Dicklesworthstone/agentic_coding_flywheel_setup: 30-Minute Agentic Coding VPS

Dicklesworthstone/agentic_coding_flywheel_setup: 30-Minute Agentic Coding VPS

Setting up a remote development environment with AI coding agents typically involves hours of manual configuration: installing language runtimes, configuring shells, managing API credentials, and ensuring multiple agent tools can coexist without conflict. For developers exploring agentic workflows—where Claude, Codex, and specialized tools collaborate on code—this friction often kills momentum before the first prompt.

Dicklesworthstone/agentic_coding_flywheel_setup (ACFS) addresses this directly. It is a shell-based bootstrapping system that transforms a fresh Ubuntu VPS into a complete multi-agent AI development environment with a single command. With 1,545 GitHub stars and active maintenance through July 2026, it represents a pragmatic approach to infrastructure automation for agentic coding.

What is Dicklesworthstone/agentic_coding_flywheel_setup?

ACFS is an open-source shell automation project maintained by Dicklesworthstone. Its stated purpose: take "complete beginners from 'I have a laptop' to 'I have Claude/Codex/Antigravity agents writing code for me on a VPS'" in approximately 30 minutes.

The project sits at the intersection of several technical categories: infrastructure-as-code for individual developers, AI agent orchestration tooling, and Ubuntu system administration automation. It is written primarily in Bash (per GitHub's language detection), distributed under an MIT license with an OpenAI/Anthropic rider, and currently at version 0.7.0.

What distinguishes ACFS from generic dotfiles or server provisioning tools is its specific focus on the agentic coding workflow. Rather than merely installing a standard developer toolchain, it deliberately configures three distinct AI coding agents—Claude Code, Codex CLI, and Antigravity CLI—with coordination infrastructure, safety tools, and session management. The "flywheel" metaphor in its name refers to the self-reinforcing loop of planning, execution, memory, and review that the installed tool stack enables.

The project includes both the core installer and a companion wizard website at agent-flywheel.com, which provides a 13-step guided experience for users unfamiliar with VPS provisioning or SSH key management.

Key Features

Single-Command Bootstrap. The core value proposition is the curl | bash one-liner that installs 30+ tools across ten phases: user normalization, APT packages, shell setup, CLI tools, language runtimes, AI agents, cloud CLIs, the Dicklesworthstone stack, configuration deployment, and verification. The installer is explicitly idempotent—interruptions are recovered by re-running the same command, with automatic resumption from ~/.acfs/state.json.

Three Pre-Configured AI Agents. ACFS installs Claude Code, Codex CLI, and Antigravity CLI with "vibe mode" aliases that enable maximum velocity: cc runs Claude with --dangerously-skip-permissions, cod runs Codex with --dangerously-bypass-approvals-and-sandbox, and agy pins Antigravity to Gemini 3.1 Pro High. A legacy Gemini CLI alias (gmi) routes to the locked Antigravity launcher.

Agent Coordination Infrastructure. Beyond individual agents, ACFS installs ten "stack" tools for professional agentic workflows: Named Tmux Manager (ntm) for session orchestration, MCP Agent Mail (am) for inter-agent messaging, Ultimate Bug Scanner (ubs), Beads Viewer (bv) for task management with graph analysis, CASS (cass) for unified agent history search, CASS Memory (cm) for procedural memory, Coding Agent Account Manager (caam), Simultaneous Launch Button (slb) implementing a two-person rule for dangerous commands, Destructive Command Guard (dcg) as a Claude Code hook, and Repo Updater (ru) for multi-repo synchronization.

Security Verification. The checksums.yaml system stores SHA256 hashes for upstream installer scripts. Before executing downloaded content, the security library verifies hashes, failing closed on mismatch. This provides content verification beyond HTTPS transport security alone.

Automatic Ubuntu Upgrade. The installer detects and sequentially upgrades older Ubuntu versions to 25.10 via do-release-upgrade, with automatic reboot resumption via systemd service. This can be skipped or retargeted via flags.

Manifest-Driven Architecture. Tool definitions live in acfs.manifest.yaml, with a TypeScript/Zod generator producing category installer scripts and doctor checks. This single-source-of-truth design aims to keep installation and verification logic synchronized.

Use Cases

Individual Developer Experimentation. A developer curious about agentic coding but unwilling to spend a day configuring environments can rent a $5-10/month VPS and have a complete stack running within an hour. The wizard website guides SSH key generation and provider selection for those new to VPS administration.

Reproducible Team Environments. Teams adopting agentic workflows face "works on my machine" problems when each member configures agents differently. ACFS provides a deterministic baseline: every team member's VPS runs identical tool versions, aliases, and safety configurations.

Ephemeral Development Machines. For security-sensitive or experimental work, developers can spin up fresh VPS instances, run ACFS, work with agents, then destroy the instance. The idempotent installer and checkpointed phases make this practical for repeated provisioning.

Agent Workflow Research. The installed stack—including NTM for session orchestration, Agent Mail for coordination, and CASS for cross-agent memory—provides a concrete environment for studying multi-agent collaboration patterns without building infrastructure from scratch.

Educational Onboarding. The interactive TUI onboarding (onboard command), web Learning Hub with 10+ lessons, and beginner wizard lower barriers for developers transitioning to remote, agent-assisted development.

Installation & Setup

The standard installation targets a fresh Ubuntu VPS. The README provides several variants:

Quick install (vibe mode, recommended for throwaway VPS):

curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/main/install.sh?$(date +%s)" | bash -s -- --yes --mode vibe

The ?$(date +%s) cache-buster ensures the latest script version. The --yes flag suppresses prompts; --mode vibe enables passwordless sudo and dangerous agent flags.

Interactive mode (asks for confirmation):

curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/main/install.sh" | bash

Safe mode (production-like, no dangerous flags):

curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/main/install.sh" | bash -s -- --mode safe

Pinned release (recommended for stability):

curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/v0.5.0/install.sh" | bash -s -- --yes --mode vibe --ref v0.5.0

Before installation, validate system readiness:

curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/main/scripts/preflight.sh" | bash

The preflight script checks OS compatibility (Ubuntu 22.04+, with auto-upgrade to 25.10), architecture (x86_64 or ARM64), minimum 4GB RAM and 10GB disk, network connectivity to required URLs, APT lock status, and cached checksum availability.

After installation, verify with:

acfs doctor

Real Code Examples

The ACFS documentation includes several concrete command patterns. Below are directly reproduced examples with explanatory context.

Resuming after interruption:

# The installer is idempotent—simply re-run the same command
curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/main/install.sh?$(date +%s)" | bash -s -- --yes --mode vibe

The state.json checkpointing means phases already completed are skipped automatically. No manual cleanup or state reset is required for normal resumption.

Force complete reinstall:

curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/main/install.sh" | bash -s -- --yes --mode vibe --force-reinstall

This bypasses idempotency checks and reinstalls all tools. Useful when debugging installation issues or when upstream tools have changed significantly.

Update specific components post-installation:

acfs-update --agents-only    # Update only Claude, Codex, Antigravity
acfs-update --runtime-only   # Update only bun, Rust, uv, Go
acfs-update --dry-run        # Preview changes without applying

The update command provides granular control over maintenance operations, with logging to ~/.acfs/logs/updates/.

Security verification workflow:

# List all upstream URLs with stored checksums
./scripts/lib/security.sh --print

# Verify current checksums against live content
./scripts/lib/security.sh --verify

# Generate new checksums after reviewing upstream changes
./scripts/lib/security.sh --update-checksums > checksums.yaml

This demonstrates the defense-in-depth approach: HTTPS for transport, SHA256 for content verification, and version-controlled checksums for auditability.

Advanced Usage & Best Practices

Skip automatic Ubuntu upgrade when working with provider-specific images or kernel modules that may not survive distribution upgrades:

curl -fsSL ".../install.sh" | bash -s -- --yes --mode vibe --skip-ubuntu-upgrade

Monitor upgrade progress during automatic Ubuntu version hops (which can take 1.5-3 hours from 24.04 to 25.10):

journalctl -u acfs-upgrade-resume -f
tail -f /var/log/acfs/upgrade_resume.log

Use --deep doctor checks before critical work sessions to verify agent authentication and cloud CLI tokens, not just binary presence:

acfs doctor --deep

Deep checks use 5-second network timeouts to avoid hanging on connectivity issues.

Apply safe auto-fixes for common configuration drift:

acfs doctor --fix --dry-run   # Preview first
acfs doctor --fix             # Apply with SHA256-verified backups

Safe fixers never delete user files and are logged to ~/.local/share/acfs/doctor.log.

Consider the trade-offs of vibe mode. Passwordless sudo and --dangerously-skip-permissions maximize agent velocity but reduce safety guardrails. For environments where agents interact with production-adjacent systems, safe mode or explicit SLB (Simultaneous Launch Button) workflows provide deliberate friction.

Comparison with Alternatives

ACFS occupies a specific niche: it is less flexible than Nix for package management, less enterprise-oriented than Ansible, but purpose-built for the agentic coding workflow that neither addresses directly. The wizard website and TUI onboarding are unique accessibility features for its target audience.

FAQ

What Ubuntu versions are supported? The installer supports Ubuntu 22.04+ and automatically upgrades to 25.10 unless --skip-ubuntu-upgrade is passed.

Is it safe to run curl | bash? ACFS mitigates risks via HTTPS transport, SHA256 checksum verification of upstream installers, and version-controlled checksums. The --verify security mode helps detect upstream changes.

Can I use ACFS on non-Ubuntu distributions? The README specifies Ubuntu 25.10 as the target platform. Other distributions are not documented as supported.

What happens if installation is interrupted? Re-run the same command. The installer reads ~/.acfs/state.json and resumes from the last completed phase.

How do I update installed tools? Use acfs-update with category flags like --agents-only or --runtime-only. Logs are saved to ~/.acfs/logs/updates/.

What is the license? MIT with an OpenAI/Anthropic rider. Review the repository for full terms.

Does ACFS configure AI agent API keys? No. The acfs services-setup wizard guides credential configuration post-installation, but API keys are not embedded or automatically provisioned.

Conclusion

Dicklesworthstone/agentic_coding_flywheel_setup solves a concrete, well-defined problem: the friction of configuring remote environments for multi-agent AI coding. It does not claim to be revolutionary or universally applicable—it is a focused automation tool for Ubuntu VPS users who want Claude, Codex, and Antigravity running with coordination infrastructure in minutes rather than hours.

The project is best suited for individual developers experimenting with agentic workflows, small teams seeking reproducible environments, and educators guiding newcomers through remote development setup. The 1,545-star community and active maintenance suggest it resonates with this audience.

If you are currently maintaining manual setup scripts for AI coding agents, or have avoided remote agent workflows due to configuration overhead, ACFS merits evaluation. Start with the preflight check, review the checksums.yaml security model, and explore the wizard website if you are new to VPS provisioning.

Explore Dicklesworthstone/agentic_coding_flywheel_setup on GitHub