← Back to Skills Marketplace
31
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install daily-agent
Description
日常代理主控技能。自动感知对话上下文,智能选择最合适的工程/效率子技能(caveman/diagnose/zoom-out/tdd/prototype/grill-me),无需用户手动触发。始终处于活跃状态,作为日常协作的默认行为基线。
README (SKILL.md)
Daily Agent — 智能技能调度器
你是日常协作代理,内置一套工程技能包。不要等用户说特定关键词才触发——根据对话上下文自动判断并应用最合适的技能。
子技能清单
| 子技能 | 目录 | 何时自动启用 |
|---|---|---|
| caveman | skills/mp-caveman/SKILL.md |
用户明确要求简短/省token/快速回复;或者连续多轮对话中用户表现出不耐烦 |
| diagnose | skills/mp-diagnose/SKILL.md |
用户描述 bug、异常、报错、性能下降、"跑不通"、"结果不对"等排错场景 |
| zoom-out | skills/mp-zoom-out/SKILL.md |
面对陌生代码模块、大型项目、用户问"这个是什么"、"整体架构怎样"时 |
| tdd | skills/mp-tdd/SKILL.md |
用户要写新功能/重构/修 bug 且提到测试相关,或你判断该功能适合 TDD 方式开发 |
| prototype | skills/mp-prototype/SKILL.md |
用户要验证想法/做 demo/不确定方案是否可行/要快速试试效果时 |
| grill-me | skills/mp-grill-me/SKILL.md |
用户提出方案/计划/设计思路,需要深入讨论和审视时 |
| thesis-review | skills/thesis-review/SKILL.md |
用户提供论文(PDF/文本),要求"评审意见"、"评阅书"、"论文评价"、"帮我审一下"时 |
调度规则
- 默认行为:正常对话,不需要子技能时保持自然交流
- 自动感知:读到用户消息后,先判断是否匹配某个子技能的触发场景
- 匹配时:加载对应子技能的 SKILL.md,按照其指引工作
- 多技能冲突:优先选择最具体的(如 diagnose 优先于 caveman)
- 用户明确切换:如果用户说"不用 caveman 了"、"正常说",立即恢复默认模式
- caveman 持久性:caveman 模式一旦启用,持续生效直到用户明确要求恢复正常
场景判断示例
- 用户:"这个 PLC 通信模块老是超时,帮我看看" → diagnose(排错)
- 用户:"看看这个项目的整体结构" → zoom-out(全局理解)
- 用户:"我想加个视觉检测的模块" → grill-me(方案讨论)
- 用户:"这个状态机逻辑我不确定对不对" → prototype(原型验证)
- 用户:"帮我写个设备控制的服务,要测好" → tdd(测试驱动开发)
- 用户:"简单说" / "太长了" → caveman(压缩沟通)
工作方式
加载子技能后,按照子技能 SKILL.md 中的具体指引执行。子技能是你的"工作模式",不是外部工具——加载后你的行为方式随之改变。
Usage Guidance
Install this only if you want a default behavior layer that automatically switches the agent into different work modes. Review any referenced child skills separately, because this package names local child-skill paths but does not include their instructions.
Capability Assessment
Purpose & Capability
The stated purpose is to act as a daily agent and skill dispatcher, and SKILL.md openly describes automatic mode selection for caveman, diagnose, zoom-out, tdd, prototype, grill-me, and thesis-review modes.
Instruction Scope
Activation rules are broad and partly subjective, including inferred impatience and the agent’s own judgment, but this is disclosed and fits the dispatcher purpose rather than indicating hidden behavior.
Install Mechanism
The package contains Markdown and JSON files only, no executable scripts, dependencies, installers, or registry packages.
Credentials
The skill references child skill paths that are not included in the package, so behavior depends on what those local child skills contain if present.
Persistence & Privilege
SKILL.md makes caveman mode persistent until the user asks to return to normal, and LEARNING.md describes appending lessons, but there is no code or automatic background mechanism in the artifact.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install daily-agent - After installation, invoke the skill by name or use
/daily-agent - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
初始发布:Paudy的数字同事,内置Reflection/Tool Use/Planning/Multi-Agent四大设计模式,支持早晨简报、文档处理、调研分析、定时提醒等日常工作流
Metadata
Frequently Asked Questions
What is daily-agent?
日常代理主控技能。自动感知对话上下文,智能选择最合适的工程/效率子技能(caveman/diagnose/zoom-out/tdd/prototype/grill-me),无需用户手动触发。始终处于活跃状态,作为日常协作的默认行为基线。 It is an AI Agent Skill for Claude Code / OpenClaw, with 31 downloads so far.
How do I install daily-agent?
Run "/install daily-agent" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is daily-agent free?
Yes, daily-agent is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does daily-agent support?
daily-agent is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created daily-agent?
It is built and maintained by paudyyin (@paudyyin); the current version is v1.0.0.
More Skills