Use when adding new error messages to React, or seeing "unknown error code" warnings.
npx skills add jjenzz/agent-skills --skill "vertical-slice-delivery"
Install specific skill from multi-skill repository
# Description
Required methodology for planning, ideating, and delivering features or tasks. Use when user asks to "plan", "break down", "continue", or "figure out steps".
# SKILL.md
name: vertical-slice-delivery
description: Required methodology for planning, ideating, and delivering features or tasks. Use when user asks to "plan", "break down", "continue", or "figure out steps".
Vertical Slice Delivery
This skill defines the default delivery methodology for planning and shipping
work as independently valuable, reviewable vertical slices.
When this skill is selected, you MUST load and obey the accompanying AGENTS.md.
When to Apply
Use this skill whenever work is being:
- planned
- split into tasks or steps
- delivered incrementally
- resumed after interruption
Any deviation from vertical slice delivery MUST be explicit and justified.
Core Loop
Each slice = one commit + one review. No slice begins until the previous is approved.
- IMPLEMENT — one behavioral slice only
- COMMIT — with slice ID
[S1] - STOP — mandatory review (do not proceed)
- WAIT — approved → next slice, blocked → fixup and re-review
Key Concepts
| Concept | Summary |
|---|---|
| Behavioral Slices | Deployable behaviors, not technical tasks. "Can a user demo this alone?" |
| Role Separation | Delivery writes code; Review audits and approves. Self-approval prohibited. |
| Git Discipline | Status before work, explicit paths, fixups for amendments |
| Code Guidelines | Improve touched code only, no speculative abstraction |
Full Document
All rules with examples and details: AGENTS.md
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.