@seclink: Share a particularly useful CLAUDE.md. Spec Driven Build system prompt - Let Claude Code / Gemini CLI design first, then code, driving AI development through the three phases: Requirements → Design → Tasks.

X AI KOLs Following Tools

Summary

Share a Spec Driven Build system prompt for Claude Code and Gemini CLI, letting AI follow a three-phase development process of Requirements → Design → Tasks, designing first before coding to improve code quality.

Share a particularly useful CLAUDE.md. Spec Driven Build system prompt - Let Claude Code / Gemini CLI design first, then code, driving AI development through the three phases: Requirements → Design → Tasks. https://t.co/St6iYirtVW
Original Article
View Cached Full Text

Cached at: 06/08/26, 01:20 PM

Sharing a particularly useful CLAUDE.md.

Spec Driven Build System Prompt - Let Claude Code / Gemini CLI design before coding, driving AI development through three phases: requirements → design → tasks.

https://t.co/St6iYirtVW


XiaomingX/systemprompt-CLAUDE

Source: https://github.com/XiaomingX/systemprompt-CLAUDE

Spec Driven Build - Claude Code / Gemini CLI System Prompt

Let AI design before coding, saying goodbye to inefficient development of “generating a bunch of garbage code with one sentence.”

What is this?

A set of Spec Driven Build system prompts for Claude Code and Gemini CLI. AI does not write code directly, but strictly goes through three phases:

Idea → Requirements Document → Architecture Design → Task List → Step-by-Step Coding

Each phase requires your review and confirmation to ensure the AI understands correctly before proceeding to the next step.

Three Steps to Get Started

1. Start Spec Mode

# Claude Code
claude -p "$(< .prompt/cli-spec-agent-system-prompt.md)"

# Gemini CLI
gemini chat --system "$(< .prompt/cli-spec-agent-system-prompt.md)"

2. Tell AI Your Idea

I want to develop a user registration and login feature

3. Review and Confirm Phase by Phase

AI will generate requirements, design, and task list in sequence. In each phase, you can supplement and modify. When satisfied, reply “Confirmed”.

Generated files are stored in the .kiro/specs/{feature_name}/ directory.

Reuse in Other Projects

# Copy prompt files to your project
mkdir -p /your/project/.prompt
cp .prompt/cli-spec-agent-system-prompt.md /your/project/.prompt/

# (Optional) Add shortcut command to ~/.zshrc
alias claude-spec='claude -p "$(< .prompt/cli-spec-agent-system-prompt.md)"'

Use Cases

  • New feature development: Produce requirements and design first to avoid rework
  • Team collaboration: Unified spec document aligns AI and human understanding
  • Complex systems: Break down into actionable task lists, advance step by step

License

MIT

Similar Articles

@yaohui12138: I've finished reading it. Here are some key takeaways I've compiled for everyone: In this session, he primarily broke down a core mechanism overlooked by 90% of users: the CLAUDE.md context injection system. This system is divided into three levels: Enterprise-level: Organization-wide mandatory rules that cannot be overridden by individual settings. Project-level: Team-shared code standards and workflows. Loc...

X AI KOLs Timeline

The article shares key insights from a workshop by Boris on using CLAUDE.md for context injection in Claude, highlighting three usage levels, specific commands like /loop, and plan mode to improve developer workflows.

@thinkszyg: https://x.com/thinkszyg/status/2066837941477920993

X AI KOLs Timeline

A practical guide for developers (especially AI coding tool users) on how to safely and efficiently use Claude Code, Codex, and other tools for multi-agent parallel development, focusing on best practices such as task decomposition, file isolation (worktree), boundary control, sequential merging, etc., to avoid file conflicts and chaos.

@seclink: Developers are shifting from "one-shot prompts" to "agentic loops"—where AI like Anthropic's Claude sets its own goals, uses tools to execute actions, observes results, and iterates until the task succeeds. Boris Chern, developer of Claude Code...

X AI KOLs Following

Developers are moving from one-shot prompts to an 'agentic loop' model, letting AI autonomously set goals and iteratively execute tasks using tools. Claude Code developer Boris Cherny has abandoned traditional IDEs entirely, now running hundreds of agents to monitor issues and merge PRs—even doing so from his phone.