[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-78035":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":9,"languages":9,"totalLinesOfCode":9,"stars":10,"forks":11,"watchers":12,"openIssues":11,"contributorsCount":11,"subscribersCount":11,"size":11,"stars1d":11,"stars7d":11,"stars30d":11,"stars90d":11,"forks30d":11,"starsTrendScore":11,"compositeScore":11,"rankGlobal":9,"rankLanguage":9,"license":13,"archived":14,"fork":14,"defaultBranch":15,"hasWiki":16,"hasPages":14,"topics":17,"createdAt":9,"pushedAt":9,"updatedAt":18,"readmeContent":19,"aiSummary":20,"trendingCount":11,"starSnapshotCount":11,"syncStatus":21,"lastSyncTime":22,"discoverSource":23},78035,"oop-architect","ZhongliangGuo\u002Foop-architect","ZhongliangGuo","A Claude Code skill for OOP architecture planning and tracking — live Mermaid UML diagrams in CLAUDE.md that update as you code.",null,101,0,104,"MIT License",false,"main",true,[],"2026-06-12 02:03:45","# OOP Architect\n\nEnglish | [中文](README.zh.md)\n\nA [Claude Code](https:\u002F\u002Fclaude.ai\u002Fcode) skill that designs, documents, and tracks object-oriented software architecture using live Mermaid UML diagrams inside `CLAUDE.md`.\n\n## What It Does\n\nWhen you start or return to a project, you need to instantly understand: what's built, what's in progress, how things connect, and what constraints exist. This skill turns `CLAUDE.md` into a self-maintaining architecture blueprint — updated automatically as you code.\n\n### Four Modes\n\n| Situation | What happens |\n|---|---|\n| New project, no code yet | Design class hierarchy from scratch, generate `CLAUDE.md` |\n| Existing code, no `CLAUDE.md` | Scan code and reverse-generate architecture documentation |\n| Existing project, adding a feature | Design new components on top of the existing architecture |\n| Code has drifted from `CLAUDE.md` | Full resync — scan code and update all diagrams |\n\n### What Gets Generated\n\n- **Class diagrams** with full typed signatures, visibility markers (`+` `-` `#`), and per-class implementation status (✅ complete \u002F 🔶 partial \u002F 🔲 not started)\n- **Sequence diagrams** for key workflows (auth, request lifecycle, etc.)\n- **Component diagrams** for multi-module projects\n- **State diagrams** for entities with explicit lifecycle or status transitions\n- **ER diagrams** for data-heavy projects with ORM or non-trivial data models\n- **Progress table**, directory structure, design decisions, and constraint notes\n- **Embedded maintenance rules** so `CLAUDE.md` stays current on its own — no skill activation needed for routine updates\n\n### Language Support\n\nPython · Java · C# · Kotlin · TypeScript · JavaScript · Go · Rust · C++ · Swift · Ruby · PHP · Generic OOP fallback\n\n---\n\n## Install\n\n**Via Claude Code marketplace** (recommended):\n```bash\nclaude plugin marketplace add ZhongliangGuo\u002Foop-architect\n```\n\n**Per-project** (committed to the repo):\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FZhongliangGuo\u002Foop-architect .claude\u002Fskills\u002Foop-architect\n```\n\n**Global** (available in all your projects):\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FZhongliangGuo\u002Foop-architect ~\u002F.claude\u002Fskills\u002Foop-architect\n```\n\nTo update:\n```bash\ncd .claude\u002Fskills\u002Foop-architect   # or ~\u002F.claude\u002Fskills\u002Foop-architect\ngit pull\n```\n\n---\n\n## Usage\n\nThe skill activates automatically when you say things like:\n\n- *\"Plan the architecture for this project\"*\n- *\"Design the classes for a task queue system\"*\n- *\"I want to add a new feature: background job retries\"*\n- *\"Resync the architecture\"* \u002F *\"Update CLAUDE.md from the code\"*\n- *\"Generate CLAUDE.md for this existing codebase\"*\n\nOr trigger manually: `\u002Foop-architect`\n\nAfter initial setup, `CLAUDE.md` maintains itself. Claude Code reads the embedded maintenance rules on every task and updates diagrams when public interfaces change — no skill activation needed.\n\n---\n\n## Background\n\nThis skill was born from a vibe coding problem: sessions get interrupted. You come back after hours or days and have no idea what's half-built, what depends on what, or what constraints you'd already figured out. Reading through code to reconstruct that mental model wastes time.\n\nThe solution is a `CLAUDE.md` that acts as a living blueprint — Mermaid UML diagrams that stay in sync with the code, showing exactly what's implemented (✅), partially done (🔶), or planned but not started (🔲).\n\n**The key design decision** was where to put the \"keep CLAUDE.md updated\" logic. The obvious answer is the skill — but that means the skill needs to be active on every task. The better answer: embed the maintenance rules directly inside the generated `CLAUDE.md`. Since Claude Code always loads `CLAUDE.md` into context at the start of every task, those rules get followed automatically. The skill only needs to run for initial planning and full resyncs.\n\n**Designed for vibe coders.** Many people using Claude Code for large projects understand OOP but not necessarily formal design patterns. This skill uses good architectural patterns internally, but always explains them in plain language — not \"this uses the Factory Pattern\" but \"adding a new database adapter means creating one new file; nothing else needs to change.\" Pattern names appear only as optional side notes for users who want to look them up.\n\n---\n\n## Design\n\n### Atomic reference loading\n\nMost skills load everything upfront. This skill loads only what the current task needs:\n\n- **Always loaded:** `SKILL.md` (~88 lines) + one language file (~25 lines)\n- **Phase-specific:** the matching phase reference file + `references\u002Fuml-class.md` when diagrams are written or updated\n- **Skipped unless needed:** sequence\u002Fcomponent\u002Fstate\u002FER diagram rules, CLAUDE.md template\n\nThis keeps context usage lean across all four modes and all languages.\n\n### Self-maintaining `CLAUDE.md`\n\nThe generated `CLAUDE.md` embeds its own update rules. Claude Code reads `CLAUDE.md` at the start of every task and follows those rules automatically. The skill only runs for initial planning and full resyncs.\n\n### When NOT to use this skill\n\nIf the project is a single-file script, a one-off data transformation, or purely functional with no shared state — the skill will say so and suggest a simpler approach rather than forcing a class hierarchy.\n\n---\n\n## File Structure\n\n```\noop-architect\u002F\n├── SKILL.md                       # Entry point — mode detection and orchestration\n└── references\u002F\n    ├── new-project.md             # Design from scratch (Phase 1)\n    ├── resync.md                  # Full resync procedure (Phase 2)\n    ├── generate-from-code.md      # Reverse-generate from existing code (Phase 3)\n    ├── extend-design.md           # Extend architecture with new feature (Phase 4)\n    ├── uml-class.md               # Class diagram rules (loaded when writing or updating diagrams)\n    ├── uml-sequence.md            # Sequence diagram rules (loaded when needed)\n    ├── uml-component.md           # Component diagram rules (loaded when needed)\n    ├── uml-state.md               # State diagram rules (loaded when needed)\n    ├── uml-er.md                  # ER diagram rules (loaded when needed)\n    ├── claude-md-template.md      # CLAUDE.md template (loaded when creating new)\n    └── langs\u002F\n        ├── python.md\n        ├── java-csharp-kotlin.md\n        ├── typescript-javascript.md\n        ├── go.md\n        ├── rust.md\n        ├── cpp.md\n        ├── swift.md\n        ├── ruby.md\n        ├── php.md\n        └── generic.md             # Fallback for unlisted languages\n```\n","OOP Architect 是一个用于面向对象软件架构设计、文档编写和跟踪的Claude Code技能，它能够在CLAUDE.md中生成并自动更新Mermaid UML图。该项目支持多种编程语言，包括Python、Java等，并能根据代码自动生成类图、序列图、组件图等多种UML图表，同时提供实现状态跟踪与项目结构说明等功能。适用于需要可视化管理复杂软件架构的开发场景，特别是在持续迭代开发过程中保持架构文档与实际代码同步的需求下特别有用。通过自动化的维护规则，即使在长时间中断后也能快速恢复对项目的理解。",2,"2026-06-11 03:56:23","CREATED_QUERY"]