AgentSkillsCN

Framework Detector

针对6大框架的多信号框架检测与置信度评分

SKILL.md
--- frontmatter
name: Framework Detector
version: 1.0.0
framework_versions:
  min: 1.0.0
  recommended: 1.0.0
compatible_agents:
  backend-developer: ">=3.0.0"
  frontend-developer: ">=3.0.0"
  tech-lead-orchestrator: ">=2.5.0"
description: Multi-signal framework detection with confidence scoring for 6 major frameworks
frameworks:
  - framework-detector
languages:
  - javascript
  - typescript
category: utility
updated: 2025-10-22

Framework Detector Skill

Quick Reference

When to Use: Automatically detect framework in project before loading framework-specific skills

Supported Frameworks: NestJS, React, Phoenix, Rails, .NET/ASP.NET Core, Blazor

Detection Method: Multi-signal analysis with weighted confidence scoring

Usage

Basic Detection

javascript
const FrameworkDetector = require('./detect-framework');

const detector = new FrameworkDetector('/path/to/project');
const result = await detector.detect();

console.log(result.primary);    // "nestjs"
console.log(result.confidence); // 0.92
console.log(result.alternates); // [{ framework: "dotnet", confidence: 0.45 }]

CLI Usage

bash
# Detect framework in current directory
./detect-framework.js

# Detect framework in specific project
./detect-framework.js /path/to/project

# Output format (JSON)
{
  "primary": "react",
  "confidence": 0.89,
  "alternates": [],
  "details": { ... }
}

Detection Signals

1. Package Manager (Weight: 10)

  • •Node.js: package.json dependencies
  • •Ruby: Gemfile gems
  • •Elixir: mix.exs dependencies
  • •.NET: *.csproj PackageReferences

2. Files (Weight: 8-9)

  • •Required: Framework-specific config files
  • •Optional: Common project structure files
  • •Wildcards: Pattern matching (e.g., *.csproj)

3. Imports (Weight: 7-8)

  • •Code pattern analysis in source files
  • •Regex-based matching
  • •File sampling for performance (max 20 files)

4. Boost Factors (Multiplier: 1.2-1.6x)

  • •Strong indicators multiply confidence
  • •Framework-specific patterns
  • •Examples:
    • •NestJS: nest-cli.json (+50%)
    • •React: JSX files (+40%)
    • •Blazor: .razor files (+60%)

Confidence Threshold

Default: 0.8 (80% confidence required)

Interpretation:

  • •≥ 0.9: Very high confidence
  • •0.8-0.9: High confidence
  • •0.6-0.8: Medium confidence (below threshold)
  • •< 0.6: Low confidence

Framework-Specific Patterns

NestJS Detection

code
✓ @nestjs/core in package.json
✓ nest-cli.json exists
✓ @Module decorator in .ts files
✓ @Controller decorator in .ts files

React Detection

code
✓ react in package.json
✓ .jsx or .tsx files exist
✓ useState or useEffect in code
✓ createRoot in code

Phoenix Detection

code
✓ {:phoenix, in mix.exs
✓ config/config.exs exists
✓ Phoenix.Endpoint in .ex files
✓ Phoenix.Router in .ex files

Rails Detection

code
✓ gem 'rails' in Gemfile
✓ config/application.rb exists
✓ Rails.application in code
✓ ActiveRecord::Base in code

.NET Detection

code
✓ Microsoft.AspNetCore in .csproj
✓ Program.cs exists
✓ [ApiController] in .cs files
✓ using Microsoft.AspNetCore

Blazor Detection

code
✓ Microsoft.AspNetCore.Components in .csproj
✓ .razor files exist
✓ @page directive in .razor files
✓ ComponentBase in code

Performance

Optimizations:

  • •File sampling (10-20 files max per check)
  • •Glob ignore patterns (node_modules, dist, build)
  • •Early exit on strong matches
  • •Async/await for parallel checks

Typical Detection Time: 100-500ms

Error Handling

No Frameworks Detected:

javascript
{
  primary: null,
  confidence: 0,
  alternates: [],
  details: {}
}

Multiple Frameworks (e.g., monorepo):

javascript
{
  primary: "react",
  confidence: 0.91,
  alternates: [
    { framework: "nestjs", confidence: 0.87 }
  ]
}

Integration with SkillLoader

javascript
const { SkillLoader } = require('../skill-loader');
const FrameworkDetector = require('./detect-framework');

async function loadFrameworkSkill(projectRoot) {
  // 1. Detect framework
  const detector = new FrameworkDetector(projectRoot);
  const result = await detector.detect();

  // 2. Handle low confidence
  if (result.confidence < 0.8) {
    // Prompt user or use alternates
    console.warn('Low confidence detection');
  }

  // 3. Load skill
  const loader = new SkillLoader({
    agentName: 'backend-developer',
    agentVersion: '3.0.0'
  });

  const skill = await loader.loadSkill(result.primary, 'quick');
  return skill;
}

Configuration

Edit framework-patterns.json to:

  • •Add new frameworks
  • •Adjust detection weights
  • •Modify confidence threshold
  • •Update boost factors

See Also