Neura Market
prompt logo

prompt

Free
FreeFree tier
Type
Open Source
Visit Website GitHub (40 stars)

About prompt

Opinionated Agent Team Designer is a prompt template for designing multi-role tooling systems within AI coding agents (such as Claude Code, Codex CLI, Gemini CLI, Cursor). Inspired by Garry Tan's open-source software factory (gstack), the prompt advocates replacing a single generalist AI assistant with a team of narrow, high-agency roles—CEO/Strategist, Designer, Eng Manager, Release Manager, Doc Engineer, and QA/Tester. It defines each role's mandate, trigger, input/output contracts, and anti-scope to prevent role creep. The prompt also outlines a review lattice (plan reviews, code reviews, pre-ship reviews), an invocation protocol (slash-commands, automatic triggers), and context-passing rules, all designed to ensure roles review each other before shipping.

Key Features

Defines six executive roles: CEO/Strategist, Designer, Eng Manager, Release Manager, Doc Engineer, QA/Tester
Each role has a one-line mandate, trigger condition, input/output contracts, and anti-scope
Designs a review lattice: plan reviews, code reviews, and pre-ship reviews with mandatory cross-role sign-offs
Slash-command invocation protocol for explicit role invocation (e.g., /plan-ceo-review, /qa-only)
Automatic triggers based on file-type patterns, commit-message tags, and CI state changes
Anti-scope rules to prevent role creep and ensure each role focuses on its core responsibility

Pros & Cons

Pros
  • Narrow, specialized roles produce more consistent and higher-quality outcomes than a single generalist
  • Explicit review lattice ensures no role ships work without at least one other role's sign-off
  • Anti-scope definitions prevent role creep and maintain focus
  • Slash-command protocol makes role invocation clear and reproducible
Cons
  • Complex to set up and maintain for small teams or simple projects
  • Requires multiple role definitions and explicit invocation, which may slow down quick tasks
  • Assumes the AI coding agent environment supports custom role definitions and slash-commands

Best For

Structuring AI coding agents for complex software projects requiring specialized oversightImproving code quality and consistency through multi-role review workflowsDesigning AI-assisted development pipelines that separate strategy, design, engineering, and QAImplementing opinionated best practices for AI-driven software factoriesEnhancing collaboration in AI pair-programming by assigning explicit roles

FAQ

What is the core thesis of this prompt?
The core thesis is that a coding agent should not be a single generalist; it should be a team of narrow, high-agency roles that are invoked explicitly, review each other, and ship with checks.
Which roles are defined in the minimum set?
The minimum role set includes: CEO/Strategist, Designer, Eng Manager, Release Manager, Doc Engineer, and QA/Tester.
What is anti-scope?
Anti-scope defines what each role is NOT allowed to do, preventing role creep and ensuring each role only handles its core responsibilities.
How is the review lattice structured?
The review lattice includes plan reviews (before execution), code reviews (after implementation), and pre-ship reviews (before deployment), with no role shipping its own work without explicit sign-off from another role.