platonic-coding

by caesar0301

Intelligent orchestrator for Platonic Coding workflow. Auto-detects project state and runs appropriate phases—init for new projects, recover specs from existing code, refine RFCs, implement from specs with guides and tests, or review code compliance. Single entry point for the complete specification-driven development lifecycle.

3.8k编码与调试未扫描2026年3月30日

安装

claude skill add --url https://github.com/openclaw/skills

文档

Platonic Coding

Intelligent orchestrator for the complete specification-driven development lifecycle. Auto-detects project state and executes the appropriate workflow phases—initialization, specification management, implementation, or review.

When to Use This Skill

Use this skill when you need to:

  • Bootstrap a new project with Platonic Coding infrastructure
  • Adopt Platonic Coding for an existing codebase (recover specs from code)
  • Manage RFC specifications (validate, refine, generate indices)
  • Implement features from RFC specs with guides and tests
  • Review code for spec compliance
  • Run the full workflow from design → RFC → code → review

Keywords: platonic coding, specs, RFC, implementation, review, workflow, spec-driven, initialization

Intelligent Auto-Detection

When invoked without specific instructions, this skill automatically detects project state and suggests the next action.

Detection Algorithm

code
1. Does .platonic.yml exist?
   → NO:  Run INIT mode
          • Has source code? → init-recover (generate specs from code)
          • No code? → init-scaffold (create infrastructure)

2. Has specs directory but no RFCs?
   → Has design drafts? → WORKFLOW Phase 1 (generate RFCs)
   → No drafts? → WORKFLOW Phase 0 (conceptual design)

3. Has RFCs but no implementation guides?
   → WORKFLOW Phase 2 (implement specs)

4. Has both specs and code?
   → REVIEW mode (check compliance)

5. In middle of workflow?
   → Resume from current phase

User Override

Users can explicitly specify operations to bypass auto-detection:

  • --init or init: Run initialization/scaffold
  • --init-recover: Run recovery (scan existing code)
  • --specs <operation>: Run specs management operation
  • --impl <operation>: Run implementation operation
  • --review: Run spec compliance review
  • --workflow: Run full 4-phase workflow
  • --phase <N>: Start workflow at specific phase

Core Workflow Phases

The Platonic Coding workflow follows four phases:

PhaseFocusOutputMode
0Conceptual DesignDesign draft (docs/drafts/)WORKFLOW
1RFC SpecificationRFCs (docs/specs/)WORKFLOW / SPECS
2ImplementationGuide + Code + TestsWORKFLOW / IMPL
3Spec Compliance ReviewReview reportWORKFLOW / REVIEW

Operation Modes

INIT Mode

Purpose: Bootstrap Platonic Coding infrastructure

Operations (see references/INIT/):

  • init-scaffold: Create .platonic.yml, directories, templates
  • init-scan: Analyze existing codebase (recovery mode)
  • init-plan-modular-specs: Propose RFC dependency graph
  • init-recover-conceptual: Generate conceptual design RFC
  • init-recover-architecture: Generate architecture design RFCs
  • init-recover-impl-interface: Generate impl interface RFCs (optional)

Auto-detection: Runs when .platonic.yml is missing

Examples:

code
# Auto-detect: new vs existing project
Use platonic-coding to set up my new project.

# Explicit: existing codebase recovery
Use platonic-coding init-recover to scan this codebase and recover specs.

# Explicit: greenfield scaffold
Use platonic-coding init-scaffold for project "Acme" (TypeScript/Next.js).

Reference: See references/REFERENCE.md → INITIALIZATION section


SPECS Mode

Purpose: Manage RFC specifications

Operations (see references/SPECS/):

  • specs-refine: Comprehensive validation and update
  • specs-generate-history: Update RFC change history
  • specs-generate-index: Update RFC index
  • specs-generate-namings: Update terminology reference
  • specs-validate-consistency: Check cross-references and metadata
  • specs-check-taxonomy: Verify terminology usage
  • specs-check-compliance: Validate against RFC standard

Auto-detection: Suggested when specs exist but need validation

Examples:

code
# Auto-detect: refine all specs
Use platonic-coding to validate and update all specifications.

# Explicit: specific operation
Use platonic-coding specs-generate-index to update the RFC index.

# Explicit: comprehensive refinement
Use platonic-coding specs-refine to run all validation and generation operations.

Reference: See references/REFERENCE.md → SPECIFICATION section


IMPL Mode

Purpose: Translate RFCs into implementation guides and code

Operations (see references/IMPL/):

  • impl-full: End-to-end: spec → guide → plan → code + tests (default)
  • impl-create-guide: Generate implementation guide from RFC
  • impl-code: Implement code from existing guide
  • impl-validate-guide: Check guide against RFC for contradictions
  • impl-update-guide: Update guide when RFC changes

Auto-detection: Suggested when RFCs exist but no implementation guides

Confirmation Gates: By default, pauses after impl guide and coding plan for user confirmation. Can be overridden with "no confirmations" or auto-mode.

Examples:

code
# Auto-detect: implement from RFC
Use platonic-coding to implement RFC-0042 (Message Queue) in the acme-queue module.

# Explicit: create guide only
Use platonic-coding impl-create-guide for RFC-0001, guide only, no coding.

# Explicit: implement from existing guide
Use platonic-coding impl-code from docs/impl/RFC-0001-impl.md.

# Auto-mode: no confirmations
Use platonic-coding impl-full for RFC-003 without stopping for confirmation.

Reference: See references/REFERENCE.md → IMPLEMENTATION section


REVIEW Mode

Purpose: Validate code implementation against specifications

Default Behavior: Generates compliance report WITHOUT modifying code

Review Process:

  1. Understand specifications (RFCs, impl guides)
  2. Generate functionality checklist
  3. Map specs to code locations
  4. Review implementation for each item
  5. Identify discrepancies (missing, inconsistent, partial, extra)
  6. Generate prioritized report with recommendations

Auto-detection: Suggested when both specs and code exist

Examples:

code
# Auto-detect: review specific RFC implementation
Use platonic-coding to review src/auth/ against RFC-0001.md.

# Explicit: comprehensive review
Use platonic-coding review to audit all code against all RFCs in docs/specs/.

# Explicit: gap analysis
Use platonic-coding review to identify gaps between specs/ and src/.

Reference: See references/REFERENCE.md → REVIEW section


WORKFLOW Mode

Purpose: Orchestrate the full 4-phase workflow from design to review

Phases:

  • Phase 0: Conceptual design (interactive chat, design draft)
  • Phase 1: Generate RFC from draft → call specs-refine
  • Phase 2: Call impl-full (guide → plan → code + tests)
  • Phase 3: Call review for spec compliance
  • FINISHED: Summary and recommendations

Auto-detection: Suggested when project is initialized and ready for new features

Phase Visibility: Always shows current phase at each step

Examples:

code
# Run full workflow from Phase 0
Use platonic-coding workflow to implement a user preferences feature.

# Start at specific phase
Use platonic-coding workflow --phase 2 to implement RFC-0042.

# Resume workflow (auto-detected)
Use platonic-coding to continue from where we left off.

Reference: See references/REFERENCE.md → WORKFLOW section

Default Paths

ArtifactDefault PathConfigurable in .platonic.yml
Design draftsdocs/drafts/Yes
RFC specsdocs/specs/Yes
Implementation guidesdocs/impl/Yes

Templates

All templates are provided in the assets/ directory:

  • Project scaffolding: assets/templates/
  • RFC templates: assets/specs/
  • Implementation templates: assets/implementation/
  • Review templates: assets/review/

Templates use {{PLACEHOLDER}} syntax. See individual reference files for details.

Best Practices

  1. Trust auto-detection: Let the skill suggest next steps based on project state
  2. Override when needed: Use explicit mode flags for specific operations
  3. Review generated artifacts: All generated RFCs and guides are Draft—review before use
  4. Run refine regularly: Keep specs validated and indices updated
  5. Use confirmation gates: Default behavior pauses for review—don't skip unless confident
  6. Report-only by default: Review mode generates reports, modify code only when explicitly requested

Dependencies

  • Read/write access to project directories
  • Read access to .platonic.yml for configuration
  • Understanding of target language and framework
  • Ability to scan and read source code files (for recovery and review)

Integration Example

Complete workflow from greenfield to reviewed implementation:

code
# Day 1: Initialize
Use platonic-coding to set up my new project "Acme" (TypeScript/Next.js).

# Day 2: Start workflow (Phase 0)
Use platonic-coding workflow to design a user authentication feature.

# Day 2: Continue workflow (Phase 1-3)
# Agent auto-runs: Phase 1 (RFC) → Phase 2 (impl) → Phase 3 (review)

# Day 3: Maintenance
Use platonic-coding to refine all specs and validate consistency.

# Day 4: New feature
Use platonic-coding workflow --phase 0 to add a notification system.

See references/REFERENCE.md for detailed operation guides.

相关 Skills

前端设计

by anthropics

Universal
热门

面向组件、页面、海报和 Web 应用开发,按鲜明视觉方向生成可直接落地的前端代码与高质感 UI,适合做 landing page、Dashboard 或美化现有界面,避开千篇一律的 AI 审美。

想把页面做得既能上线又有设计感,就用前端设计:组件到整站都能产出,难得的是能避开千篇一律的 AI 味。

编码与调试
未扫描111.1k

网页构建器

by anthropics

Universal
热门

面向复杂 claude.ai HTML artifact 开发,快速初始化 React + Tailwind CSS + shadcn/ui 项目并打包为单文件 HTML,适合需要状态管理、路由或多组件交互的页面。

在 claude.ai 里做复杂网页 Artifact 很省心,多组件、状态和路由都能顺手搭起来,React、Tailwind 与 shadcn/ui 组合效率高、成品也更精致。

编码与调试
未扫描111.1k

网页应用测试

by anthropics

Universal
热门

用 Playwright 为本地 Web 应用编写自动化测试,支持启动开发服务器、校验前端交互、排查 UI 异常、抓取截图与浏览器日志,适合调试动态页面和回归验证。

借助 Playwright 一站式验证本地 Web 应用前端功能,调 UI 时还能同步查看日志和截图,定位问题更快。

编码与调试
未扫描111.1k

相关 MCP 服务

GitHub

编辑精选

by GitHub

热门

GitHub 是 MCP 官方参考服务器,让 Claude 直接读写你的代码仓库和 Issues。

这个参考服务器解决了开发者想让 AI 安全访问 GitHub 数据的问题,适合需要自动化代码审查或 Issue 管理的团队。但注意它只是参考实现,生产环境得自己加固安全。

编码与调试
83.0k

by Context7

热门

Context7 是实时拉取最新文档和代码示例的智能助手,让你告别过时资料。

它能解决开发者查找文档时信息滞后的问题,特别适合快速上手新库或跟进更新。不过,依赖外部源可能导致偶尔的数据延迟,建议结合官方文档使用。

编码与调试
51.7k

by tldraw

热门

tldraw 是让 AI 助手直接在无限画布上绘图和协作的 MCP 服务器。

这解决了 AI 只能输出文本、无法视觉化协作的痛点——想象让 Claude 帮你画流程图或白板讨论。最适合需要快速原型设计或头脑风暴的开发者。不过,目前它只是个基础连接器,你得自己搭建画布应用才能发挥全部潜力。

编码与调试
46.2k

评论