21.3K ·GitHub
What is this?
A Chinese-community edition of the agency-agents library: 277 plug-and-play AI expert personas covering 20 departments — from CEO and CTO to engineering, design, security and finance — that install into AI coding tools like Claude Code, Cursor and Copilot.
A Chinese-community edition of the agency-agents library: 277 plug-and-play AI expert personas covering 20 departments — from CEO and CTO to engineering, design, security and finance — that install into AI coding tools like Claude Code, Cursor and Copilot.So is this a standalone app, or files I add to the AI tool I already use?Not an app — it's a library of agent definition files: each expert is a prompt with its own identity, rules, workflow and deliverables that you install into your AI coding tool and activate with natural language.
Not an app — it's a library of agent definition files: each expert is a prompt with its own identity, rules, workflow and deliverables that you install into your AI coding tool and activate with natural language.How do I install the agents into my tool?One command installs into any of the 20 supported tools: Claude Code and GitHub Copilot install directly, while other tools first need ./scripts/convert.sh to convert the format. You can also copy files manually or just browse and adapt the prompts.
One command installs into any of the 20 supported tools: Claude Code and GitHub Copilot install directly, while other tools first need ./scripts/convert.sh to convert the format. You can also copy files manually or just browse and adapt the prompts.Do I need to install anything before I can use it?No special prerequisites — the agents run inside whichever of the 20 supported AI coding tools you already have. If you'd rather avoid the terminal, there's a native desktop client that needs no Node.js at all.
No special prerequisites — the agents run inside whichever of the 20 supported AI coding tools you already have. If you'd rather avoid the terminal, there's a native desktop client that needs no Node.js at all.Is this only useful if I work with Chinese platforms?Not only: 213 agents are translations of the upstream English library, while 64 are China-market originals covering Xiaohongshu, Douyin, WeChat, Bilibili, Feishu and DingTalk operations, plus verticals like cross-border e-commerce and medical compliance.
Not only: 213 agents are translations of the upstream English library, while 64 are China-market originals covering Xiaohongshu, Douyin, WeChat, Bilibili, Feishu and DingTalk operations, plus verticals like cross-border e-commerce and medical compliance.Can I look through the experts without installing anything?Yes — ao.aiolaola.com/experts is an online library where you can search, filter by department and copy any expert's full prompt right in the browser; a native desktop client for macOS, Windows and Linux is also available.
Yes — ao.aiolaola.com/experts is an online library where you can search, filter by department and copy any expert's full prompt right in the browser; a native desktop client for macOS, Windows and Linux is also available.How is this different from a plain prompt that says 'you are an expert'?A generic prompt just assigns a role; each agent here defines how the expert thinks, works and what they deliver. The security engineer audits code against OWASP Top 10, and the Xiaohongshu specialist outputs a full content strategy plus an influencer collaboration plan.
A generic prompt just assigns a role; each agent here defines how the expert thinks, works and what they deliver. The security engineer audits code against OWASP Top 10, and the Xiaohongshu specialist outputs a full content strategy plus an influencer collaboration plan.What is the Agency Orchestrator the README keeps mentioning?The companion no-code orchestrator: describe the job in one sentence and multiple experts team up automatically along a DAG, running in parallel with checkpoint resume. It supports 10 LLMs (7 of them without an API key) and ships ready-made templates.
The companion no-code orchestrator: describe the job in one sentence and multiple experts team up automatically along a DAG, running in parallel with checkpoint resume. It supports 10 LLMs (7 of them without an API key) and ships ready-made templates.How does the OpenClaw integration work?OpenClaw is described as the community's most popular integration: each agent is split into three files — SOUL.md for the persona, AGENTS.md for capabilities, IDENTITY.md for the intro — which natively supports multi-agent orchestration. Restart the OpenClaw gateway after installing.
OpenClaw is described as the community's most popular integration: each agent is split into three files — SOUL.md for the persona, AGENTS.md for capabilities, IDENTITY.md for the intro — which natively supports multi-agent orchestration. Restart the OpenClaw gateway after installing.Is there training on how to use all 277 experts as a team?Yes — the author's official free 35-lesson course walks through solo experts, auto-team assembly and one-person-company workflows, with the desktop client taught zero-code. The same platform hosts 13 courses (648 lessons) in total, all free.
Yes — the author's official free 35-lesson course walks through solo experts, auto-team assembly and one-person-company workflows, with the desktop client taught zero-code. The same platform hosts 13 courses (648 lessons) in total, all free.What are similar projects?superpowers-zh, system-prompts-and-models-of-ai-tools, nuwa-skill, token-monitor, easy-vibe, zhangxuefeng-skill, 9router, agents
superpowers-zh, system-prompts-and-models-of-ai-tools, nuwa-skill, token-monitor, easy-vibe, zhangxuefeng-skill, 9router, agentsWhich topics is it in?cursor-rules, system-prompt, system-prompts, workbuddy, feishu, prompt-library, xiaohongshu, prompts, copilot, no-code, hermes-agent, wechat
cursor-rules, system-prompt, system-prompts, workbuddy, feishu, prompt-library, xiaohongshu, prompts, copilot, no-code, hermes-agent, wechat