← 返回 Skills 市场
OMNI — All-In-One Master Skill
作者
Eternal0404
· GitHub ↗
· v1.0.0
· MIT-0
104
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install omni-master
功能描述
The all-in-one master skill — unified interface for every capability domain. Use when the user asks for anything that spans multiple domains or when no singl...
安全使用建议
This skill is intentionally broad and functions as a router across many domains; that explains much of its wide I/O and network behavior, but also raises the risk that it will read workspace files, write persistent logs, run shell commands, and request external service credentials at runtime. Before installing: 1) Inspect scripts/router.sh (and any included scripts) to confirm there are no unexpected execs or external endpoints. 2) Decide a policy for credentials — only provide oauth/API keys on a per-task basis, and prefer scoped, revocable tokens. 3) Ensure the platform will prompt you before destructive actions (deletes, installs, posts, financial ops). 4) Verify memory/ logging paths (memory/*.json) don't capture secrets; if needed, configure or sandbox the skill's storage location. 5) If you need tighter guarantees, prefer more focused domain-specific skills that explicitly declare required credentials and narrower permissions. If you want, I can inspect the contents of scripts/router.sh (if you paste it) and point out any risky commands or endpoints.
功能分析
Type: OpenClaw Skill
Name: omni-master
Version: 1.0.0
The 'omni-master' skill bundle is a comprehensive framework designed to provide an AI agent with structured domain knowledge and cognitive protocols. It features a master router (SKILL.md and router.sh) that directs the agent to 35+ specialized reference files covering topics from coding and system administration to ethics and reasoning. The bundle includes a 'Brain' protocol (brain.md) that implements error logging, loop detection, and anti-hallucination measures, while the 'ethics-safety.md' file explicitly instructs the agent to avoid destructive actions and protect sensitive data. No evidence of malicious intent, data exfiltration, or unauthorized execution was found.
能力评估
Purpose & Capability
The name/description (an all-in-one router) aligns with the included references covering many domains. That breadth reasonably explains why the SKILL.md references many tools and workflows. However, many reference files show example integrations that require external credentials and platform-specific CLIs (Google Calendar/Gmail via 'gog', Slack/Discord skills, IMAP/SMTP tools, cloud providers, etc.) while the skill declares no required environment variables or primary credential. It's plausible a router would defer credential handling to other skills, but the lack of any declared expected credentials is worth noting.
Instruction Scope
The SKILL.md and references explicitly direct the agent to read and write local memory files (memory/mistakes.json, decisions.json, daily logs), to 'read the code in the workspace' for debugging, and to run CLI/network verification commands (curl, sqlite3, pip/npm commands) as part of anti-hallucination and precision protocols. Those instructions give the skill the ability to access arbitrary workspace files, execute shell commands, and perform network calls. While these actions can be necessary for a multi-domain assistant, they also expand the surface for accidental data exposure or destructive actions unless additional guards are enforced (explicit confirmations, masking of secrets, avoid destructive commands by default).
Install Mechanism
This is instruction-first with no install spec, so nothing is downloaded or written by an installer. That keeps install-time risk low. There is a scripts/router.sh file included (present in the manifest), which should be reviewed because instruction-only skills that include runnable scripts can still cause runtime effects if executed.
Credentials
The skill declares no required env vars or credentials, which reduces upfront privilege requests. However, the references and examples show many operations that typically require API keys, OAuth tokens, or mail/calendar credentials. Expect the skill to ask for or attempt to use credentials at runtime via other platform skills. The manifest not declaring expected credential names is not necessarily malicious but reduces clarity and increases the chance the agent will request sensitive tokens ad hoc — the user should be prepared to deny/provide credentials carefully.
Persistence & Privilege
always:false (no forced global inclusion) and disable-model-invocation:false (normal autonomous invocation) are reasonable. The skill's 'Perfect Memory' and 'Brain' protocols instruct writing persistent logs to memory/*.json, so it will create/append local files by design. This is expected for a router skill, but users should review what gets logged (the references claim to avoid logging secrets). There is no indication the skill modifies other skills' configs or requests system-wide privileges.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install omni-master - 安装完成后，直接呼叫该 Skill 的名称或使用
/omni-master触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release: 38 files covering every domain.
## What's Included
### Adaptive Brain (Always-On)
- Error memory with root cause analysis & prevention rules
- Loop detection & breaking protocols
- Anti-hallucination with confidence calibration
- Precision code guidelines (Python/JS/Bash)
- 5-step systematic debug protocol
- Self-improving agent patterns (detection triggers, promotion workflow, dedup)
### Advanced Prompting Engine
- Transform vague/broken/contradictory prompts into precise execution
- Prompt repair techniques, CLEAR framework
- Auto-complete missing context, disambiguate homonyms
### 37 Domain References
Brain, Advanced Prompting, Data Literacy, Reasoning, Math Engine,
Research, Quality Assurance, Performance, Ethics & Safety,
Health & Fitness, Career, Learning, Network & Cloud, Creative,
Multi-Agent, Realtime, Automation, Coding, Data & Databases,
Web Scraping, Finance, Security, Media Generation, Music/Audio,
Email & Messaging, Calendar, Documents, Translation, Writing,
Project Management, Deployment/DevOps, IoT/Smart Home,
Communications, Knowledge Management, System Admin, API Design
### Infrastructure
- Router v3: 37-classify-rule domain classifier with fallback
- Memory system: mistakes.json, patterns.json, decisions.json, feature-requests.json
- 53 installed skills catalog
- All code blocks validated (Python/Bash/JS/JSON)
元数据
常见问题
OMNI — All-In-One Master Skill 是什么？
The all-in-one master skill — unified interface for every capability domain. Use when the user asks for anything that spans multiple domains or when no singl... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 104 次。
如何安装 OMNI — All-In-One Master Skill？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install omni-master」即可一键安装，无需额外配置。
OMNI — All-In-One Master Skill 是免费的吗？
是的，OMNI — All-In-One Master Skill 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
OMNI — All-In-One Master Skill 支持哪些平台？
OMNI — All-In-One Master Skill 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 OMNI — All-In-One Master Skill？
由 Eternal0404（@eternal0404）开发并维护，当前版本 v1.0.0。
推荐 Skills