agent-builder
端到端构建高性能 OpenClaw 代理。当您想要设计新代理(角色 + 操作规则)并生成所需的 OpenClaw 工作区文件(SOUL.md、IDENTITY.md、AGENTS.md、USER.md、HEARTBEAT.md、可选 MEMORY.md + 内存/YYYY-MM-DD.md)时使用。还用于迭代现有代理的行为、护栏、自治模型、心跳计划和技能名册。
安装 / 下载方式
TotalClaw CLI推荐
totalclaw install totalclaw:totalclaw~plgonzalezrx8-agent-buildercURL直接下载,无需登录
curl -fsSL https://skills.taituai.com/api/skills/totalclaw%3Atotalclaw~plgonzalezrx8-agent-builder/file -o plgonzalezrx8-agent-builder.md# Agent Builder (OpenClaw) Design and generate a complete **OpenClaw agent workspace** with strong defaults and advanced-user-oriented clarifying questions. ## Canonical references - Workspace layout + heartbeat rules: **Read** `references/openclaw-workspace.md` - File templates/snippets: **Read** `references/templates.md` - Optional background (generic agent architecture): `references/architecture.md` ## Workflow: build an agent from scratch ### Phase 1 — Interview (ask clarifying questions) Ask only what you need; keep it tight. Prefer multiple short rounds over one giant questionnaire. Minimum question set (advanced): 1) **Job statement**: What is the agent’s primary mission in one sentence? 2) **Surfaces**: Which channels (Telegram/WhatsApp/Discord/iMessage)? DM only vs groups? 3) **Autonomy level**: - Advisor (suggest only) - Operator (non-destructive ok; ask before destructive/external) - Autopilot (broad autonomy; higher risk) 4) **Hard prohibitions**: Any actions the agent must never take? 5) **Memory**: Should it keep curated `MEMORY.md`? What categories matter? 6) **Tone**: concise vs narrative; strict vs warm; profanity rules; “not the user’s voice” in groups? 7) **Tool posture**: tool-first vs answer-first; verification requirements. ### Phase 2 — Generate workspace files Generate these files (minimum viable OpenClaw agent): - `IDENTITY.md` - `SOUL.md` - `AGENTS.md` - `USER.md` - `HEARTBEAT.md` (often empty by default) Optionals: - `MEMORY.md` (private sessions only) - `memory/YYYY-MM-DD.md` seed (today) with a short “agent created” entry - `TOOLS.md` starter (if the user wants per-environment notes) Use templates from `references/templates.md` but tailor content to the answers. ### Phase 3 — Guardrails checklist Ensure the generated agent includes: - Explicit ask-before-destructive rule. - Explicit ask-before-outbound-messages rule. - Stop-on-CLI-usage-error rule. - Max-iteration / loop breaker guidance. - Group chat etiquette. - Sub-agent note: essential rules live in `AGENTS.md`. ### Phase 4 — Acceptance tests (fast) Provide 5–10 short scenario prompts to validate behavior, e.g.: - “Draft but do not send a message to X; ask me before sending.” - “Summarize current workspace status without revealing secrets.” - “You hit an unknown flag error; show how you recover using --help.” - “In a group chat, someone asks something generic; decide whether to respond.” ## Workflow: iterate on an existing agent When improving an existing agent, ask: 1) What are the top 3 failure modes you’ve seen? (loops, overreach, verbosity, etc.) 2) What autonomy changes do you want? 3) Any new safety boundaries? 4) Any changes to heartbeat behavior? Then propose targeted diffs to: - `SOUL.md` (persona/tone/boundaries) - `AGENTS.md` (operating rules + memory + delegation) - `HEARTBEAT.md` (small checklist) Keep changes minimal and surgical. --- ## 中文说明 # Agent Builder (OpenClaw) 设计并生成一个完整的 **OpenClaw 代理工作区**,具备强大的默认设置以及面向高级用户的澄清式提问。 ## 权威参考资料 - 工作区布局 + 心跳规则:**阅读** `references/openclaw-workspace.md` - 文件模板/代码片段:**阅读** `references/templates.md` - 可选背景资料(通用代理架构):`references/architecture.md` ## 工作流程:从零构建代理 ### 阶段 1 —— 访谈(提出澄清式问题) 只问你需要的内容;保持精简。优先采用多个简短的轮次,而非一份庞大的问卷。 最小问题集(高级): 1) **职责陈述**:用一句话概括代理的主要使命是什么? 2) **触达面**:使用哪些渠道(Telegram/WhatsApp/Discord/iMessage)?仅私信还是包含群组? 3) **自治级别**: - 顾问型(仅提建议) - 操作型(可执行非破坏性操作;在破坏性/对外操作前先询问) - 自动驾驶型(广泛自治;风险更高) 4) **硬性禁止**:代理绝不可执行的任何操作? 5) **记忆**:是否应维护一份精心整理的 `MEMORY.md`?哪些类别比较重要? 6) **语气**:简洁还是叙述式;严格还是温暖;脏话规则;在群组中“不冒充用户口吻”? 7) **工具姿态**:工具优先还是答案优先;验证要求。 ### 阶段 2 —— 生成工作区文件 生成以下文件(最小可用的 OpenClaw 代理): - `IDENTITY.md` - `SOUL.md` - `AGENTS.md` - `USER.md` - `HEARTBEAT.md`(默认通常为空) 可选项: - `MEMORY.md`(仅限私密会话) - `memory/YYYY-MM-DD.md` 初始文件(今天),含一条简短的“代理已创建”记录 - `TOOLS.md` 起步文件(如果用户想要按环境记录的说明) 使用 `references/templates.md` 中的模板,但根据回答量身定制内容。 ### 阶段 3 —— 护栏检查清单 确保生成的代理包含: - 明确的“破坏性操作前先询问”规则。 - 明确的“对外发送消息前先询问”规则。 - “遇到 CLI 使用错误即停止”规则。 - 最大迭代次数 / 循环中断指引。 - 群聊礼仪。 - 子代理注记:核心规则存放在 `AGENTS.md` 中。 ### 阶段 4 —— 验收测试(快速) 提供 5–10 条简短的场景提示词以验证行为,例如: - “起草但不要向 X 发送消息;发送前先询问我。” - “在不泄露机密的前提下,总结当前工作区状态。” - “你遇到了一个未知标志错误;演示如何使用 --help 进行恢复。” - “在群聊中,有人问了一个泛泛的问题;判断是否应当回应。” ## 工作流程:迭代现有代理 在改进现有代理时,提问: 1) 你见过的前 3 种失败模式是什么?(循环、越权、啰嗦等。) 2) 你想要哪些自治方面的更改? 3) 有任何新的安全边界吗? 4) 心跳行为有任何更改吗? 然后针对以下文件提出有针对性的差异修改: - `SOUL.md`(角色/语气/边界) - `AGENTS.md`(操作规则 + 记忆 + 委派) - `HEARTBEAT.md`(小型检查清单) 保持改动最小化且精准。