market-analyst - SKILL.md Agent Skill

name: market-analyst description: Conduct product market analysis, competitive intelligence, and feature benchmarking. Use when the user asks for market research, competitor comparison, product teardown, TAM/SAM/SOM estimation, or executive-ready analysis reports. Generates structured Markdown preview first, then exports PDF/DOCX on request. Do NOT use for financial trading, stock picks, or investment advice. homepage: https://github.com/z22811584/market-analyst-skill metadata: {"openclaw":{"emoji":"📊","requires":{"bins":["curl","jq"],"anyBins":["pandoc","wkhtmltopdf"],"env":[]},"primaryEnv":"SIMILARWEB_API_KEY","install":[{"id":"jq-brew","kind":"brew","formula":"jq","bins":["jq"],"label":"Install jq (brew)"},{"id":"pandoc-brew","kind":"brew","formula":"pandoc","bins":["pandoc"],"label":"Install pandoc (brew)"}]}}

Market Analyst — 產品市場分析 & 競品報告

Generate executive-ready market analysis reports covering product positioning, competitive landscape, and feature benchmarking. Output a Markdown preview in chat first, then export PDF or DOCX on user request.

When to Trigger

Activate on any of these intents:

"分析 XX 產品的市場" / "market analysis for XX"
"XX 跟 YY 的競品比較" / "competitor comparison"
"XX 的功能分析" / "product feature teardown"
"TAM SAM SOM" / "市場規模估算"
"產品報告" / "executive report" / "簡報型分析"

Workflow (5 Steps)

Step 1 — Scope Definition

Collect from user (ask if missing):

Target product — the product or company to analyze
Competitors — list of 2-5 competitors (or ask agent to identify top 3-5)
Focus areas — pick from: market sizing, competitive landscape, feature comparison, pricing, SWOT, go-to-market strategy
Audience — default: executive summary for C-level / 高管簡報

Store answers as variables for subsequent steps.

Step 2 — Data Collection (Multi-Source)

Execute data gathering in parallel where possible. Use the strategies below in priority order.

2a. Web Research (always available)

Use the browser tool or web search to collect:

Company "About" pages, press releases, blog announcements
Crunchbase / PitchBook profiles (funding, valuation, headcount)
G2 / Capterra / TrustRadius reviews and ratings
Industry analyst reports (Gartner, Forrester, IDC — summaries only)
News articles from last 12 months
App store listings and changelogs

Search queries to run (replace {product} with target):

{product} market share 2025
{product} vs {competitor} comparison
{product} pricing plans
{product} series funding valuation
{product} G2 reviews rating
{product} product roadmap announcements
{industry} TAM SAM SOM market size
{product} monthly visits traffic similarweb
{product} DAU MAU active users
{industry} top sites traffic ranking
site:similarweb.com {product}
site:semrush.com {product} traffic

2b. API Data (when keys configured)

If SIMILARWEB_API_KEY is set in env or config:

bash {baseDir}/scripts/fetch_similarweb.sh "{domain}" "$SIMILARWEB_API_KEY"

Returns: monthly visits, traffic sources, geo distribution, top referrals.

If CRUNCHBASE_API_KEY is set:

bash {baseDir}/scripts/fetch_crunchbase.sh "{company_name}" "$CRUNCHBASE_API_KEY"

Returns: funding rounds, investors, employee count, categories.

If no API keys configured — skip gracefully, rely on web research only. Do NOT ask user for API keys unless they mention wanting API-sourced data.

2c. User-Uploaded Documents

If user provides files (PDF, Excel, CSV, PPTX):

Read the file using appropriate tool (file read / pdf extraction)
Extract relevant data points: revenue figures, user counts, feature lists, internal metrics
Integrate into analysis — label as "internal data" in the report

Step 3 — Analysis & Structuring

Process collected data through these analysis frameworks. See {baseDir}/references/analysis-frameworks.md for detailed methodology.

Run each applicable framework:

Market Sizing — TAM → SAM → SOM with sources and assumptions
Traffic & User Analysis — Top N players, monthly/daily visits, DAU/MAU, traffic sources, traffic ceiling estimation
Competitive Positioning — 2x2 matrix (pick relevant axes per industry)
Feature Comparison — normalized scoring table (1-5 scale)
SWOT Analysis — for target product relative to market
Pricing Intelligence — tier comparison table with value-per-dollar notes
Key Findings — 3-5 bullet executive takeaways

Step 4 — Report Generation (Chat Preview)

Output the report in Markdown directly in chat using this structure. See {baseDir}/references/report-template.md for the full template.

Report Structure:

# {Product} 市場分析報告
> 報告日期: {date} | 分析師: OpenClaw Market Analyst

## TL;DR — 執行摘要
(3-5 bullet points, each ≤2 sentences)

## 1. 市場概覽
### 1.1 市場規模 (TAM/SAM/SOM)
### 1.2 市場趨勢與驅動力

## 2. 流量與用戶分析
### 2.1 市場 Top N 玩家流量排名
(table: 排名 | 網站/平台 | 月訪問量 | 日均訪問量 | DAU | MAU | 主要流量來源 | YoY 變化)
### 2.2 流量天花板估算
(Based on TAM user base × conversion × frequency; provide optimistic/base/conservative scenarios)
### 2.3 流量結構分析
(traffic source breakdown: direct/search/social/referral; geo distribution; device split)
(If target market includes a specific region like Chinese-speaking, add regional Top N platforms and user characteristics)

## 3. 競品格局
### 3.1 競品定位矩陣
(Use table-based matrix with clear X/Y axis labels -- NEVER use ASCII art charts, they break in PDF)
### 3.2 競品概覽表
(table: 公司 | 成立年份 | 融資 | 估值 | 員工數 | 核心產品)

## 4. 功能對比分析
### 4.1 功能評分矩陣
(table: 功能 | {product} | {comp1} | {comp2} | ... scored 1-5)
### 4.2 差異化亮點

## 5. 定價分析
(table: 方案 | {product} | {comp1} | {comp2} | ...)

## 6. SWOT 分析
(2x2 grid: Strengths | Weaknesses | Opportunities | Threats)

## 7. 關鍵發現與建議
(numbered list, 3-5 actionable items)

## 附錄
### 資料來源
### 方法論說明
### 免責聲明

Formatting rules for executive readability:

Use tables for any comparison data — executives scan tables, not paragraphs
Bold the single most important number in each section
Keep body text ≤3 sentences per paragraph
Use PDF-safe text indicators in matrices: [H]=High/Good, [M]=Medium, [L]=Low/Weak (avoid emoji like 🟢🟡🔴 — wkhtmltopdf cannot render them, causing blank cells)
Chinese (zh-TW) as default language; switch to English if user writes in English

Step 5 — Export (On Request)

When user asks to export ("匯出", "下載", "export", "PDF", "DOCX"):

PDF export (preferred for executives):

bash {baseDir}/scripts/export_report.sh pdf "{report_path}" "{output_path}"

DOCX export:

bash {baseDir}/scripts/export_report.sh docx "{report_path}" "{output_path}"

Save output to ~/.openclaw/workspace/outputs/market-analyst/ with filename pattern: {product}-market-analysis-{YYYY-MM-DD}.{ext}

After export, send the file to the user in the current chat channel.

Guardrails

NEVER present estimates as facts — always label assumptions and confidence level
NEVER fabricate data points — if data is unavailable, state "data not available" explicitly
ALWAYS include source attribution in the appendix
ALWAYS add disclaimer: "本報告基於公開資訊分析，不構成投資建議"
If user asks for stock/investment recommendations, decline and suggest consulting a financial advisor
Rate-limit API calls: max 5 requests per source per analysis run

Failure Handling

API key missing → skip that data source, note in report: "（{source} 數據未取得，因未設定 API 金鑰）"
API rate limited → retry once after 30s, then skip with note
No competitors found → ask user to provide 2-3 competitor names
Insufficient data for a section → include section header with note: "數據不足，建議補充內部資料"
Export tool missing → fall back to raw Markdown file output