Skip to content

DrenLea/hackathon-topic-advisor

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🏆 Hackathon Topic Advisor

Stop building "another todo app." Find a blue ocean topic before you write a single line of code.

A cross-platform AI Skill that validates your hackathon project idea through multi-round dialogue, multi-channel search, social demand discovery, and three-dimensional scoring. Works with TRAE, Claude Code, Cursor, Windsurf, and any tool that reads Markdown.


Why This Exists

66% of hackathon projects are "reskins" — todo apps, weather dashboards, GPT wrappers. They don't win. The #1 factor that determines hackathon success is topic selection, not code quality. This skill helps you pick a topic that is:

  • Rare — not already built by 500 other teams
  • Original — not a reskin of a common pattern
  • Meaningful — solves a real problem with real users

How It Works

User describes idea
       │
       ▼
┌──────────────┐
│  1. Clarify   │  AI asks 2-3 sharp questions
│  (反问)       │  Who's the user? What's the core scenario?
└──────┬───────┘
       ▼
┌──────────────┐
│  2. Search    │  GitHub · Devpost · Product Hunt
│  (同类搜索)   │  How many similar projects exist?
└──────┬───────┘
       ▼
┌──────────────┐
│  3. Discover  │  Reddit · Twitter · V2EX · Zhihu
│  (社媒需求)   │  "I wish there was a tool that..."
└──────┬───────┘
       ▼
┌──────────────┐
│  4. Score     │  Python script → deterministic score
│  (三维评分)   │  Scarcity + Originality + Meaning
└──────┬───────┘
       ▼
┌──────────────┐
│  5. Report    │  🌊 Blue / ⚡ Yellow / 🔥 Red
│  (分析报告)   │  + differentiation strategies
└──────┬───────┘
       ▼
┌──────────────┐
│  6. Iterate   │  Adjust idea → re-analyze
│  (迭代优化)   │  until you find a winner
└──────────────┘

Scoring System

Composite = Scarcity × 40% + Originality × 35% + Meaning × 25%

Score Ocean Type What It Means
≥ 70 🌊 Blue Ocean Scarce, original, high social value — go build it!
45–69 Yellow Ocean Some competition, but you can differentiate
< 45 🔥 Red Ocean Saturated. Pivot or find a unique angle.

What Makes Each Dimension

Dimension Weight How It's Calculated
Scarcity 40% GitHub + Devpost search result count and hit ratio. 0 results = 95 pts; 5000+ results = 10 pts.
Originality 35% Base 70. Subtract penalty for matched common patterns (25 patterns in library). Add bonus for originality boosters (accessibility, AI agent, AR, etc.).
Meaning 25% Base 50. Add bonus for social impact boosters (elderly, disaster relief, rural, etc.). New: social demand modifier from Reddit/Twitter quotes (+10 / -10).

Quick Start

Install

TRAE (project-level):

cp -r hackathon-topic-advisor/ .trae/skills/

Claude Code (project-level):

cp -r hackathon-topic-advisor/ .claude/skills/

Cursor / Windsurf / others: see install/other-tools.md

Use

Just talk naturally — the skill auto-activates when you mention hackathons:

You: "我想参加黑客松,想做一个帮老年人管理用药的App"

AI: "这个方向有意思。在搜索之前我想确认几点:1. 你的目标用户是独居老人还是养老院?2. 核心功能是提醒吃药还是药物相互作用检测?..."

Or explicitly invoke: /hackathon-topic-advisor

Example Output

# 🏆 选题分析报告

## 📊 三维评分
| 维度 | 得分 | 说明 |
|------|------|------|
| 原创性 | 82/100 | 匹配 accessibility + deaf 加分 |
| 稀缺度 | 89/100 | GitHub 仅5个同类仓库 |
| 意义感 | 76/100 | 社媒强需求 +10 |
| **综合** | **83/100** | |

## 🌊 海域判定
🌊 蓝海项目!这个方向稀缺度高,社会价值显著,有很大机会脱颖而出。

## 💬 社媒需求发现
- 需求强度: Strong
- 用户原话摘录:
  > "I wish there was a real-time sign language translator..." — Reddit user
  > "要是有一个能把手语翻译成文字的工具就好了" — V2EX 用户

What's Inside

hackathon-topic-advisor/
├── SKILL.md                    # Core methodology (6-phase workflow)
├── references/
│   ├── concept-map.md          # 110 CN→EN keyword mappings
│   ├── common-patterns.md     # 25 common "reskin" patterns + differentiation strategies
│   ├── blue-ocean.md           # 8 blue ocean directions + 18 originality boosters
│   └── scoring-rules.md        # Scoring algorithm with worked examples
├── scripts/
│   └── topic_scorer.py         # Pure Python scorer (stdlib only, no pip install)
├── tests/
│   └── test_topic_scorer.py    # 9 unit tests (all passing)
├── install/
│   ├── trae.md                 # TRAE setup
│   ├── claude-code.md          # Claude Code setup
│   └── other-tools.md          # Cursor / Windsurf / AGENTS.md
└── README.md                   # You are here

Data Assets

Asset Count Source
CN→EN concept mappings 110 From HackCheck's conceptMap
Common project patterns 25 From HackCheck's commonPatterns
Blue ocean directions 8 From HackCheck's blueOceanDirections
Originality boosters 18 From HackCheck's originalityBoosters

Try the Scorer Standalone

echo '{"description":"AI sign language glove","search_total_count":5,"search_hit_ratio":0.2,"matched_patterns":[],"matched_boosters":["accessibility","deaf"],"social_demand_level":"strong"}' | python scripts/topic_scorer.py

Output:

{
  "originality": 82,
  "scarcity": 89,
  "meaning": 76,
  "composite": 83,
  "ocean_type": "blue",
  "verdict": "蓝海项目!这个方向稀缺度高,社会价值显著,有很大机会脱颖而出。"
}

Cross-Platform Compatibility

Platform Status How It Works
TRAE ✅ Native Reads .trae/skills/ — uses name + description fields
Claude Code ✅ Native Reads .claude/skills/ — uses full frontmatter (triggers, tags, allowed-tools)
Cursor ✅ Adapted Copy SKILL.md body to .cursor/rules/
Windsurf ✅ Adapted Copy SKILL.md body to .windsurfrules
Any AGENTS.md tool ✅ Adapted Reference SKILL.md in AGENTS.md

The SKILL.md frontmatter includes fields for both TRAE and Claude Code. Each platform reads what it needs and ignores the rest — one file, all platforms.

Requirements

  • Python 3.6+ (only for the scoring script; AI falls back to manual scoring if unavailable)
  • Any AI coding tool that supports custom instructions/skills

Origin

Derived from HackCheck — a hackathon full-process assistant web app. The topic validation module was extracted, enhanced with social demand discovery and multi-round dialogue, and packaged as a portable skill.

Enhancements over the original web app:

  • AI asks clarifying questions before searching (multi-round dialogue)
  • Searches social media for real user demand quotes (Reddit, Twitter, V2EX, Zhihu)
  • Cross-platform — works in any vibe-coding tool, not just a browser
  • Lightweight Python scoring script — deterministic, testable, no dependencies

License

MIT

About

a skill that can be your hackathon-topic-advisor

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages