Agency Orchestrator is a framework designed to automate complex workflows by orchestrating multiple AI agents. It targets users who need to generate comprehensive plans or analyses by leveraging diverse AI expertise without writing code. The primary benefit is the rapid generation of detailed outputs by simulating a collaborative team of specialized AI roles, accessible via simple natural language prompts or zero-code YAML configurations.

How It Works

The core of Agency Orchestrator lies in its ability to define and execute multi-agent workflows using YAML configurations. Users can leverage a vast library of pre-defined AI roles (over 211 in Chinese, 170+ in English) or define custom ones. The system automatically constructs a Directed Acyclic Graph (DAG) from the workflow definition, enabling intelligent parallel execution of independent tasks. It integrates with numerous Large Language Model (LLM) providers, crucially supporting several that do not require API keys, thus reducing adoption friction and cost.

Quick Start & Requirements

• Primary install: npm install -g agency-orchestrator
• Prerequisites: Node.js and npm.
• LLM Providers: Supports 10 LLM providers including Claude (Max/Pro), Gemini, GitHub Copilot, ChatGPT (Plus/Pro), Ollama (local), DeepSeek, and OpenAI. Notably, 7 providers (Claude Code, Gemini CLI, Copilot CLI, Codex CLI, OpenClaw, Hermes CLI, Ollama) require no direct API key, leveraging existing subscriptions or free tiers.
• Links: Community channels include WeChat Official Account "AI不止语", QQ Group (1071280067), and a WeChat group accessible via the official account.

Highlighted Details

• Features 211+ specialized Chinese AI roles and 170+ English roles.
• Enables zero-code workflow definition and execution via natural language prompts (ao compose) or YAML files.
• Automated DAG generation for efficient parallel task execution.
• Extensive LLM provider support, with a focus on free or subscription-based options to minimize operational costs.
• ao compose command intelligently generates complex workflows from a single sentence prompt.
• Supports iterative refinement of workflows using a resume functionality.

Maintenance & Community

The project is actively developed, with a roadmap outlining future enhancements like a Web UI and visual DAG editor. Community engagement is facilitated through a WeChat official account ("AI不止语"), a QQ group (1071280067), and a WeChat group.

Licensing & Compatibility

The project is licensed under the Apache-2.0 license, which generally permits commercial use and integration into proprietary software, provided attribution and license terms are met.

Limitations & Caveats

Current versions lack a graphical user interface (GUI) and visual DAG editor, which are slated for future releases (v0.6). While many LLM providers are free to use via existing subscriptions or tiers, users are still dependent on the availability, rate limits, and terms of service of these third-party LLM providers.