What changed
The jnMetaCode/agency-agents-zh repository hosts a collection of 277 AI expert roles designed for plug-and-play integration. These agents are compatible with over 20 tools, including popular options like Claude Code, Cursor, and Copilot. The project emphasizes broad applicability, covering 20 departments such as engineering, design, marketing, and finance. A notable feature is the inclusion of 64 original AI agents specifically developed for the Chinese market, addressing platforms like Xiaohongshu, Douyin, WeChat, Feishu, DingTalk, and applications involving Qt-based upper machines and mechanical design.
Complementing the agent definitions is the agency-orchestrator, a component that facilitates automatic collaboration among multiple experts. This orchestration is based on a Directed Acyclic Graph (DAG) structure, allowing for complex workflows to be initiated with simple, single-sentence commands. The repository is primarily written in Shell and has a MIT license. The latest release, v1.2.6, was made on June 16, 2026.
Why it matters for builders
For developers and AI engineers, this repository offers a significant head start in building multi-agent systems. The extensive library of pre-configured AI experts eliminates the need to define basic agent functionalities from scratch. The inclusion of agents tailored to specific Chinese market applications provides a competitive edge for projects targeting that region. Furthermore, the integration with agency-orchestrator simplifies the complex task of managing inter-agent communication and workflow execution, enabling builders to focus on higher-level logic and innovation.
Practical impact
Builders can leverage these AI expert roles to rapidly assemble sophisticated AI agents for diverse tasks. For instance, a marketing team could deploy a suite of agents for social media campaign management, with specific agents handling content creation for Douyin or Xiaohongshu. The DAG-based orchestration means that complex, multi-step processes can be automated efficiently. Developers can integrate these agents into existing applications or use them as the foundation for new AI-powered products. The project's MIT license allows for broad adoption and modification.
Caveats and source limits
The provided repository metadata indicates that the latest release was 155 days ago, suggesting a potential lag in recent updates or active development. While the repository lists many supported tools and departments, specific details on the performance or limitations of individual agents are not detailed in the summary. The agency-orchestrator functionality is mentioned, but its specific implementation, requirements, and ease of use for complex DAGs are not elaborated upon. Documentation is marked as not present, and examples are also absent, which may pose challenges for new users trying to understand and implement the system. The repository is primarily in Shell, which might limit its direct applicability for developers working exclusively in other languages without additional integration layers.
Sources
Claim check: 7/7 supported claims - 7 evidence links - 100% avg confidence
- Offers 277 plug-and-play AI expert roles.supported - github.com
- Supports over 20 tools including Claude Code, Cursor, and Copilot.supported - github.com
- Includes 64 original AI agents for the Chinese market.supported - github.com
- Integrates with agency-orchestrator for DAG-based multi-agent collaboration.supported - github.com
- Repository has 22166 stars.supported - github.com
- Repository has 3556 forks.supported - github.com
- Latest release was v1.2.6 on June 16, 2026.supported - github.com
Caveats
- Single-source caution: verify critical details at the linked source.
Radar score 86/100 - how it was calculated
- Reliability 82: GitHub metadata supports source trust
- Freshness 92: Fresh GitHub activity
- Novelty 71: Novelty blends source metadata and enrichment
- Technical 89: Repository technical metadata
- Developer 96: Developer tooling signals
- Ecosystem 66: Developer-oriented GitHub signal
- Confidence 100: Claims have reliable evidence
Discussion
Loading comments...