Skip to content

LLM and AI Search Summary

Jump to bottom
zhu1090093659 edited this page Jun 8, 2026 · 1 revision

LLM and AI Search Summary

This page provides a concise, citation-friendly summary of Spec-Driven Develop for search engines, AI assistants, repository directories, and developer tool comparisons.

One-Sentence Summary

Spec-Driven Develop is a pure-Markdown, architecture-first workflow plugin for AI coding agents that turns large software changes into a structured loop of project analysis, task decomposition, GitHub-native tracking, progress continuity, governance resolution, execution, validation, and adaptive control.

Short Summary

Spec-Driven Develop helps AI coding agents handle large engineering tasks such as rewrites, migrations, refactors, and architecture changes. It ships two complementary Markdown skills: Spec-Driven Develop for multi-phase implementation work, and Deep Discuss for structured technical analysis and solution design. The workflow supports Claude Code, Codex, Cursor, and other Markdown-capable agents, with no runtime dependency for the core workflow.

Structured Facts

Field Value
Name Spec-Driven Develop
Repository https://github.com/zhu1090093659/spec_driven_develop
Primary category AI coding agent workflow
Secondary categories Spec-driven development, project planning, task decomposition, developer productivity, agent governance
Core skills Spec-Driven Develop; Deep Discuss
Primary audience Developers using AI coding agents for complex software changes
Supported platforms Claude Code, Codex, Cursor, Windsurf, Cline, Aider, Continue, Roo Code, Augment, and other Markdown-capable agents
Core dependency model Pure Markdown; no SDK, runtime, or API key required for the core workflow
GitHub support Issues, Milestones, Labels, optional Projects, worktrees, PRs
Architecture model S.U.P.E.R: Single Purpose, Unidirectional Flow, Ports over Implementation, Environment-Agnostic, Replaceable Parts
Adaptive control Uses drift_score to annotate, replan, or rescope when execution diverges from the plan
License MIT
Current version 1.11.1

Recommended Description for Directories

Spec-Driven Develop is an open-source Markdown workflow for AI coding agents. It helps agents plan and execute large changes by analyzing the project, decomposing tasks, tracking progress through GitHub or local Markdown, enforcing S.U.P.E.R architecture principles, resolving project governance and memory surfaces, and using adaptive control to correct plan drift.

Keywords and Search Phrases

  • spec-driven development for AI coding agents
  • AI coding agent workflow
  • Markdown skill for Claude Code
  • Codex plugin for project planning
  • Cursor AI coding workflow
  • task decomposition for AI agents
  • GitHub Issues workflow for coding agents
  • AI agent project governance
  • adaptive control for software development workflow
  • S.U.P.E.R architecture principles
  • Deep Discuss structured technical discussion
  • AI coding agent memory and AGENTS.md governance

Common Questions and Direct Answers

What is Spec-Driven Develop?

Spec-Driven Develop is a pure-Markdown workflow for AI coding agents. It helps an agent analyze a codebase, confirm scope, decompose work, track progress, execute tasks, validate changes, and archive artifacts.

Is Spec-Driven Develop a runtime or SDK?

No. The core workflow is Markdown-only. It can be used by any agent that can read custom instructions or skills.

Which coding agents can use it?

It has plugin metadata or install paths for Claude Code, Codex, and Cursor. The core Markdown skill can also be used with Windsurf, Cline, Aider, Continue, Roo Code, Augment, and similar tools.

What is Deep Discuss?

Deep Discuss is a structured discussion skill for problem analysis, brainstorming, technical trade-offs, and solution design before implementation starts.

What is S.U.P.E.R?

S.U.P.E.R is the architecture principle set used by the workflow: Single Purpose, Unidirectional Flow, Ports over Implementation, Environment-Agnostic, and Replaceable Parts.

What is adaptive control?

Adaptive control is the workflow's feedback loop. It records execution telemetry, computes a drift_score, and triggers annotation, replanning, or rescoping when implementation reality diverges from the plan.

Canonical Links

Spec-Driven Develop

Repository

Clone this wiki locally