Founder Competitive Analysis

Preamble (run first)

_FF_DIR="${FFOUNDER_DIR:-$(find ~/.claude/skills -maxdepth 1 -name 'f-founder' -type d 2>/dev/null | head -1)}"
[ -z "$_FF_DIR" ] && _FF_DIR="$(find .claude/skills -maxdepth 1 -name 'f-founder' -type d 2>/dev/null | head -1)"
if [ -n "$_FF_DIR" ] && [ -f "$_FF_DIR/bin/f-founder-update-check" ]; then
  _UPD=$("$_FF_DIR/bin/f-founder-update-check" 2>/dev/null || true)
  [ -n "$_UPD" ] && echo "$_UPD" || true
fi
_FF_VER=$(cat "$_FF_DIR/VERSION" 2>/dev/null | tr -d '[:space:]' || echo "unknown")
echo "F-FOUNDER: v$_FF_VER"

If output shows UPGRADE_AVAILABLE: tell user to upgrade before proceeding.

founder-competitive-analysis: Competitive Analysis for Founders

Generate a deep competitive analysis report that identifies competitors, researches their products, growth, and user feedback, then produces a structured Markdown + PDF report with opportunity insights.

Overview

What it produces:

• report.md — comprehensive Markdown report with cross-competitor comparisons and insights
• report.pdf — professionally styled PDF (brand-matched if pitch-deck exists, default style otherwise)

Key features:

• Problem-first approach — starts from the user problem, not product categories
• Three-tier competitor discovery — direct, indirect, and potential competitors
• Parallel deep research — one subagent per competitor, all running simultaneously
• Real user feedback — sourced from G2, Capterra, Reddit, Product Hunt
• Opportunity synthesis — market gaps, differentiation angles, risk flags
• Brand-consistent PDF — reuses pitch-deck design tokens when available

Workflow: 4 steps with 3 founder confirmation gates.

---

Tool Dependencies

| Tool | Required | Purpose | |------|----------|---------| | WebSearch | Yes | Discovery, product info, funding, news | | WebFetch | Yes | Fetch pages (company sites, G2, Reddit posts) | | Agent | Yes | Parallel subagent deep research | | Browse | Optional | JS-heavy page scraping | | pdf | Optional | Parse PRD PDF input + PDF report generation |

---

反讨好规则

竞品分析必须客观，禁止：

• 淡化竞争威胁（"虽然竞争激烈但仍有机会" → 说清楚具体哪里有机会、有多大）
• 给市场空白加乐观包装（"存在差异化空间" → 说清楚空白是什么、为什么没人做、是真空白还是伪需求）
• 隐瞒不利数据（竞品融了 $100M 就说 $100M，不要说"获得了可观的融资"）
• 编造或推测数据（找不到就写"未找到公开数据"）

必须：

• 竞品强就说强，附证据
• 市场拥挤就说拥挤，列出玩家数量和融资总额
• 用户反馈差就如实呈现差评，不要只挑好的
• 机会分析要有数据支撑，不要空谈

Step 1 — Problem Definition

Understand what problem the founder is trying to solve. This frames ALL subsequent research.

复用已有输出

如果 founder 之前已经跑过 /founder-requirements-clarification，检测当前项目目录下是否有需求澄清的输出。如果有：

• 直接复用已确认的核心问题、目标用户、关键需求
• 跳过问题定义的对话，告知 founder "检测到你之前的需求澄清结果，直接使用"
• 仅确认是否需要调整

如果没有，按正常流程执行问题定义。

Input Detection

If the source arg is provided:

• .md file → read directly, extract problem definition
• .pdf file → invoke pdf skill to parse, then extract
• .txt / .docx file → read directly
• No file extension → treat as free-text description

If no source arg, ask the founder: > "What problem are you trying to solve? You can describe it in your own words, or point me to a PRD file."

Processing

Extract and structure into:

## Problem Definition

**Core Problem:** One sentence — the user pain point
**Target Users:** Who has this problem (be specific)
**Use Cases:** When/where users encounter this problem
**Key Needs:** 3-5 core user needs (these become search keywords)
**Current Solutions:** How users solve this today (if mentioned)

If input is a PRD:

• Also extract: product name, positioning, target market
• Use product name to EXCLUDE from competitor search results

Gate

Present the structured problem definition to the founder: > "Here is my understanding of the problem. Is this accurate? Any adjustments?"

Wait for confirmation before proceeding.

---

Step 2 — Competitor Discovery

Find products that address the same or similar user needs. Cast a wide net, then narrow.

Reference: Read references/search-strategies.md Section 1 for complete query templates.

2a. Construct Search Queries

From Step 1's Key Needs and Core Problem, construct queries for three search paths:

Path 1 — General Web Search (WebSearch):

• "best [need keywords] tools/apps [current year]"
• "[need keywords] alternatives"
• "[core problem] solution/software"

Path 2 — Reddit (WebSearch with site: filter):

• "site:reddit.com [need keywords] recommendation"
• "site:reddit.com best [product type]"
• WebFetch top threads to extract product names and sentiment

Path 3 — App Market / SaaS Reviews (adaptive):

• If mobile-relevant: search App Store / Google Play via WebSearch
• If SaaS/Web: search G2 / Capterra via WebSearch
• If unclear: search both

2b. Supplementary Sources

After initial results, check:

• AlternativeTo — search by keyword or discovered competitor name
• Product Hunt — search related products
• G2 / Capterra — search by category

2c. Process Results

1. Deduplicate — merge same product found across multiple sources
2. Score relevance — how directly does this product address the founder's core problem?
3. Classify into three tiers:

| Tier | Definition | |------|-----------| | Direct | Solves the same problem with a similar approach | | Indirect | Satisfies the same need through a different method | | Potential | Adjacent product that could expand into this space |

1. Present to founder as a table:

## Competitor Discovery Results

### Direct Competitors
| # | Product | Description | Relevance | Sources |
|---|---------|-------------|-----------|---------|

### Indirect Competitors
| # | Product | Description | Relevance | Sources |
|---|---------|-------------|-----------|---------|

### Potential Competitors
| # | Product | Description | Relevance | Sources |
|---|---------|-------------|-----------|---------|

Gate

> "Here are the competitors I found. Do you want to add, remove, or adjust any before I start deep research? Once confirmed, I will research each one in depth."

Wait for confirmation. Apply any changes the founder requests.

---

Step 3 — Deep Research (Parallel Subagents)

Research each confirmed competitor in depth using parallel Agents.

Reference: Read references/subagent-prompt.md for the complete prompt template. Reference: Read references/search-strategies.md Section 2 for per-module query templates.

3a. Dispatch Agents

For each competitor in the confirmed list, launch one Agent in parallel:

Agent(
  description="Research [CompetitorName]",
  prompt=,
  subagent_type="general-purpose"
)

All Agents launch in a single message for maximum parallelism.

Each Agent prompt includes:

• Competitor name and one-line description
• Core problem definition (from Step 1)
• Competitor tier (direct/indirect/potential)
• Standardized research task list (Module A: Basics, Module B: Growth, Module C: Feedback)
• Required output format

3b. Collect and Quality-Check Results

After all Agents complete:

1. Scan for completeness — check each result for critical fields:

• Core Features (Module A) — MUST have data
• Pricing (Module B) — MUST have data
• User Feedback (Module C) — SHOULD have data from at least one source

1. Flag data gaps — if a competitor has 3+ critical fields marked "No public data found", run a supplementary Agent with narrower queries.

1. Cross-check consistency — if two Agents reference the same product differently (e.g., conflicting founding dates), flag for the final report.

3c. Progress Update

After all Agents complete: > "Deep research complete. Researched [n] competitors. [Brief summary of data quality — e.g., 'Strong data for 5, limited public info for 2.'] Generating report now."

---

Step 4 — Report Generation

Assemble findings into a structured Markdown report, then generate PDF.

Reference: Read references/report-structure.md for complete report template and PDF generation instructions.

4a. Assemble Markdown Report

Follow the template in references/report-structure.md Section 1:

1. Executive Summary — write LAST, after all other sections. Synthesize the 3-5 most important findings and the overall opportunity signal.

1. Cross-Competitor Overview — build comparison tables from Agent outputs:

• Basic Info table (launch, funding, team, pricing)
• Feature Matrix (use Key Needs from Step 1 as rows)
• Growth Comparison table

1. Detailed Competitor Analysis — paste each Agent's full output, grouped by tier (Direct → Indirect → Potential)

1. User Feedback Insights — SYNTHESIZE across all competitors:

• Cross-competitor patterns (what do users consistently praise/complain about?)
• Sentiment ranking
• Unmet needs table with frequency and quotes
• Flag needs NO competitor addresses

1. Opportunity Analysis — connect dots between:

• Unmet user needs × market gaps → differentiation angles
• Competitor weaknesses × founder's strengths → strategic entry points
• Big company moves × regulatory changes → risk flags

1. Data Sources — consolidate all reference URLs from all Agents

4b. Save Markdown Report

mkdir -p competitive-analysis

Write the report to competitive-analysis/report.md.

4c. Generate PDF

Design token detection (see references/report-structure.md Section 3):

1. Look for pitch-deck index.html in current project — extract CSS custom properties
2. Look for design-tokens.json
3. If neither found, use defaults:

• Primary: #1a1a2e, Accent: #e94560, BG: #ffffff
• Font: Inter + Noto Sans SC

Generation:

1. Read assets/report-template.html
2. Replace {{PLACEHOLDER}} tokens with design values:

• {{REPORT_TITLE}} — "[Product/Project Name] Competitive Analysis Report"
• {{PRIMARY}} — primary brand color
• {{ACCENT}} — accent color
• {{BG}} — background color
• {{BG_SECONDARY}} — secondary background
• {{TEXT}} — primary text color
• {{TEXT_SECONDARY}} — secondary text color
• {{BORDER}} — border color
• {{FONT_MAIN}} — main font stack
• {{FONT_MONO}} — monospace font stack
• {{GOOGLE_FONTS_LINK}} — full `` tag
• {{REPORT_CONTENT}} — Markdown converted to HTML

1. Convert report.md to HTML: transform markdown headings, tables, lists, blockquotes, and inline formatting into corresponding HTML elements
2. Inject the HTML into the template's {{REPORT_CONTENT}} placeholder
3. Write to competitive-analysis/report.html

PDF output:

• If browse skill is available: use it to open report.html and print to PDF → competitive-analysis/report.pdf
• Otherwise: tell the founder to open report.html in browser and use Ctrl+P / Cmd+P to save as PDF

4d. 附录：中间材料

最终报告必须包含以下附录：

附录 A：问题定义

• 核心问题、目标用户、关键需求的结构化输出

附录 B：竞品发现原始结果

• 所有搜索查询（实际使用的 query）
• 每条 query 返回的原始结果摘要
• 去重和筛选前的完整列表

附录 C：Subagent 原始输出

• 每个竞品的 Agent 原始返回（未经编辑）
• 标注哪些字段是 Agent 原始输出，哪些是主进程补充的

附录 D：Founder 确认记录

• Step 1 问题定义确认
• Step 2 竞品列表确认（含 founder 的增删调整）

Gate

> "Report generated: competitive-analysis/report.md and competitive-analysis/report.html. > [If PDF generated:] PDF saved to competitive-analysis/report.pdf. > [If no Browse:] Open report.html in your browser and print to PDF (Cmd+P). > > Would you like me to dig deeper into any specific competitor?"

If the founder requests deeper research on specific competitors, dispatch targeted Agents and update the report.

下一步建议

报告完成后，根据 founder 的情况推荐：

• 如果发现了明确的市场空白 → "建议使用 /founder-synthetic-research 对目标用户做合成访谈，验证这些机会是否真实"
• 如果已经验证了方向 → "建议使用 /founder-pitch-deck 生成 pitch deck，竞品分析数据可以直接用于 Competition slide"
• 如果问题定义还不够清晰 → "建议先使用 /founder-requirements-clarification 梳理需求"

---

Quick Reference

Trigger: /founder-competitive-analysis or describe a competitive analysis need Input: Free text, or source arg pointing to a PRD file Output: competitive-analysis/report.md + report.html + report.pdf Duration: Deep research mode — thorough analysis with parallel subagents Gates: 3 confirmation points (problem → competitor list → final report)

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: fourierwang66666
• Source: fourierwang66666/f-founder
• License: MIT

Install and usage instructions live in the source repository linked above.