oh-my-openclaw

Curated ClawHub skills for OpenClaw, tested via GitHub Actions + Telegram.

Install

As OpenClaw plugin (recommended):

openclaw plugins install @m4d3bug/oh-my-openclaw

Or one-line script:

curl -fsSL https://raw.githubusercontent.com/m4d3bug/oh-my-openclaw/master/install.sh | bash

Included skills (44)

Security

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | skill-vetter | 1.0.0 | 2,000 | ✅ Benign | Security-first vetting — review any skill before install | | agent-sentinel | 0.1.1 | 3 | ✅ Benign | Safety circuit breaker — budget constraints, rate limits | | arc-skill-scanner | 1.4.0 | 0 | ✅ Benign | Auto-scan skills for vulnerabilities before install |

Memory & Persistence

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | self-improving-agent | 3.0.4 | 3,700 | ✅ Benign | Log errors, corrections, and learnings across sessions | | agent-memory | 1.0.0 | 262 | ✅ Benign | Persistent SQLite memory — store, recall, track entities | | ontology | 1.0.4 | 388 | ✅ Benign | Typed knowledge graph for structured agent memory | | arc-wake-state | 1.0.0 | 0 | ✅ Benign | Auto-recover agent state after crash/restart | | agent-wal | 1.0.1 | 5 | ✅ Benign | Write-Ahead Log — atomic state change guarantees |

Agent Behavior

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | proactivity | 1.0.1 | 18 | ✅ Benign | Anticipate needs, keep work moving, improve over time | | self-reflection | 1.1.1 | 106 | ✅ Benign | Continuous self-improvement through structured reflection | | agent-team-orchestration | 1.0.0 | 125 | ✅ Benign | Multi-agent team coordination — roles, handoffs, reviews | | reflection | 1.1.0 | 23 | ✅ Benign | Self-critique before responding — fewer revision rounds |

Session Management

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | smart-context | 1.0.0 | 0 | ✅ Benign | Token-efficient context pruning, response sizing, delegation | | summarize | 1.0.0 | 3,800 | ✅ Benign | Summarize conversations, code changes, and sessions | | daily-digest | 1.0.0 | 68 | ✅ Benign | Generate daily journal from memory files |

Code Intelligence

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | github | 1.0.0 | 2,863 | ✅ Benign | Git workflow — PRs, issues, CI runs via gh CLI | | test-runner | 1.0.0 | 71 | ✅ Benign | Run tests — Jest, pytest, Playwright, XCTest | | planning-with-files | 2.22.0 | 70 | ⚠️ Suspicious | Task planning — task_plan.md / findings.md / progress.md | | docsync | 1.0.1 | 0 | ✅ Benign | Auto-generate docs from code, detect doc drift | | simplify-and-harden | 1.0.1 | 6 | ✅ Benign | Post-completion self-review — simplify, harden, micro-doc | | env-setup | 1.0.0 | 2 | ✅ Benign | Scan env vars, generate .env.example, validate .gitignore |

DevOps

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | cron-mastery | 1.0.3 | 103 | ✅ Benign | Cron scheduling and system maintenance | | docker-essentials | 1.0.0 | 209 | ✅ Benign | Docker commands, container management, debugging | | cicd-pipeline | 1.0.0 | 24 | ✅ Benign | Manage GitHub Actions CI/CD pipelines directly |

Context & Search

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | multi-search-engine | 2.0.1 | 796 | ✅ Benign | 17 search engines (8 CN + 9 Global), no API key needed | | agent-reach | 1.1.0 | 246 | ⚠️ Suspicious | Read 14+ platforms (Twitter, Reddit, YouTube, Bilibili, etc.) | | agent-browser | 0.2.0 | 2,588 | ✅ Benign | Headless browser — navigate, click, type, screenshot, record | | cli-anything | 2026-03-17 | — | GitHub | Natural language to CLI commands for any tool | | clawddocs | 1.2.2 | 413 | ✅ Benign | OpenClaw documentation assistant with decision tree routing | | humanizer | 1.0.0 | 760 | ✅ Benign | Remove AI writing patterns — make text sound human |

Creative & Planning

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | brainstorming | 0.1.0 | 88 | ✅ Benign | Explore intent, requirements and design before implementation | | writing-plans | 0.1.0 | 43 | ✅ Benign | Generate multi-step implementation plans from specs | | copywriting | 0.1.0 | 124 | ✅ Benign | Marketing copy — AIDA/PAS/FAB, headlines, CTAs, ad copy | | content-strategy | 0.1.0 | 45 | ✅ Benign | Content marketing strategy, calendar, distribution, metrics |

Autonomous Execution

| Skill | Version | Installs | OpenClaw Scan | What it does | |-------|---------|----------|---------------|-------------| | better-ralph | 1.0.0 | — | ✅ Benign | PRD-driven autonomous coding iteration | | agent-weave | 1.0.0 | — | ✅ Benign | Master-Worker parallel task execution cluster | | forge | 1.0.0 | — | ✅ Benign | Autonomous QA swarm — E2E testing, self-healing fix loops | | agent-autonomy-kit | 1.0.0 | 121 | ✅ Benign | Stop waiting for prompts — keep working autonomously | | codebase-documenter | 0.1.0 | 5 | ✅ Benign | Generate README, architecture docs, API docs from codebase | | debug-pro | 1.0.0 | 114 | ✅ Benign | 7-step debugging protocol with multi-language support |

skill-vetter is installed first to audit all subsequent installs.

Local setup

1. Install OpenClaw + plugin

npm install -g openclaw
openclaw plugins install @m4d3bug/oh-my-openclaw

2. Configure model + Telegram

Pick one model backend:

OpenAI-compatible (e.g. Aliyun DashScope):

openclaw config set models.providers.custom-openai.baseUrl "https://coding.dashscope.aliyuncs.com/v1"
openclaw config set models.providers.custom-openai.apiKey "your-key"
openclaw config set models.providers.custom-openai.api "openai-completions"
openclaw config set agents.defaults.model.primary "custom-openai/MiniMax-M2.5"

Anthropic-compatible (e.g. SiliconFlow):

openclaw config set models.providers.custom-anthropic.baseUrl "https://api.siliconflow.cn/v1"
openclaw config set models.providers.custom-anthropic.apiKey "your-key"
openclaw config set models.providers.custom-anthropic.api "anthropic-messages"
openclaw config set agents.defaults.model.primary "custom-anthropic/Pro/MiniMaxAI/MiniMax-M2.5"

Telegram (get token from @BotFather):

openclaw config set channels.telegram.enabled true
openclaw config set channels.telegram.dmPolicy open
openclaw config set channels.telegram.allowFrom '["*"]'   # skip pairing, anyone can chat
# or restrict to your Telegram ID only:
# openclaw config set channels.telegram.allowFrom '["YOUR_TELEGRAM_ID"]'

Both local and GitHub Actions use dmPolicy: open + allowFrom: ["*"] to skip device pairing. Change ["*"] to your Telegram user ID for production use.

3. Enable full tools + start

openclaw config set tools.profile full
TELEGRAM_BOT_TOKEN="your-bot-token" openclaw gateway run --verbose

Chat with your bot on Telegram. Skills are loaded automatically from the plugin.

GitHub Actions + Telegram

1. Fork this repo

2. Set Secrets (Settings > Secrets > Actions):

   | Secret | How to get it | |--------|--------------| | TELEGRAM_BOT_TOKEN | @BotFather > /newbot |

   Pick one model backend (or both — OpenAI takes priority):

   OpenAI-compatible (e.g. Aliyun DashScope): | Secret | Example | |--------|---------| | OPENAI_API_KEY | your key | | OPENAI_BASE_URL | https://coding.dashscope.aliyuncs.com/v1 | | OPENAI_MODEL | MiniMax-M2.5 |

   Anthropic-compatible (e.g. SiliconFlow): | Secret | Example | |--------|---------| | ANTHROPIC_API_KEY | your key | | ANTHROPIC_BASE_URL | https://api.siliconflow.cn/v1 | | ANTHROPIC_MODEL | Pro/MiniMaxAI/MiniMax-M2.5 |

3. Push any commit or run workflow manually (Actions > Test Drive > Run workflow)

4. Chat with your bot on Telegram

New runs automatically cancel previous ones.

Default configuration

tools.profile = full

Managing skills

clawhub install <slug>       # Install a skill
clawhub search "keyword"     # Search ClawHub
clawhub list                 # List installed
clawhub update --all         # Update all

How it works

Push / Manual trigger
  └─ GitHub Actions
       ├─ docker pull ghcr.io/openclaw/openclaw
       ├─ clawhub install (from skills.txt)
       ├─ git clone (from github-skills.txt)
       ├─ docker run ... bash /omocw/docker/entrypoint.sh
       │    ├─ gen-config.py         → ~/.openclaw/openclaw.json
       │    ├─ copy skills/*/SKILL.md → ~/.openclaw/skills/
       │    ├─ openclaw config set tools.profile full
       │    └─ openclaw gateway run (Telegram polling)
       └─ Keep alive 15 min, then cleanup

License

MIT