Agent Flywheel — AI Agents Coding For You

Tagline: "AI Agents Coding For You" — from zero to agentic coding in 30 minutes.

Author: Jeffrey Emanuel (jeffreyemanuel.com, X: @doodlestein, GitHub: Dicklesworthstone)

Key Facts

• 30+ tools installed
• 3 AI agents pre-configured (Claude Code, OpenAI Codex CLI, Google Gemini CLI)
• ~30 minutes setup time
• Free and open-source

Tools Mentioned

Claude Code, Codex CLI, Gemini CLI, Bun, Rust, Go, tmux, zsh

Featured Guides

1. The Flywheel Methodology

Described as "the definitive guide to planning-first agentic development," covering decomposition of complex projects into "beads," detecting convergence, coordinating agent swarms. Contains 25 in-depth sections and 9 interactive visualizations.

2. The Core Flywheel

A beginner-friendly starting point using just three core tools: Agent Mail, beads, and bv. Contains 3 core tools, 4 interactive visualizations.

Feature Breakdown

• One-liner Install: A single curl command. "No manual configuration, no dependency hell."
• Three AI Agents: Claude Code, Codex CLI, Gemini CLI "all configured with optimal settings for coding."
• Idempotent & Safe: Rerunnable, phases resume on failure, SHA256-verified installers.
• Vibe Mode: Described as "Passwordless sudo with dangerous flags enabled for maximum velocity on throwaway VPS environments."
• Modern Shell: zsh + oh-my-zsh + powerlevel10k with lsd, atuin, fzf, zoxide.
• Interactive Tutorial: Running onboard after setup gives guided lessons "from Linux basics to full agentic workflows."

The Ecosystem: "Agentic Coding Flywheel"

Twenty interconnected tools:

"13 Steps to Liftoff" Wizard Sequence

1. Choose OS
2. Install Terminal
3. Generate SSH Key
4. Rent VPS
5. Create Instance
6. SSH Connect
7. Set Up Accounts
8. Pre-Flight Check
9. Run Installer
10. Reconnect
11. Verify Key
12. Status Check
13. Launch Onboard

"Why a VPS?" Section

Three arguments:

• Not Your Laptop: "Each agent uses ~2GB RAM. With 10+ agents, you need 48-64GB." Running locally drains battery and slows everything.
• Not AWS/GCP/Azure: "A 64GB VPS costs ~$40-56/month flat. Equivalent cloud resources would cost 3-5x more."
• Works While You Sleep: "AI agents can refactor, test, and iterate autonomously—compounding progress overnight."

"Is This For You?" — Honest Assessment

For you if: - You want AI writing "real, production code" — "Full implementations, not just suggestions" - You find sites like Lovable.dev too limiting - You're willing to invest roughly $500/month in subscriptions - You can follow step-by-step instructions — "No coding experience required, just patience"

Not for you if: - You want a completely free solution - You only want occasional AI help with snippets - You're looking for mobile-first development - You need "enterprise compliance out of the box"

Pricing / Investment Breakdown

The page notes: "a junior developer costs $5,000+/month." The pitch is that for under $700 you get "10+ AI agents working 24/7."

About Section

Jeffrey Emanuel states he built this because he was "being inundated with requests from friends, older relatives, and strangers on the internet" asking for help getting started with AI for software development. He wanted a single resource covering everything "from soup to nuts" even for people with "almost no computer expertise, just motivation and desire."

He originally built the open-source agentic coding tools for himself to move faster in consulting work with Private Equity and Hedge Funds.

External Links

• GitHub repo: Dicklesworthstone/agentic_coding_flywheel_setup
• NTM: github.com/Dicklesworthstone/ntm
• Agent Mail: github.com/Dicklesworthstone/mcp_agent_mail
• Jeffrey's personal site and X profile