agency-orchestrator 0.1.0 → 0.2.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +39 -26
- package/README.zh-CN.md +7 -0
- package/dist/index.js +2 -1
- package/package.json +2 -1
package/README.md
CHANGED
|
@@ -1,13 +1,15 @@
|
|
|
1
1
|
# Agency Orchestrator
|
|
2
2
|
|
|
3
|
-
> **The YAML-first multi-agent orchestrator — 186
|
|
3
|
+
> **The YAML-first multi-agent orchestrator — 186 ready-to-use AI roles, zero code, any LLM**
|
|
4
|
+
>
|
|
5
|
+
> **不用写代码的 AI 团队 — 186 个角色开箱即用，YAML 编排，支持 DeepSeek/Claude/OpenAI/Ollama**
|
|
4
6
|
|
|
5
7
|
[](https://github.com/jnMetaCode/agency-orchestrator/actions)
|
|
6
8
|
[](https://www.npmjs.com/package/agency-orchestrator)
|
|
7
9
|
[](./LICENSE)
|
|
8
10
|
[](./CONTRIBUTING.md)
|
|
9
11
|
|
|
10
|
-
[中文文档](./README.zh-CN.md)
|
|
12
|
+
**English** | [中文文档](./README.zh-CN.md)
|
|
11
13
|
|
|
12
14
|
---
|
|
13
15
|
|
|
@@ -21,7 +23,7 @@ Agency Orchestrator turns a YAML file into a multi-agent pipeline. No Python. No
|
|
|
21
23
|
|
|
22
24
|
```python
|
|
23
25
|
# CrewAI: ~50 lines of Python, write every role from scratch
|
|
24
|
-
researcher = Agent(role="PM", goal="...", backstory="...(
|
|
26
|
+
researcher = Agent(role="PM", goal="...", backstory="...(write it yourself)...")
|
|
25
27
|
task = Task(description="...", agent=researcher)
|
|
26
28
|
crew = Crew(agents=[researcher], tasks=[task])
|
|
27
29
|
crew.kickoff()
|
|
@@ -63,37 +65,41 @@ export DEEPSEEK_API_KEY=your-key # or ANTHROPIC_API_KEY, OPENAI_API_KEY
|
|
|
63
65
|
npx ao run workflows/story-creation.yaml --input premise='A time travel story'
|
|
64
66
|
```
|
|
65
67
|
|
|
66
|
-
## Demo:
|
|
68
|
+
## Demo: 3 AI Roles Reviewing a Product in 22 Seconds
|
|
67
69
|
|
|
68
70
|
```
|
|
69
|
-
$ ao run workflows/
|
|
71
|
+
$ ao run workflows/product-review.yaml -i prd_content="Build an AI-powered code review tool"
|
|
70
72
|
|
|
71
|
-
|
|
72
|
-
|
|
73
|
+
Workflow: Product Review
|
|
74
|
+
Steps: 4 | Concurrency: 2 | Model: deepseek-chat
|
|
73
75
|
──────────────────────────────────────────────────
|
|
74
76
|
|
|
75
|
-
── [1/4]
|
|
76
|
-
|
|
77
|
-
|
|
77
|
+
── [1/4] analyze (product/product-manager) ──
|
|
78
|
+
Done | 8.2s | 2,341 tokens
|
|
79
|
+
Key requirements extracted: 1) GitHub/GitLab integration for PR webhooks
|
|
80
|
+
2) Multi-language AST parsing 3) LLM-based review with context window...
|
|
78
81
|
|
|
79
|
-
── [2/4]
|
|
80
|
-
|
|
81
|
-
|
|
82
|
+
── [2/4] tech_review (engineering/engineering-software-architect) ── ← parallel
|
|
83
|
+
Done | 6.8s | 1,892 tokens
|
|
84
|
+
Technical feasibility: HIGH. Recommended stack: Node.js + tree-sitter
|
|
85
|
+
for parsing, streaming API for real-time review feedback...
|
|
82
86
|
|
|
83
|
-
── [3/4]
|
|
84
|
-
|
|
85
|
-
|
|
87
|
+
── [3/4] design_review (design/design-ux-researcher) ── ← parallel
|
|
88
|
+
Done | 6.8s | 1,756 tokens
|
|
89
|
+
UX risks: inline comments may overwhelm developers. Suggest progressive
|
|
90
|
+
disclosure — show summary first, expand details on click...
|
|
86
91
|
|
|
87
|
-
── [4/4]
|
|
88
|
-
|
|
89
|
-
|
|
92
|
+
── [4/4] summary (product/product-manager) ──
|
|
93
|
+
Done | 5.1s | 2,014 tokens
|
|
94
|
+
GO with conditions: strong technical feasibility, address UX concern
|
|
95
|
+
about comment density before launch...
|
|
90
96
|
|
|
91
97
|
==================================================
|
|
92
|
-
|
|
98
|
+
Done: 4/4 steps | 21.9s | 8,003 tokens
|
|
93
99
|
==================================================
|
|
94
100
|
```
|
|
95
101
|
|
|
96
|
-
Steps 2 & 3 ran **in parallel** (auto-detected from DAG).
|
|
102
|
+
Steps 2 & 3 ran **in parallel** (auto-detected from DAG). 3 specialized AI roles (PM, Architect, UX Researcher) collaborated to review the product.
|
|
97
103
|
|
|
98
104
|
## How It Works
|
|
99
105
|
|
|
@@ -231,13 +237,13 @@ console.log(result.totalTokens); // { input: 1234, output: 5678 }
|
|
|
231
237
|
Each run saves to `.ao-output/<name>-<timestamp>/`:
|
|
232
238
|
|
|
233
239
|
```
|
|
234
|
-
.ao-output
|
|
240
|
+
.ao-output/Product-Review-2026-03-21T16-30-00/
|
|
235
241
|
├── summary.md # Final step output
|
|
236
242
|
├── steps/
|
|
237
|
-
│ ├── 1-
|
|
238
|
-
│ ├── 2-
|
|
239
|
-
│ ├── 3-
|
|
240
|
-
│ └── 4-
|
|
243
|
+
│ ├── 1-analyze.md
|
|
244
|
+
│ ├── 2-tech_review.md
|
|
245
|
+
│ ├── 3-design_review.md
|
|
246
|
+
│ └── 4-summary.md
|
|
241
247
|
└── metadata.json # Duration, token usage, step status
|
|
242
248
|
```
|
|
243
249
|
|
|
@@ -253,6 +259,13 @@ agency-orchestrator (this project — YAML workflow engine)
|
|
|
253
259
|
DeepSeek / Claude / OpenAI / Ollama
|
|
254
260
|
```
|
|
255
261
|
|
|
262
|
+
## Sister Projects
|
|
263
|
+
|
|
264
|
+
| Project | Description |
|
|
265
|
+
|---------|-------------|
|
|
266
|
+
| [agency-agents-zh](https://github.com/jnMetaCode/agency-agents-zh) | 186 AI role definitions — the role library this orchestrator is built on |
|
|
267
|
+
| [superpowers-zh](https://github.com/jnMetaCode/superpowers-zh) | AI coding superpowers — 17 skills to make your AI coding assistant actually productive |
|
|
268
|
+
|
|
256
269
|
## Roadmap
|
|
257
270
|
|
|
258
271
|
- [x] **v0.1** — YAML workflow, DAG engine, 4 LLM connectors, CLI, real-time output
|
package/README.zh-CN.md
CHANGED
|
@@ -210,6 +210,13 @@ agency-orchestrator（本项目 — YAML 工作流引擎）
|
|
|
210
210
|
DeepSeek / Claude / OpenAI / Ollama
|
|
211
211
|
```
|
|
212
212
|
|
|
213
|
+
## 姊妹项目
|
|
214
|
+
|
|
215
|
+
| 项目 | 说明 |
|
|
216
|
+
|------|------|
|
|
217
|
+
| [agency-agents-zh](https://github.com/jnMetaCode/agency-agents-zh) | 186 个 AI 角色定义 — 本编排引擎的角色库 |
|
|
218
|
+
| [superpowers-zh](https://github.com/jnMetaCode/superpowers-zh) | AI 编程超能力 · 中文版 — 17 个 skills，让你的 AI 编程助手真正会干活 |
|
|
219
|
+
|
|
213
220
|
## 路线图
|
|
214
221
|
|
|
215
222
|
- [x] **v0.1** — YAML 工作流、DAG 引擎、4 个 LLM 连接器、CLI、实时输出
|
package/dist/index.js
CHANGED
|
@@ -124,11 +124,12 @@ function resolveAgentsDir(agentsDir, workflowPath) {
|
|
|
124
124
|
const relToWorkflow = resolve(dirname(workflowPath), agentsDir);
|
|
125
125
|
if (existsSync(relToWorkflow))
|
|
126
126
|
return relToWorkflow;
|
|
127
|
-
// 3.
|
|
127
|
+
// 3. 常见位置（包括 npm 依赖自带的）
|
|
128
128
|
const candidates = [
|
|
129
129
|
resolve('agency-agents-zh'),
|
|
130
130
|
resolve('../agency-agents-zh'),
|
|
131
131
|
resolve('node_modules/agency-agents-zh'),
|
|
132
|
+
resolve(dirname(new URL(import.meta.url).pathname), '../../node_modules/agency-agents-zh'),
|
|
132
133
|
];
|
|
133
134
|
for (const dir of candidates) {
|
|
134
135
|
if (existsSync(dir))
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "agency-orchestrator",
|
|
3
|
-
"version": "0.
|
|
3
|
+
"version": "0.2.0",
|
|
4
4
|
"description": "基于 agency-agents 角色定义的轻量多智能体编排引擎 — 用 YAML 定义工作流，自动调度 AI 角色协作完成复杂任务",
|
|
5
5
|
"keywords": [
|
|
6
6
|
"agency-agents",
|
|
@@ -56,6 +56,7 @@
|
|
|
56
56
|
},
|
|
57
57
|
"dependencies": {
|
|
58
58
|
"@anthropic-ai/sdk": "^0.52.0",
|
|
59
|
+
"agency-agents-zh": "^1.0.0",
|
|
59
60
|
"js-yaml": "^4.1.0"
|
|
60
61
|
},
|
|
61
62
|
"devDependencies": {
|