Agent Skills

A collection of agent skills that I have found to be useful in my day-to-day work. Each skill is a self-contained module that can be invoked by the agent when appropriate. The skills are designed to be composable, so that they can be combined to solve more complex problems.

Installation

Install any skill from this repo into another repo with:

npx skills@latest nzepeda/skills/{skill-name}

Install the full feature workflow into another repo with one command:

curl -fsSL https://raw.githubusercontent.com/nzepeda/skills/main/scripts/install-feature-suite.sh | bash

That installer performs a non-interactive local project install for Codex by default. To install for Claude, pass claude:

curl -fsSL https://raw.githubusercontent.com/nzepeda/skills/main/scripts/install-feature-suite.sh | bash -s -- claude

Manual project install paths:

• Codex: .agents/skills/<skill-name>/SKILL.md
• Claude: .claude/skills/<skill-name>/SKILL.md

Skills

• grill-me: Stress-test a plan or design through a rigorous interview that walks each decision branch to a shared understanding.
• write-a-prd: Create a PRD through user interviews, codebase exploration, and module design, then save it under ./prds/.
• prd-to-plan: Turn an existing PRD into a phased implementation plan using tracer-bullet vertical slices, written to ./plans/.
• feature-owner: Orchestrate a feature end-to-end, starting with a mandatory clarification interview, then require explicit approval before kickoff, technical planning, and implementation.
• feature-to-brief: Normalize a rough feature request into a structured feature brief.
• design-brief-generator: Turn a feature brief into a design review artifact with flows, states, and open questions.
• tech-spec-generator: Turn an approved brief and design direction into a technical proposal with tradeoffs.
• implementation-planner: Break an approved spec into reviewable implementation chunks.
• code-implementer: Implement one planned chunk at a time and keep the diff narrow.
• code-reviewer: Review a chunk for regressions, scope creep, and missing tests before it is considered complete.

Feature Automation Flow

The feature automation skills are meant to be composed in this order:

1. grill-me
2. feature-owner
3. feature-to-brief
4. design-brief-generator
5. tech-spec-generator
6. implementation-planner
7. code-implementer
8. code-reviewer

The intended operating model is:

• You describe the feature once.
• feature-owner first uses grill-me to clarify outstanding questions and reach a shared understanding.
• You explicitly approve that shared understanding.
• feature-owner asks for explicit approval before starting the workflow.
• The workflow creates ./docs/features/<feature-slug>/.
• You explicitly approve the design direction.
• You explicitly approve the technical shape.
• You explicitly approve implementation before any code work starts.
• The remaining work proceeds one reviewable chunk at a time.

feature-owner is intended to act as the orchestrator agent. The stage skills can be delegated through the agent runtime's available subagent or task facilities when supported, but only feature-owner should own workflow state and approval-gate transitions.

If you want the whole workflow available in a target project, use the installer script above. It installs:

• grill-me
• feature-owner
• feature-to-brief
• design-brief-generator
• tech-spec-generator
• implementation-planner
• code-implementer
• code-reviewer

Equivalent direct command:

npx skills@latest add nzepeda/skills \
  --skill grill-me \
  --skill feature-owner \
  --skill feature-to-brief \
  --skill design-brief-generator \
  --skill tech-spec-generator \
  --skill implementation-planner \
  --skill code-implementer \
  --skill code-reviewer \
  --agent codex \
  --yes

For Claude, use the same command with --agent claude.