Skip to content

README.md

Latest commit

 

History

History
72 lines (54 loc) · 2.48 KB

README.md

File metadata and controls

72 lines (54 loc) · 2.48 KB

Commands glossary

Pipeline

Commands form a feature pipeline — each step feeds into the next. Every predecessor is optional input (except /code-execute, which requires a plan), so each command also works standalone.

/explore → /one-pager → /critique → /code-explore → /code-plan → /code-execute → /code-review → /code-review-peer → /document

All commands

Command Mode Description
/explore Ask Product discovery — validates the problem, maps 5-7 competitors, surfaces risks.
/one-pager Agent Writes a product one-pager (Concept, Problem/Solution, or Decision format).
/critique Ask Senior leadership stress-test — challenges need, approach, and risk. Rates 1-5.
/code-explore Ask Analyzes codebase to understand how a feature fits — dependencies, structure, edge cases.
/code-plan Plan Produces a structured implementation plan with progress tracking.
/code-execute Agent Implements the plan step by step, updating the tracking document as it goes.
/code-review Ask Thorough code review focused on production readiness and security.
/code-review-peer Ask Evaluates peer review findings — verifies each against real code, separates signal from noise.
/document Agent Updates documentation after code changes. Reads actual code, never trusts existing docs.

Standalone (not part of the pipeline):

Command Mode Description
/code-explain Ask Three-level explanation (core concept, how it works, deep dive) for designer.
/copy-review Ask Reviews interface copy against a guidelines file and UX writing best practices — suggests fixes.
/create-presentation Ask Builds presentation structure, slides, and speaker notes using the Kapterev dramaturgy method.

How to create a good command?

  • 2500 characters max (including spaces and formatting)
  • Use the template below
  • Keep it very simple and to the point, reference results of other commands that might add useful context
---
name: command-name
description: One-line description of what this command does
---

# Command name

## Persona
Act as a... 

## Input
**Required:** ... 
**Optional:** ...

## Your task

### Mode: Ask / Agent / Plan / Debug
E.g. Don't make any changes until my explicit approval

### 1. Step one name
- ...
- ...

### 2. Step two name
- ...
- ...

### N. Step N name
- ...
- ...

## Output
...