |
openclaw-self-optimizing |
Self-Optimizing | 自我进化引擎 | Meta-learning, automatic reflection, behavioral adaptation, skill synthesis | 元学习、自动反思、行为适应、技能合成 |
chinese-model-compat |
Chinese Model Compat | 国产模型兼容性 | Resolves 400/401 errors for Kimi, GLM, Bailian models | 解决 Kimi、GLM、百炼模型的 400/401 错误 |
unified-memory |
Unified Memory | 跨渠道统一记忆 | Cross-channel memory sharing for Telegram, Discord, WebChat | 跨渠道记忆共享,Telegram/Discord/WebChat |
session-robust |
Session Robust | Session 健壮性 | Manual compaction, backup, restore for session management | 手动压缩、备份、恢复 session |
skill-registry |
Skill Registry | Skills 注册表 | Recommendation engine, ClawHub CLI, skills.sh Top 50 | 推荐引擎、ClawHub CLI、Top 50 索引 |
chinese-community-kit |
Chinese Community Kit | 中文社区工具包 | Publishing toolkit for SegmentFault, Zhihu, CSDN, WeChat | SegmentFault、知乎、CSDN、微信公众号发布工具 |
| Requirement | Version | 说明 |
|---|---|---|
| Node.js | 22+ | OpenClaw runtime |
| OpenClaw | Latest | Agent framework |
| Python | 3.8+ | Scripts runtime |
| OS | Cross-platform | Windows/macOS/Linux |
# Clone the repository
git clone https://github.com/YOUR_USERNAME/openclaw-skills.git
cd openclaw-skills
# Copy all skills to your workspace
cp -r skills/* ~/.openclaw/workspace-main/skills/
# Or copy specific skills
cp -r skills/openclaw-speed-optimizer ~/.openclaw/workspace-main/skills/# Navigate to your workspace
cd ~/.openclaw/workspace-main/skills/
# Clone individual skill
git clone https://github.com/YOUR_USERNAME/openclaw-skills.git skills/openclaw-speed-optimizer
# Then move the skill folder up:
mv openclaw-skills/* .
rm -rf openclaw-skills# Install via ClawHub (if skill is published)
clawhub install openclaw-speed-optimizer
clawhub install chinese-model-compatopenclaw-skills/
├── skills/
│ ├── openclaw-speed-optimizer/ # Speed & latency optimizer
│ ├── openclaw-compute-guardian/ # Resource monitor
│ ├── openclaw-self-optimizing/ # Self-improvement engine
│ ├── chinese-model-compat/ # Chinese model compatibility
│ ├── unified-memory/ # Cross-channel memory
│ ├── session-robust/ # Session management
│ ├── skill-registry/ # Skill discovery
│ └── chinese-community-kit/ # Community publishing
├── GITHUB_UPLOAD_README.md # This file
└── LICENSE # MIT License
# Check speed optimization status
python skills/openclaw-speed-optimizer/scripts/speed_check.py --status
# Monitor resource usage
python skills/openclaw-compute-guardian/scripts/resource_monitor.py --watch
# Run self-optimization
python skills/openclaw-self-optimizing/scripts/meta_learning.py# Test Chinese model compatibility
python skills/chinese-model-compat/scripts/model_tester.py --all
# Apply safe configuration
python skills/chinese-model-compat/scripts/config_generator.py --apply# Start unified memory server
python skills/unified-memory/scripts/memory_server.py
# Backup current session
python skills/session-robust/scripts/session_backup.pyContributions are welcome! Please read the guidelines below.
欢迎贡献!请阅读以下指南。
- Fork the repository
- Create a branch for your feature:
git checkout -b skill/my-new-skill - Follow the structure:
skills/my-new-skill/ ├── SKILL.md # Required: skill definition ├── README.md # Required: bilingual (EN/CN) documentation ├── scripts/ # Optional: executable scripts └── docs/ # Optional: additional docs - SKILL.md format — Must include:
name,description,version,author- Use when triggers (EN + CN)
- Commands / usage examples
- File structure
- README.md — Bilingual format with EN/CN for every section
- Test your scripts before submitting
- Submit a PR with a clear description
- ✅ All scripts must be cross-platform compatible
- ✅ No hardcoded absolute paths (use
~/.openclaw/or env vars) - ✅ No API keys or personal information in code
- ✅ Bilingual documentation (English + Chinese)
- ✅ Clear error messages and help text
MIT License — see LICENSE file for details.
MIT 许可证 — 详见 LICENSE 文件。
Maintained by the OpenClaw community. 由 OpenClaw 社区维护。