idea-reality-mcp-validation

v2026.09.25

Pre-build reality check for AI coding agents — scan GitHub, HN, npm, PyPI, Product Hunt to validate ideas before building

GitHub
安装命令
npx skhub add reason-machines/idea-reality-mcp-validation
Markdown
SKILL.md

idea-reality-mcp-validation

Skill by ara.so — MCP Skills collection.

idea-reality-mcp is an MCP server that validates project ideas before you write code. It scans GitHub, Hacker News, npm, PyPI, Product Hunt, and Stack Overflow to return a 0–100 reality score, trend detection, top competitors, and pivot suggestions.

Installation

Quick Start (uvx)

uvx idea-reality-mcp

Add to Claude Code

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

Add to Claude Desktop / Cursor

Edit your MCP config file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Cursor: .cursor/mcp.json

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

Smithery (Remote)

npx -y @smithery/cli install idea-reality-mcp --client claude

Setup & Configuration

First-Time Setup

idea-reality setup

Interactive wizard that:

  1. Shows terms acceptance
  2. Detects your platform (Claude Desktop, Cursor, etc.)
  3. Generates config snippet
  4. Runs health check

Platform-Specific Config

idea-reality config              # interactive menu
idea-reality config claude_code  # auto-installs
idea-reality config cursor       # prints Cursor config
idea-reality config raw_json     # generic MCP JSON

Health Check

idea-reality doctor        # core checks (~2s)
idea-reality doctor --full # + API validation, all 6 sources

Optional Environment Variables

export GITHUB_TOKEN=ghp_...        # Higher GitHub API rate limits
export PRODUCTHUNT_TOKEN=...       # Enable Product Hunt (deep mode)

Core MCP Tool: idea_check

Tool Schema

Parameters:

  • idea_text (string, required): Natural-language description of your idea
  • depth (string, optional): "quick" (default) or "deep"

Modes:

  • quick: GitHub + Hacker News (< 3 seconds)
  • deep: All 6 sources (GitHub, HN, npm, PyPI, Product Hunt, Stack Overflow)

Using from AI Agent

When a user says "check if this idea already exists", use the idea_check tool:

{
  "tool": "idea_check",
  "arguments": {
    "idea_text": "a CLI tool that converts Figma designs to React components",
    "depth": "deep"
  }
}

Response Structure

{
  "reality_signal": 72,
  "duplicate_likelihood": "high",
  "trend": "accelerating",
  "sub_scores": {
    "market_momentum": 73
  },
  "evidence": [
    {
      "source": "github",
      "type": "repo_count",
      "query": "figma react converter CLI",
      "count": 342
    },
    {
      "source": "github",
      "type": "max_stars",
      "query": "figma react converter",
      "count": 15000
    },
    {
      "source": "hackernews",
      "type": "mention_count",
      "query": "figma react",
      "count": 18
    },
    {
      "source": "npm",
      "type": "package_count",
      "query": "figma-react",
      "count": 56
    },
    {
      "source": "pypi",
      "type": "package_count",
      "query": "figma-react",
      "count": 23
    },
    {
      "source": "producthunt",
      "type": "product_count",
      "query": "figma react",
      "count": 8
    },
    {
      "source": "stackoverflow",
      "type": "question_count",
      "query": "figma react converter",
      "count": 120
    }
  ],
  "top_similars": [
    {
      "name": "figma-to-react/figma-to-react",
      "url": "https://github.com/figma-to-react/figma-to-react",
      "stars": 15000,
      "description": "Convert Figma designs to React components"
    }
  ],
  "pivot_hints": [
    "High competition. Consider a niche differentiator like supporting design tokens or specific component libraries.",
    "The leading project may have gaps in TypeScript support or accessibility features."
  ]
}

REST API (No MCP Required)

Python Client

import httpx

response = httpx.post(
    "https://idea-reality-mcp.onrender.com/api/check",
    json={
        "idea_text": "AI code review tool",
        "depth": "deep"
    },
    timeout=30.0
)

result = response.json()
print(f"Reality Score: {result['reality_signal']}/100")
print(f"Trend: {result['trend']}")
print(f"Top Competitor: {result['top_similars'][0]['name']} ({result['top_similars'][0]['stars']} ⭐)")

cURL

curl -X POST https://idea-reality-mcp.onrender.com/api/check \
  -H "Content-Type: application/json" \
  -d '{
    "idea_text": "a markdown-based static site generator with live reload",
    "depth": "quick"
  }'

Common Patterns

Pre-Build Validation

Before starting a new project:

# User says: "I want to build a CLI tool for GitHub issue management"
# Agent calls:
{
  "tool": "idea_check",
  "arguments": {
    "idea_text": "CLI tool for GitHub issue management with labels and milestones",
    "depth": "deep"
  }
}

# If reality_signal > 80:
#   → Suggest niche differentiation or pivot
# If reality_signal 40-80:
#   → Validate unique features, check top_similars
# If reality_signal < 40:
#   → Green light, low competition

Feature Validation

Check if a feature is already widely implemented:

{
  "tool": "idea_check",
  "arguments": {
    "idea_text": "add real-time collaborative editing to my markdown editor",
    "depth": "quick"
  }
}

Market Trend Analysis

Understand if a space is growing or declining:

{
  "tool": "idea_check",
  "arguments": {
    "idea_text": "browser automation library using Chrome DevTools Protocol",
    "depth": "deep"
  }
}

Check the trend field:

  • "accelerating" → Growing market, act fast
  • "stable" → Mature market, differentiation critical
  • "declining" → Consider pivoting

Auto-Trigger in Agent Instructions

Add to .cursorrules, CLAUDE.md, or .github/copilot-instructions.md:

When starting a new project, use the idea_check MCP tool to check if similar projects already exist.

CI/CD Integration

GitHub Action for Pull Requests

Create .github/workflows/idea-check.yml:

name: Idea Reality Check
on:
  issues:
    types: [opened]

jobs:
  check:
    if: contains(github.event.issue.labels.*.name, 'proposal')
    runs-on: ubuntu-latest
    steps:
      - uses: mnemox-ai/idea-check-action@v1
        with:
          idea: ${{ github.event.issue.title }}
          github-token: ${{ secrets.GITHUB_TOKEN }}

This auto-validates feature proposals labeled proposal.

Interpreting Results

Reality Signal (0–100)

  • 0–30: Low competition, potentially novel idea
  • 31–60: Moderate competition, validate unique angle
  • 61–85: High competition, niche differentiation required
  • 86–100: Saturated market, strong pivot recommended

Duplicate Likelihood

  • low: Few similar projects found
  • medium: Several similar projects exist
  • high: Many similar projects, established category
  • very_high: Extremely crowded space

Market Momentum (sub_scores)

Measures recent growth in the space:

  • < 40: Declining interest
  • 40–60: Stable
  • > 60: Growing/accelerating

Scoring Weights

Quick Mode

SourceWeight
GitHub repos60%
GitHub stars20%
Hacker News20%

Deep Mode

SourceWeight
GitHub repos22%
GitHub stars9%
Hacker News14%
npm18%
PyPI13%
Product Hunt14%
Stack Overflow10%

If a source fails, weights redistribute automatically.

Troubleshooting

"MCP server not found"

Cursor/Claude Desktop:

  1. Restart the application completely
  2. Check config file location and syntax
  3. Run idea-reality doctor

Claude Code:

claude mcp list  # verify installation
claude mcp remove idea-reality
claude mcp add idea-reality -- uvx idea-reality-mcp

"GitHub API rate limit exceeded"

Set a GitHub token:

export GITHUB_TOKEN=ghp_your_token_here
idea-reality doctor --full  # verify

Generate token at: https://github.com/settings/tokens (no scopes needed for public data)

"Product Hunt data missing (deep mode)"

Product Hunt requires authentication:

export PRODUCTHUNT_TOKEN=your_token_here

Or use depth: "quick" which skips Product Hunt.

"Irrelevant results"

The tool uses 3-stage keyword extraction. If results are off:

  1. Be more specific in idea_text:

    • ❌ "productivity app"
    • ✅ "CLI time-tracking tool for developers with Git integration"
  2. Report the issue: https://github.com/mnemox-ai/idea-reality-mcp/issues/new?template=inaccurate-result.yml

"Tool call timeout"

Deep mode can take 10–15 seconds. Increase timeout:

# Python httpx client
httpx.post(..., timeout=30.0)

Or use depth: "quick" (< 3 seconds).

Example Workflows

1. Pre-Project Kickoff

User: "I want to build a Rust-based SQL formatter with auto-fix"

Agent:
1. Call idea_check with depth="deep"
2. If reality_signal > 70:
   - Show top competitors (e.g. sqlformat, prettier-plugin-sql)
   - Suggest niches from pivot_hints (e.g. Rust performance angle)
3. If reality_signal < 50:
   - Proceed with project scaffolding

2. Feature Gap Analysis

User: "Should I add Vim keybindings to my editor?"

Agent:
1. Call idea_check: "text editor with vim keybindings"
2. Check evidence[].count for npm/PyPI packages
3. High count → already solved, suggest integration
4. Low count → potential differentiator

3. Market Validation

User: "Is AI code review still worth building?"

Agent:
1. Call idea_check with depth="deep"
2. Check trend field
3. If "accelerating" → market growing, move fast
4. If "declining" → suggest pivot to specific niche

Advanced Usage

Batch Validation

import httpx

ideas = [
    "AI-powered commit message generator",
    "Real-time Markdown collaboration",
    "GitHub issue templates manager"
]

async with httpx.AsyncClient() as client:
    tasks = [
        client.post(
            "https://idea-reality-mcp.onrender.com/api/check",
            json={"idea_text": idea, "depth": "quick"}
        )
        for idea in ideas
    ]
    results = await asyncio.gather(*tasks)

for idea, resp in zip(ideas, results):
    data = resp.json()
    print(f"{idea}: {data['reality_signal']}/100 ({data['trend']})")

Custom Analysis

def should_build(result: dict) -> str:
    score = result['reality_signal']
    trend = result['trend']
    
    if score < 40:
        return "BUILD: Low competition, novel idea"
    elif score < 70 and trend == "accelerating":
        return "BUILD WITH NICHE: Growing market, differentiate"
    elif score > 85:
        return "PIVOT: Saturated market"
    else:
        return "RESEARCH: Validate unique angle"

response = httpx.post(...)
verdict = should_build(response.json())

Best Practices

  1. Always use idea_check before scaffolding — prevents wasted effort
  2. Start with depth="quick" — faster iteration, upgrade to deep if uncertain
  3. Read pivot_hints — often contains actionable niche suggestions
  4. Check top_similars — study competitors before building
  5. Monitor trend — timing matters as much as uniqueness
  6. Set GITHUB_TOKEN — avoids rate limits on larger scans

Resources

发现
标签

此技能尚未发布标签。

版本
最新版本元数据

版本

v2026.09.25

发布时间

2026年9月25日

分类

未分类

许可证

NOASSERTION

源路径

skills/idea-reality-mcp-validation

默认分支

main

最新提交

329e67c

Tree SHA

01fd22f