English | 中文

• What It Does
• Deep Discuss — Structured Deep Discussion
• S.U.P.E.R — The Architectural Backbone
• Platform Compatibility
• Installation
• Usage
• Project Structure
• Friendly Links
• License

A structured methodology for AI coding agents. Pure Markdown. Any platform. Architecture-first.

Spec-Driven Develop is a platform-agnostic AI agent plugin that ships two complementary skills:

• Spec-Driven Develop — automates the full pre-development pipeline for large-scale complex tasks: deep project analysis, phased task decomposition, document-driven progress tracking, and task-specific sub-SKILL generation — all before a single line of implementation code is written.
• Deep Discuss — a structured deep-discussion workflow for problem analysis, brainstorming, and solution design through disciplined multi-phase thinking.

No SDK. No runtime. No dependencies. Just Markdown files that any AI coding agent can read and execute.

When you tell your agent something like "rewrite this project in Rust" or "migrate to a microservice architecture", Spec-Driven Develop kicks in with a 7-phase preparation pipeline:

Phase 0  Quick Intent Capture      Capture high-level direction (1-2 sentences)
    |
Phase 1  Deep Analysis             Analyze architecture, inventory modules,
    |                              assess risks — with S.U.P.E.R health evaluation
    |
Phase 2  Intent Refinement         Ask targeted questions grounded in analysis,
    |                              confirm scope, priorities, and constraints
    |
Phase 3  Task Decomposition        Break work into phases, tasks, parallel lanes —
    |                              each task annotated with S.U.P.E.R design drivers
    |
Phase 4  Progress Tracking         Generate MASTER.md + per-phase detail files
    |                              for cross-conversation continuity
    |
Phase 5  Sub-SKILL Generation      Create a project-level SKILL with inlined
    |                              S.U.P.E.R principles and code review checklist
    |
Phase 6  Handoff                   Present all artifacts, confirm readiness
    |
   ...  Development Phases         Iterative implementation with progress tracking
    |
Phase 7  Archive                   Preserve all artifacts for traceability

A master progress file (docs/progress/MASTER.md) serves as the agent's memory anchor across conversations. No matter how many sessions a task spans, the agent always knows where things stand.

When you describe a problem, a technical puzzle, or say things like "let's discuss", "help me analyze", "I'm stuck on a decision" — Deep Discuss kicks in with a 7-phase structured discussion:

Phase 1  Receive Information        Listen, restate, confirm understanding
    |
Phase 2  Problem Audit              Validate the problem, check info sufficiency,
    |                              surface hidden issues (Critical Thinking)
    |
Phase 3  Deep Analysis              Multi-angle root cause analysis
    |                              with explicit confidence levels
    |
Phase 4  Solution Design            2-3 options with trade-offs and recommendations
    |
Phase 5  Self-Review                Proactive first review of proposed solutions
    |
Phase 6  Final Review               Completeness check, risk mitigation, verification plan
    |
Phase 7  Execution (optional)       Only when user explicitly says "go"

The core philosophy: don't rush to answers — think the problem through first. Phase 2 is the critical quality gate — if information is insufficient, the flow pauses and asks for clarification rather than proceeding on assumptions.

S.U.P.E.R is not a footnote — it is the design philosophy that drives every phase of the workflow and every line of code the agent produces.

Write code like building with LEGO — each brick has a single job, a standard interface, a clear direction, runs anywhere, and can be swapped at will.

S.U.P.E.R isn't just a reference document the agent might read — it's woven into the workflow at every level:

• Phase 1 — Analysis: Every module gets a per-principle compliance score (S🟢 U🟡 P🔴 E🟢 R🟡). The risk assessment includes a S.U.P.E.R Architecture Health Summary with violation hotspots.

• Phase 2 — Intent Refinement: Analysis findings are presented to the user so they can make informed decisions about scope and S.U.P.E.R priorities before task decomposition begins.

• Phase 3 — Planning: Each task is annotated with its S.U.P.E.R design drivers (which principles matter most for that task). Early phases prioritize fixing violation hotspots before building new features.

• Phase 5 — Sub-SKILL: The full S.U.P.E.R principles are inlined verbatim into the generated sub-SKILL (not just referenced), along with a mandatory 10-point code review checklist that the agent must pass after every task:

  Check	Principle
  Every new module/file has exactly one responsibility	S
  No function does more than one conceptual thing	S
  Data flows input → processing → output, no reverse deps	U
  No circular imports introduced	U
  Cross-module interfaces are schema-defined	P
  Module I/O is serializable	P
  No hardcoded paths, URLs, keys, or config values	E
  All new dependencies explicitly declared	E
  New modules can be replaced without changes to others	R
  All tests pass after the change	—

  All pass = proceed. 1-2 fail = fix before marking complete. 3+ fail = stop and refactor.

The SKILL prompt is written in a generic, platform-neutral way. It gracefully degrades on platforms without certain capabilities — for example, if sub-agents aren't available, it falls back to sequential execution automatically.

Tested platforms with install scripts:

• Claude Code — installed as a plugin (with enhanced agent/command support)
• Codex (OpenAI) — installed as a skill
• Cursor — installed as a global or project-level skill

Any other agent — copy SKILL.md (plus the references/ directory if you want full template and protocol support) to wherever your agent reads instructions. The files have no external dependencies and no platform-specific logic. Works with Windsurf, Cline, Aider, Continue, Roo Code, Augment, or any other coding agent that reads Markdown-based skills or system prompts.

/plugin marketplace add zhu1090093659/spec_driven_develop
/plugin install spec-driven-develop@spec-driven-develop

After installation, run /reload-plugins to activate.

Use the built-in skill installer (inside a Codex session):

$skill-installer install https://github.com/zhu1090093659/spec_driven_develop/tree/main/plugins/spec-driven-develop/skills/spec-driven-develop

Or install via shell:

bash <(curl -sL https://raw.githubusercontent.com/zhu1090093659/spec_driven_develop/main/scripts/install-codex.sh)

bash <(curl -sL https://raw.githubusercontent.com/zhu1090093659/spec_driven_develop/main/scripts/install-cursor.sh)

Or clone the repo and run locally:

git clone https://github.com/zhu1090093659/spec_driven_develop.git
bash spec_driven_develop/scripts/install-cursor.sh

For any other coding agent, grab the SKILL file and place it where your agent reads instructions:

# Download the SKILL.md
curl -sL https://raw.githubusercontent.com/zhu1090093659/spec_driven_develop/main/plugins/spec-driven-develop/skills/spec-driven-develop/SKILL.md -o SKILL.md

Where to place it depends on your agent:

If your agent doesn't have a formal "skills" directory, you can paste the content of SKILL.md into its system prompt or custom instructions field — the effect is the same.

Simply describe your task to the agent. Each skill triggers on different keywords:

Spec-Driven Develop — large-scale transformation tasks:

• English: "rewrite", "migrate", "overhaul", "refactor entire project", "transform", "rebuild in [language]"
• Chinese: "改造", "重写", "迁移", "重构", "大规模"

Deep Discuss — problem analysis and brainstorming:

• English: "let's discuss", "help me analyze", "I have a problem", "what do you think", "I'm torn between"
• Chinese: "讨论一下", "帮我分析", "我遇到一个问题", "你觉得怎么样", "帮我想想", "我在纠结"

/spec-dev rewrite this Python project in Rust
/dp Our API response times have been spiking recently

When working on a long-running task across multiple conversations, the agent reads docs/progress/MASTER.md at the start of each new session to restore context and continue from where it left off.

When the agent starts a work session, it automatically loads the current phase's pending tasks into the platform's native task tracking tool (e.g. TodoWrite in Claude Code). You get real-time visual progress in your IDE sidebar — no need to open Markdown files manually. MASTER.md remains the persistent source of truth across conversations; the native tool provides in-session visibility.

An optional script exports your progress data to structured JSON, making it easy to import into external project management tools (Linear, Jira, Notion, etc.):

python scripts/export-progress.py docs/progress/

When all tasks are marked complete, the agent archives all workflow artifacts (analysis, plan, progress, sub-SKILL) into docs/archives/<project-name>/ and updates an index at docs/archives/README.md. Nothing is deleted — everything is preserved for traceability.

spec_driven_develop/
├── plugins/spec-driven-develop/              # Self-contained Claude Code plugin
│   ├── skills/
│   │   ├── spec-driven-develop/
│   │   │   ├── SKILL.md                      # Core workflow — works on ANY platform
│   │   │   └── references/
│   │   │       ├── super-philosophy.md       # S.U.P.E.R architecture principles
│   │   │       ├── parallel-protocol.md      # Parallel execution protocol
│   │   │       ├── behavioral-rules.md       # Non-negotiable workflow rules
│   │   │       └── templates/                # Document templates (one per concern)
│   │   │           ├── analysis.md           # Phase 1: with S.U.P.E.R health assessment
│   │   │           ├── plan.md               # Phase 3: with S.U.P.E.R design constraints
│   │   │           ├── progress.md           # Phase 4: cross-conversation tracking
│   │   │           ├── archive.md            # Phase 7: artifact preservation
│   │   │           └── sub-skill.md          # Phase 5: with inlined S.U.P.E.R + checklist
│   │   └── deep-discuss/
│   │       └── SKILL.md                      # Structured deep discussion workflow
│   ├── agents/                               # Claude Code sub-agents (optional)
│   │   ├── project-analyzer.md
│   │   ├── task-architect.md
│   │   └── task-executor.md
│   └── commands/                             # Slash commands (Claude Code)
│       ├── spec-dev.md                       # /spec-dev — launch spec-driven workflow
│       └── dp.md                             # /dp — launch deep discussion
├── scripts/                                  # Installation & utility scripts
│   ├── install-cursor.sh
│   ├── install-codex.sh
│   ├── install-all.sh
│   └── export-progress.py                    # Export progress to JSON
└── LICENSE

The essential files for cross-platform use are the SKILL.md files and the references/ directory. Everything else — agents, commands, plugin manifests — is platform-specific enhancement for Claude Code.

• linux.do

MIT