idea-reality-mcp MCP — 사용 사례, 설치 & 라이브 데모

왜 쓰나요

핵심 기능

One tool: idea_check with quick (<3s) or deep mode
Scans 6 sources: GitHub, HN, npm, PyPI, Product Hunt, SO
Returns 0-100 reality score + trend direction + pivot ideas
Zero auth needed for basic use; optional tokens for higher limits

라이브 데모

실제 사용 모습

idea-reality.replay ▶ 준비됨

0/0

설치

클라이언트 선택

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "idea-reality": {
      "command": "uvx",
      "args": [
        "idea-reality-mcp"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config 열기. 저장 후 앱 재시작.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "idea-reality": {
      "command": "uvx",
      "args": [
        "idea-reality-mcp"
      ],
      "_inferred": true
    }
  }
}

Cursor는 Claude Desktop과 동일한 mcpServers 스키마 사용. 프로젝트 설정이 전역보다 우선.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "idea-reality": {
      "command": "uvx",
      "args": [
        "idea-reality-mcp"
      ],
      "_inferred": true
    }
  }
}

Cline 사이드바의 MCP Servers 아이콘 클릭 후 "Edit Configuration" 선택.

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "idea-reality": {
      "command": "uvx",
      "args": [
        "idea-reality-mcp"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop과 같은 형식. Windsurf 재시작 후 적용.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "idea-reality",
      "command": "uvx",
      "args": [
        "idea-reality-mcp"
      ]
    }
  ]
}

Continue는 맵이 아닌 서버 오브젝트 배열 사용.

~/.config/zed/settings.json

{
  "context_servers": {
    "idea-reality": {
      "command": {
        "path": "uvx",
        "args": [
          "idea-reality-mcp"
        ]
      }
    }
  }
}

context_servers에 추가. 저장 시 Zed가 핫 리로드.

claude mcp add idea-reality -- uvx idea-reality-mcp

한 줄 명령. claude mcp list로 확인, claude mcp remove로 제거.

사용 사례

실전 활용법: idea-reality-mcp

Validate a SaaS idea in 10 seconds before you build

👤 Founders, indie hackers, side-project tinkerers ⏱ ~10 min beginner

언제 쓸까: That 'I should build X' feeling at 2am — run it through reality check first.

흐름

Quick check

Run idea_check on 'a mobile app that uses ML to predict plant watering needs'. Quick depth.✓ 복사됨

→ Score 0-100 + top 5 existing projects
Deep dive if promising

If the quick score is <60, run deep mode for full market scan including Product Hunt launches.✓ 복사됨

→ Expanded list with trend arrows + funding signals
Pivot if crowded

Given the top 5 competitors, suggest 3 pivot angles that are less crowded.✓ 복사됨

→ Specific, narrower positioning options

결과: 10 minutes of evidence-based triage instead of 10 weekends of parallel-inventing.

함정

Score high just because nobody built it — maybe because no one wants it — Look at HN comments + SO question volume; zero demand signal is as bad as saturation

Map existing competitors before a product kickoff

👤 PMs, founders writing a one-pager ⏱ ~20 min intermediate

언제 쓸까: You're scoping a product and need a real competitor list, not your own biased knowledge.

사전 조건

GITHUB_TOKEN for higher rate limits — github.com/settings/tokens — no scopes needed

흐름

Run deep mode

Deep idea_check on 'realtime collaborative code review tool for remote teams'.✓ 복사됨

→ List of 10+ competitors with stars/launches/dates
Cluster by approach

Group the competitors by approach (IDE plugin vs standalone vs browser extension). What's missing?✓ 복사됨

→ Gap analysis

결과: A one-pager with a real competitive map.

함께 쓰기: firecrawl

조합

다른 MCP와 조합해 10배 효율

idea-reality + firecrawl

Deep-dive a competitor's landing/pricing page for real positioning

idea_check flagged Competitor X. Use firecrawl to scrape their pricing page and summarize their plans.✓ 복사됨

idea-reality + notion

Dump validation results into a validation journal

For today's idea_check on 'AI for gardeners', create a Notion page with the score and competitor list.✓ 복사됨

도구

이 MCP가 노출하는 것

도구	입력	언제 호출	비용
idea_check	idea_text: str, depth?: 'quick'\|'deep'	Any idea worth more than 30 min of your time before building	Free public APIs; optional GitHub token raises limits

비용 및 제한

운영 비용

API 쿼터: GitHub unauth: 60/hr; with token: 5k/hr. Others: public rate limits
호출당 토큰: Quick: 500-1500; deep: 2-5k
금액: Free
팁: Always start with quick mode; deep is expensive and rarely changes the verdict on saturated spaces

보안

권한, 시크릿, 파급범위

자격 증명 저장: Optional GITHUB_TOKEN, PRODUCTHUNT_TOKEN in env

데이터 외부 송신: api.github.com, news.ycombinator.com, registry.npmjs.org, pypi.org, api.producthunt.com, api.stackexchange.com

절대 부여 금지: repo write scopes

문제 해결

자주 발생하는 오류와 해결

Rate limited (GitHub 403)

Set GITHUB_TOKEN env var with a token (no scopes required)

확인: curl -H 'Authorization: Bearer <token>' https://api.github.com/rate_limit

Deep mode returns empty Product Hunt

Requires PRODUCTHUNT_TOKEN — free to get at producthunt.com/v2/oauth

Score seems wildly off

Phrase the idea more specifically; 'AI app' matches everything and scores 0

대안

idea-reality-mcp 다른 것과 비교

대안	언제 쓰나	단점/장점
Manual Google + GitHub search	You have 2 hours to spare	Slower, less systematic
Exploding Topics	You want trend data, not competitor lists	Paid, no MCP

idea-reality-mcp