spec-driven-development_skill

This skill guides spec-driven development by turning vague feature ideas into testable requirements, design documentation, and actionable tasks for reliable
  • TypeScript

492

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill jasonkneen/kiro --skill spec-driven-development

  • SKILL.md6.3 KB

Overview

This skill implements a systematic three-phase approach to feature development: Requirements, Design, and Tasks. It converts vague ideas into testable requirements, a concrete technical design, and actionable implementation steps to reduce ambiguity and rework. The method improves collaboration, supports AI-assisted workflows, and preserves knowledge for future maintainers.

How this skill works

Start by capturing user stories and writing acceptance criteria in the EARS format (WHEN/IF patterns) to make requirements measurable. Create a design document that defines architecture, components, data models, interfaces, error handling, and test strategy. Break the design into sequenced, traceable tasks sized for 2–4 hour work increments and include decision records and checklists to validate completeness.

When to use it

  • Building complex features with multiple components or integrations
  • High-stakes work where rework costs are significant
  • Projects requiring shared understanding across teams or with AI assistants
  • When you need durable documentation for future maintainers
  • Avoid for trivial fixes, experimental prototypes, or urgent hotfixes

Best practices

  • Write requirements in EARS format and keep them implementation-agnostic
  • Validate requirements with edge cases, constraints, and measurable acceptance criteria
  • Document architecture and component responsibilities before coding
  • Sequence tasks by foundation, feature-slice, or risk-first strategies
  • Keep tasks small (2–4 hours), traceable to requirements, and testable

Example use cases

  • Designing a new account and authentication flow that touches backend, frontend, and email systems
  • Adding a multi-step checkout feature that requires data modeling and error recovery strategies
  • Refactoring a subsystem where clear decision records and tests reduce regressions
  • Onboarding AI assistants: provide explicit context and checklists so outputs are verifiable
  • Coordinating cross-team work where a shared design document prevents duplicated effort

FAQ

Use the EARS patterns (WHEN/IF THEN) to state events, preconditions, and expected system responses. Include normal, edge, and error cases and make success measurable.

What task sequencing strategy should I choose?

Choose foundation-first for stable interfaces, feature-slice for early end-to-end validation, risk-first for uncertain areas, or a hybrid tailored to project needs.

How does this integrate with AI-assisted development?

Provide project context, constraints, and goals, then work phase-by-phase with the AI. Use the quality checklists to validate AI outputs and maintain trace links between requirements, design, and tasks.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational