FAQ

FAQ

What is Spec-Driven Develop?

Spec-Driven Develop is a pure-Markdown spec-driven development workflow for AI coding agents. It helps agents analyze a project, refine scope, break work into tasks, track progress, execute changes, validate results, and archive artifacts.

How is it different from a prompt template?

A prompt template is usually a single instruction block. Spec-Driven Develop is a repeatable workflow with phases, templates, GitHub integration, progress tracking, governance resolution, architecture checks, and adaptive control.

Does it require a runtime, SDK, or API key?

No. The core workflow is Markdown-only. GitHub-native tracking uses the gh CLI when available, but the workflow can run in local-only Markdown mode.

Which AI coding agents can use it?

The repository includes support for Claude Code, Codex, and Cursor. The core Markdown skills can also be used by Windsurf, Cline, Aider, Continue, Roo Code, Augment, or any agent that reads custom instructions.

What are the two skills in this repository?

The repository ships two complementary skills:

• Spec-Driven Develop for large implementation projects.
• Deep Discuss for structured problem analysis and solution design.

When should I use Spec-Driven Develop?

Use it when the work is large enough that analysis, planning, task tracking, and validation matter. Typical examples include rewrites, migrations, broad refactors, architecture changes, and multi-phase feature projects.

When should I use Deep Discuss?

Use Deep Discuss when the problem is still unclear. It is best for debugging direction, architecture choices, product or technical trade-offs, root-cause analysis, and brainstorming.

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

S.U.P.E.R is the workflow's architecture model:

• Single Purpose.
• Unidirectional Flow.
• Ports over Implementation.
• Environment-Agnostic.
• Replaceable Parts.

What is drift_score?

drift_score measures how much execution reality has diverged from the plan. It increases when tasks take longer than expected, S.U.P.E.R improvements fail to materialize, or unplanned dependencies appear.

What happens when drift gets too high?

The workflow can annotate the next task, replan the remaining tasks, or return to scope confirmation with the user.

Does the workflow always use GitHub Issues?

No. It auto-detects GitHub availability. If gh CLI and repository access are available, it can use GitHub Issues and Milestones. If not, it uses local Markdown tracking.

Does it create a local memory file by default?

No. The workflow prefers the active coding agent's native memory surface. A repo-local fallback memory file should only be used when the project already declares one or the user explicitly selects it.

Is the project open source?

Yes. Spec-Driven Develop is licensed under MIT.

Where is the source repository?

The source repository is https://github.com/zhu1090093659/spec_driven_develop.