雇用
- 作者仓库星标 39
- 作者仓库 awesome-omni-skill
- 领域
- 人工智能
- 作者声明的智能体
- 未找到明确声明;不据此推断已兼容或已测试
- 静态检查
- 88 / 100 · 启发式扫描,不代表运行安全
- 运行环境证据
- 未执行;当前没有隔离运行测试证据
- 作者 / 版本 / 许可证
- @diegosouzapw · 未声明许可证
- 上下文用量评级
- 低消耗
- 接入复杂程度
- 即装即用
- 是否需要外部接口密钥
- 不需要
- 兼容的系统
- 未声明(默认跨平台)
- 运行环境要求
- 未检出明确要求
- 检测到的文件与系统行为
-
- 只读
- 允许写入 / 修改
- 网络行为
- 仅限本地
- 安装命令数
- 无(仅作为资料)
需要注意: 未声明工具权限范围;实际权限由运行环境决定。
第 1 步:发现可用型号
运行`openclaw models list`或检查网关配置以查看配置的内容。
第二步:按等级分类
将发现的模型映射到功能层:层·模型(示例)·最适合推理·claude-opus-*、gpt-5*、gpt-4o、deepseek-r1·策略、咨询、复杂分析、架构平衡·claude-sonnet-*、gpt-4-turbo、gpt-4o-mini·研究、写作、快速一般任务·claude-haiku-*、gpt-3.5*、local/ollama·大容量、简单的任务、草稿
第 3 步:将角色与层级匹配
基于访谈: 大量推理/咨询/策略 → 推理层 研究/写作/创意 → 平衡层 以代码为中心 → 代码层(如果不可用则平衡) 高容量/监控 → 快速层
第四步:选择并确认
选择最适合该角色的可用模型。在摘要卡中,添加:如果存在多个好的选项或者您不确定,请询问:“对于[角色类型]角色,我建议[模型](能力和成本的良好平衡)。或者如果您想要[更深入的推理/更快的响应/更低的成本],则选择[替代方案]。偏好?”
# hire
Set up a new AI team member through a guided conversation. Not a config generator - a hiring process.
## When to Use
User says something like:
- "I want to hire a new agent"
- "I need help with X" (where X implies a new agent role)
- "Let's add someone to the team"
- `/hire`
## The Interview
### 3 core questions, asked one at a time:
**Q1: "What do you need help with?"**
Let them describe the problem, not a job title. "I'm drowning in code reviews" beats "I need a code reviewer."
- Listen for: scope, implied autonomy level, implied tools needed
**Q2: "What's their personality? Formal, casual, blunt, cautious, creative?"**
Or frame it as: "If this were a human colleague, what would they be like?"
- Listen for: communication style, vibe, how they interact
**Q3: "What should they never do?"**
The red lines. This is where trust gets defined.
- Listen for: boundaries, safety constraints, access limits
### Q4: Dynamic (optional)
After Q1-Q3, assess whether anything is ambiguous or needs clarification. If so, ask ONE follow-up question tailored to what's unclear. Examples:
- "You mentioned monitoring - should they alert you immediately or batch updates?"
- "They'll need access to your codebase - any repos that are off-limits?"
- "You said 'casual' - are we talking friendly-professional or meme-level casual?"
If Q1-Q3 were clear enough, skip Q4 entirely.
## Summary Card
After the interview, present a summary:
```
🎯 Role: [one-line description]
🧠 Name: [suggested name from naming taxonomy]
🤖 Model: [selected model] ([tier])
⚡ Personality: [2-3 word vibe]
🔧 Tools: [inferred from conversation]
🚫 Boundaries: [key red lines]
🤝 Autonomy: [inferred level: high/medium/low]
```
Then ask: **"Want to tweak anything, or are we good?"**
## Model Selection
Before finalizing, select an appropriate model for the agent.
### Step 1: Discover available models
Run `openclaw models list` or check the gateway config to see what's configured.
### Step 2: Categorize by tier
Map discovered models to capability tiers:
| Tier | Models (examples) | Best for |
|------|-------------------|----------|
| **reasoning** | claude-opus-*, gpt-5*, gpt-4o, deepseek-r1 | Strategy, advisory, complex analysis, architecture |
| **balanced** | claude-sonnet-*, gpt-4-turbo, gpt-4o-mini | Research, writing, general tasks |
| **fast** | claude-haiku-*, gpt-3.5*, local/ollama | High volume, simple tasks, drafts |
| **code** | codex-*, claude-sonnet-*, deepseek-coder | Coding, refactoring, tests |
Use pattern matching on model names - don't hardcode specific versions.
### Step 3: Match role to tier
Based on the interview:
- Heavy reasoning/advisory/strategy → reasoning tier
- Research/writing/creative → balanced tier
- Code-focused → code tier (or balanced if not available)
- High-volume/monitoring → fast tier
### Step 4: Select and confirm
Pick the best available model for the role. In the summary card, add:
```
🤖 Model: [selected model] ([tier] - [brief reason])
```
If multiple good options exist or you're unsure, ask:
"For a [role type] role, I'd suggest [model] (good balance of capability and cost). Or [alternative] if you want [deeper reasoning / faster responses / lower cost]. Preference?"
### Notes
- Don't assume any specific provider - work with what's available
- Cheaper is better when capability is sufficient
- The user's default model isn't always right for every agent
- If only one model is available, use it and note it in the summary
## Optional Extras
After the summary is confirmed, offer:
1. **"Want to set up periodic performance reviews?"**
- If yes: ask preferred frequency (weekly, biweekly, monthly)
- Create a cron job that triggers a review conversation
- Review covers: what went well, what's not working, scope/permission adjustments
- At the end of each review, ask: "Want to keep this schedule, change frequency, or stop reviews?"
2. **Onboarding assignment** (if relevant to the role)
- Suggest a small first task to test the new agent
- Something real but low-stakes, so the user can see them in action
## What to Generate
Create an agent directory at `agents/<name>/` with:
### Always unique (generated fresh):
- **AGENTS.md** - Role definition, responsibilities, operational rules, what they do freely vs ask first
- **IDENTITY.md** - Name, emoji, creature type, vibe, core principles
### Start from template, customize based on interview:
- **SOUL.md** - Base from workspace SOUL.md template, customize vibe/boundaries sections
- **TOOLS.md** - Populated with inferred tools and access notes
- **HEARTBEAT.md** - Empty or with initial periodic tasks if relevant to role
### Symlink to shared (default, opinionated):
- **USER.md** → `../../USER.md` (they need to know who they work for)
- **MEMORY.md** → `../../MEMORY.md` (shared team context)
Mention to the user: "I've linked USER.md and MEMORY.md so they know who you are and share team context. You can change this later if you want them more isolated."
## Naming
Use craft/role-based names. Check TOOLS.md for the full naming taxonomy:
- Research: Scout, Observer, Surveyor
- Writing: Scribe, Editor, Chronicler
- Code: Smith, Artisan, Engineer
- Analysis: Analyst, Assessor, Arbiter
- Creative: Muse, Artisan
- Oversight: Auditor, Reviewer, Warden
Check existing agents to avoid name conflicts. Suggest a name that fits the role, but let the user override.
## Team Awareness
Before generating, check `agents/` for existing team members. Note:
- Potential overlaps with existing roles
- Gaps this new hire fills
- How they'll interact with existing agents
Mention any relevant observations: "You already have Scout for research - this new role would focus specifically on..."
## After Setup
1. Tell the user what was created and where
2. Add the agent to OpenClaw config with the selected model:
```json
{
"id": "<name>",
"workspace": "/path/to/clawd/agents/<name>",
"model": "<selected-model>"
}
```
Or guide them to run `openclaw agents add`
3. If monthly reviews were requested, confirm the cron schedule
4. Update any team roster if one exists
## Important
- This is a CONVERSATION, not a form. Be natural.
- Infer as much as possible from context. Don't ask what you can figure out.
- The user might not know what they want exactly. Help them figure it out.
- Keep the whole process under 5 minutes for the simple case.
章节与文件索引
按说明文件中的顺序排列。
何时使用
3个核心问题,一次问一个:
Q4:动态(可选)
总结卡
选型
文件/命令 -> /hire · openclaw models list · agents/<name>/ · ../../USER.md · ../../MEMORY.md · agents/ · openclaw agents add
内容校验值 -> aa57603b238c
当前原文的自动摘译,尚未人工审校。
何时使用
用户会说这样的话:“我想雇用一名新代理人”“我需要 X 方面的帮助”(其中 X 意味着新的代理人角色)“让我们向团队添加某人”
/hire3个核心问题,一次问一个:
Q1:“您需要什么帮助?”让他们描述问题,而不是职位名称。 “我淹没在代码审查中”胜过“我需要一名代码审查员”。倾听:范围、隐含的自主水平、隐含的所需工具 Q2:“他们的个性是什么?正式、随意、直率、谨慎、有创造力?”或者将其描述为:“如果这是一位人类同事,他们会是什么样子?”
Q4:动态(可选)
在 Q1-Q3 之后,评估是否有任何不明确或需要澄清的事情。如果是这样,请针对不清楚的内容提出一个后续问题。示例:“您提到监控 - 他们应该立即提醒您还是批量更新?” “他们需要访问你的代码库——任何禁止访问的存储库?” “你说‘休闲’——我们是在谈论友好的专业还是模因级的休闲?”如果 Q1-Q3 足够清楚,则完全跳过 Q4。
总结卡
面试结束后,提出一个总结:然后问:“想要调整一些东西,还是我们做得很好?”
选型
在最终确定之前,请为代理选择合适的型号。