# Multi Agent Workflow

> Run a two-stage research then planning workflow for growth campaigns. Use when the user wants to analyze past campaign data AND create a new plan based on that analysis in one connected workflow. Stage 1: analyze data and extract structured insights. Stage 2: use insights to build a campaign plan. Input: past campaign data, target, budget, timeline. Output: structured analysis followed by a specific campaign plan.

- Skill: `thaolst/multi-agent-workflow` (Agent Skill)
- Install (CLI): `npx skillmds@latest add thaolst/multi-agent-workflow`
- Raw SKILL.md: https://api.skillmd.com/api/skills/thaolst/multi-agent-workflow/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Marketing & Growth
- Author: thaolst (https://skillmd.com/u/thaolst)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/thaolst/multi-agent-workflow

---


# Multi-Agent Workflow: Research + Plan

Hai giai đoạn nối tiếp nhau. Output của giai đoạn 1 là input của giai đoạn 2.

## Giai đoạn 1 — Research Agent

Đọc data campaign và xuất ra JSON có cấu trúc:

```json
{
  "top_performing_mechanics": [],
  "underperforming_areas": [],
  "segment_insights": {},
  "recommended_focus": [],
  "data_gaps": []
}
```

Chỉ trả về JSON, không giải thích thêm.

## Giai đoạn 2 — Strategy Agent

Nhận JSON từ giai đoạn 1. Dựa trên analysis, viết campaign plan với:
- Strategy tổng thể và lý do
- Campaign cụ thể với mechanic, budget, timeline
- Giải thích tại sao mỗi quyết định dựa trên insight từ giai đoạn 1

## Nguyên tắc

Không bỏ qua data gaps từ giai đoạn 1. Mọi quyết định trong plan phải có dẫn chứng từ analysis.

---

# English

Two stages in sequence. Output of stage 1 is input of stage 2.

Stage 1: Analyze campaign data, return structured JSON with top performing mechanics, underperforming areas, segment insights, recommended focus, and data gaps.

Stage 2: Use the JSON analysis to build a specific campaign plan. Every decision must reference insights from stage 1.