
document-project
by resolve-io
An AI-assisted software engineering methodology that refracts development complexity into clear, actionable principles through coordinated AI agents. Combines proven methodologies (PSP/TSP, TDD/XP, Clean Code, DDD) into a unified framework for predictable, resilient, intentional, sustainable, and maintainable software development.
SKILL.md
name: document-project description: Use to analyze and document any project codebase. Creates comprehensive reference documentation for AI-assisted development including architecture and patterns. version: 1.0.0
Document Project Task
When to Use
- Project Documentation: Document any codebase for AI understanding
- Architecture Assessment: Analyze current system architecture
- Knowledge Transfer: Create comprehensive project documentation
- PRD Preparation: Generate context needed for PRD creation
- Onboarding: Build complete project reference for new team members
What This Skill Does
Guides you through building comprehensive project documentation:
- Configuration: Load project settings from
.prism/core-config.yaml - Classification: Detect project type, structure, and technology stack
- Scanning: Analyze codebase (quick, deep, or exhaustive)
- Documentation: Generate architecture, API, and development guides
- Indexing: Create master index for AI retrieval
Quick Start
- Run the skill and specify project root directory
- Choose scan depth (quick/deep/exhaustive)
- Confirm project classification
- Wait for documentation generation
- Review generated files at
{{output_folder}}/index.md
Output Location: docs/project/ (or as configured)
Workflow Overview
The workflow follows 10 sequential steps. For detailed instructions, see Workflow Steps Reference.
| Step | Name | Purpose |
|---|---|---|
| 0 | Load Configuration | Read .prism/core-config.yaml |
| 1 | Check Existing | Detect existing docs, choose mode |
| 2 | Select Scan Level | Quick/Deep/Exhaustive |
| 3 | Initialize State | Create tracking file |
| 4 | Detect Structure | Classify project type |
| 5 | Scan Project | Execute selected scan depth |
| 6 | Deep-Dive Mode | (If selected) Detailed area analysis |
| 6.5 | Check Existing Docs | Avoid duplicates via Smart Connections |
| 7 | Generate Index | Create master entry point |
| 8-10 | Validate & Complete | Verify, fill gaps, finalize |
Scan Levels
| Level | Time | What It Does |
|---|---|---|
| Quick | 2-5 min | Pattern-based analysis, configs only |
| Deep | 10-30 min | Critical directories, selective file reading |
| Exhaustive | 30-120 min | ALL source files (excludes node_modules) |
Project Types Detected
- claude-code-plugin: PRISM-style plugins with skills/commands/tasks
- web: React, Vue, Angular applications
- backend: Express, FastAPI, Rails servers
- mobile: React Native, Flutter apps
- cli/library/desktop: Various other project types
Generated Documentation
Core files generated (as applicable):
index.md- Master entry pointproject-overview.md- Executive summaryarchitecture.md- Technical architecturetechnology-stack.md- Complete tech stackdevelopment-guide.md- Setup and workflowapi-contracts.md- API endpointsdata-models.md- Database schemasource-tree-analysis.md- Directory structure
Key Principles
- Write-as-you-go: Write files immediately, don't accumulate
- Resumability: State file tracks progress for resume
- Batching: Process subfolders one at a time for token management
- No duplicates: Check existing docs before creating
Reference Documentation
- Workflow Steps - Detailed step-by-step instructions
Guardrails
- Always load config first (Step 0)
- Update state file after EVERY step completion
- Write documents immediately after generating content
- Purge detailed data from context after writing
- First-time startup may require explicit scan level selection
スコア
総合スコア
リポジトリの品質指標に基づく評価
SKILL.mdファイルが含まれている
ライセンスが設定されている
100文字以上の説明がある
GitHub Stars 100以上
3ヶ月以内に更新がある
10回以上フォークされている
オープンIssueが50未満
プログラミング言語が設定されている
1つ以上のタグが設定されている
レビュー
レビュー機能は近日公開予定です