菜单
Automated brownfield codebase analysis with weighted keyword scoring, three-stream analysis, and design pattern recognition. Detects project type, frameworks, dependencies, architecture patterns, and generates comprehensive project profile.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/oimiragieo/agent-studio --skill project-analyzer命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | project-analyzer |
| description | Automated brownfield codebase analysis with weighted keyword scoring, three-stream analysis, and design pattern recognition. Detects project type, frameworks, dependencies, architecture patterns, and generates comprehensive project profile. |
| version | 1.1.0 |
| model | sonnet |
| invoked_by | both |
| user_invocable | true |
| tools | ["Read","Glob","Grep","Bash"] |
| best_practices | ["Detect project root from package managers and manifest files","Identify frameworks from dependencies and directory structure","Generate comprehensive file statistics and language breakdown","Map component relationships and architecture patterns","Validate output against project-analysis.schema.json","Execute in < 30 seconds for typical projects (< 10k files)"] |
| error_handling | graceful |
| streaming | supported |
| executable | .claude/tools/analysis/project-analyzer/analyzer.mjs |
| test_suite | null |
| output_schema | .claude/schemas/project-analysis.schema.json |
| verified | false |
| lastVerifiedAt | "2026-02-19T05:29:09.098Z" |
| source | builtin |
| trust_score | 100 |
| provenance_sha | 641607b45a34f39a |
References (archive): SCAFFOLD_SKILLS_ARCHIVE_MAP.md — ProjectAnalyzer monorepo/service detection from Auto-Claude-develop analysis/analyzers.
Project Analyzer - Automated brownfield codebase analysis for rapid project onboarding and understanding. - Detecting project type (frontend, backend, fullstack, library, cli, mobile, monorepo) - Identifying frameworks and libraries from manifests and structure - Generating file statistics and language breakdown - Mapping component relationships and module structure - Detecting architecture patterns (MVC, layered, microservices, etc.) - Analyzing dependency health and outdated packages - Identifying code quality indicators (linting, testing, type safety) - Detecting technical debt and anti-patterns - Generating prioritized improvement recommendationsLocate project root by finding manifest files:
Search for package manager files:
package.json (Node.js/JavaScript/TypeScript)requirements.txt, pyproject.toml, setup.py (Python)go.mod (Go)Cargo.toml (Rust)pom.xml, build.gradle (Java/Maven/Gradle)composer.json (PHP)Identify project root:
Validate project root:
.git directoryClassify project based on manifest files and directory structure:
Frontend Projects:
src/components/, public/, assets/Backend Projects:
routes/, , , controllers/models/api/Fullstack Projects:
Library/Package Projects:
index.ts, lib/, dist/, build/library field in package.jsonCLI Projects:
bin field in package.jsonMobile Projects:
android/, ios/, mobile/Monorepo Projects:
workspaces in package.json, pnpm-workspace.yamlMicroservices Projects:
docker-compose.yml, service configsIdentify frameworks from manifest files and imports:
Read package.json dependencies (Node.js):
dependencies and devDependenciesRead requirements.txt (Python):
Analyze imports (optional deep scan):
Framework Categories:
Confidence Scoring:
Generate quantitative project statistics:
Count files by type:
node_modules/, .git/, dist/, build/Count lines of code:
Identify largest files:
Calculate averages:
Language Detection:
.ts, .tsx → TypeScript.js, .jsx → JavaScript.py → Python.go → Go.rs → Rust.java → Java.md → Markdown.json → JSON.yaml, .yml → YAMLAnalyze project structure and architecture:
Identify root directories:
src/, app/, lib/test/, __tests__/, cypress/config/, .config/docs/, documentation/dist/, build/, out/scripts/, bin/assets/, static/, public/Detect entry points:
index.ts, main.py, app.pyapp.ts, server.ts, app/page.tsxhandler.ts, lambda.tscli.ts, bin/Detect architecture pattern:
models/, views/, controllers/presentation/, business/, data/domain/, application/, infrastructure/Detect module system:
package.json for "type": "module" (ESM)import/export (ESM) vs require (CommonJS)Analyze dependency health:
Count dependencies:
Check for outdated packages (optional):
npm outdated or equivalentSecurity scan (optional):
npm audit or equivalentDetect code quality tooling:
Linting Configuration:
.eslintrc.json, eslint.config.js, ruff.tomlFormatting Configuration:
.prettierrc, pyproject.toml (Black/Ruff)Testing Framework:
Type Safety:
tsconfig.json"strict": trueIdentify common patterns and anti-patterns:
Good Practices:
Anti-Patterns:
Neutral Patterns:
Calculate technical debt score:
Debt Indicators:
Debt Score (0-100):
Remediation Effort:
Create prioritized improvement recommendations:
Categorize Recommendations:
Prioritize by Impact:
Estimate Effort and Impact:
Validate analysis output against schema:
Schema Validation:
project-analysis.schema.jsonOutput Metadata:
</execution_process>
node_modules/, .git/, dist/Integration with Other Skills:
<best_practices>
# Analyze current project
node .claude/tools/analysis/project-analyzer/analyzer.mjs
# Analyze specific directory
node .claude/tools/analysis/project-analyzer/analyzer.mjs /path/to/project
# Output to file
node .claude/tools/analysis/project-analyzer/analyzer.mjs --output .claude/context/artifacts/project-analysis.json
Agent Invocation:
# Analyze current project
Analyze this project
# Generate comprehensive analysis
Perform full project analysis and save to artifacts
# Quick analysis (manifest only)
Quick project type detection
</usage_example>
<formatting_example>
Sample Output (.claude/context/artifacts/project-analysis.json):
{
"analysis_id": "analysis-llm-rules-20250115",
"project_type": "fullstack",
"analyzed_at": "2025-01-15T10:30:00.000Z",
"project_root": "C:\\dev\\projects\\LLM-RULES",
"stats": {
"total_files": 1243,
"total_lines": 125430,
"languages": {
"JavaScript": 45230,
"TypeScript": 38120,
"Markdown": 25680,
"JSON": 12400,
"YAML": 4000
},
"file_types": {
".js": 234,
".mjs": 156,
".ts": 89,
".md": 312,
".json": 145
},
"directories": 87,
"avg_file_size_lines": 101,
"largest_files": [
{
"path": ".claude/tools/enforcement-gate.mjs",
"lines": 1520
}
]
},
"frameworks": [
{
"name": "nextjs",
"version": "14.0.0",
"category": "framework",
"confidence": 1.0,
"source": "package.json"
},
{
"name": "react",
"version": "18.2.0",
"category": "framework",
"confidence": 1.0,
"source": "package.json"
}
],
"structure": {
"root_directories": [
{
"name": ".claude",
"purpose": "config",
"file_count": 543
},
{
"name": "conductor-main",
"purpose": "source",
"file_count": 234
}
],
"entry_points": [
{
"path": "conductor-main/src/index.ts",
"type": "main"
}
],
"architecture_pattern": "modular",
"module_system": "esm"
},
"dependencies": {
"production": 45,
"development": 23
},
"code_quality": {
"linting": {
"configured": true,
"tool": "eslint"
},
"formatting": {
"configured": true,
"tool": "prettier"
},
"testing": {
"framework": "vitest",
"test_files": 89,
"coverage_configured": true
},
"type_safety": {
"typescript": true,
"strict_mode": true
}
},
"tech_debt": {
"score": 35,
"indicators": [
{
"category": "complexity",
"severity": "medium",
"description": "3 files exceed 1000 lines",
"remediation_effort": "moderate"
}
]
},
"recommendations": [
{
"priority": "P1",
"category": "maintainability",
"title": "Refactor large files",
"description": "Break down files > 1000 lines into smaller modules",
"effort": "moderate",
"impact": "high"
}
],
"metadata": {
"analyzer_version": "1.0.0",
"analysis_duration_ms": 2340,
"files_analyzed": 1243,
"files_skipped": 3420,
"errors": []
}
}
</formatting_example>
When classifying files, directories, or components into categories, use weighted keyword scoring instead of simple string matching to prevent false positives:
| Signal Source | Score Weight | Example |
|---|---|---|
| File path/URL | 3 points | /api/routes/ matches "API" category |
| File/class name | 2 points | AuthService.ts matches "Authentication" |
| File content/imports | 1 point | import express matches "Backend" |
Threshold: Require 2+ total points before assigning a category. Falls back to "other" if no category scores above threshold. This prevents weak single-signal matches from misclassifying components.
Category keywords (extend per project type):
For comprehensive project understanding, analyze three parallel streams:
Stream 1 — Code Analysis: AST patterns, framework detection, dependency graph, architecture classification. This is the existing core workflow (Steps 1-11).
Stream 2 — Documentation: README quality, API docs existence, inline doc coverage, changelog maintenance, contribution guides. Score: docFiles / totalFiles weighted by type.
Stream 3 — Community/Operations: Git activity (commit frequency, contributor count), CI/CD configuration, issue templates, PR templates, release workflow, Docker/container setup.
Combine all three streams into the output JSON under analysis.streams:
{
"streams": {
"code": { "score": 0.85, "findings": [...] },
"documentation": { "score": 0.60, "findings": [...] },
"operations": { "score": 0.75, "findings": [...] }
},
"compositeHealth": 0.73
}
Detect common design patterns with confidence scoring:
| Pattern | Detection Signal | Confidence Threshold |
|---|---|---|
| Singleton | Private constructor + static instance | 0.80 |
| Factory | create* methods returning interface types | 0.70 |
| Observer | subscribe/on/emit/addEventListener | 0.70 |
| Strategy | Interface + multiple implementations | 0.60 |
| Decorator | Wrapper classes with same interface | 0.60 |
| Repository | Data access layer abstraction | 0.70 |
| Middleware | Chain-of-responsibility in request pipeline | 0.70 |
Output detected patterns in the analysis JSON with location, confidence, and evidence:
{
"patterns": [
{
"type": "Factory",
"category": "Creational",
"confidence": 0.85,
"location": "src/services/UserFactory.ts",
"evidence": ["createUser method", "returns IUser interface"]
}
]
}
For additional detection patterns extracted from the Auto-Claude analysis framework, see:
references/auto-claude-patterns.md - Monorepo indicators, SERVICE_INDICATORS, SERVICE_ROOT_FILES, infrastructure detection, convention detectionreferences/service-patterns.md - Service type detection (frontend, backend, library), framework-specific patterns, entry point detectionreferences/database-patterns.md - Database configuration file patterns, ORM detection (Prisma, SQLAlchemy, TypeORM, Drizzle, Mongoose), connection string patternsreferences/route-patterns.md - Express, FastAPI, Flask, Django, Next.js, Go, Rust API route detection patternsThese references provide comprehensive regex patterns and detection logic for brownfield codebase analysis.
Before starting:
Read .claude/context/memory/learnings.md
After completing:
.claude/context/memory/learnings.md.claude/context/memory/issues.md.claude/context/memory/decisions.mdASSUME INTERRUPTION: If it's not in memory, it didn't happen.