turbo-flow

The rules layer for AI-written code — rig-lite: a fail-closed cross-model review gate, a written constitution, git-versioned memory. Any harness, any repo. v1-4

What is turbo-flow?

The rules layer for AI-written code — rig-lite: a fail-closed cross-model review gate, a written constitution, git-versioned memory. Any harness, any repo. v1-4 were the agentic dev environment.

Overview of turbo-flow

What it is

Turbo Flow is an open-source project by Adventure Wave Labs (Marcus Patman). It is the rules layer (rig-lite) for AI-written code. It bundles no agent and no orchestrator. It works with any tool that reads AGENTS.md and can run bash. Versions 1–4 (2025–2026) packaged an agentic development environment around Claude Code and Ruflo. By v5.2 (October 2026) the orchestration stack was fully removed: Turbo Flow is now the rules layer (rig-lite). The integrated successor, Turbo Rig, runs in private beta.

Key features

  • rig-lite: a portable governance kit that ships no harness, installs nothing, and depends only on bash.
  • gate.sh: a cross-model review gate that enforces builder ≠ reviewer, fail-closed, never merges. It supports --pr, --sweep, -C options and posts verdicts as GitHub comments.
  • protect.sh: makes the gate a GitHub rule by turning on branch protection with the gate's verdict as a required status check (gate-verdict).
  • wt.sh: creates isolated worktrees per parallel writer.
  • init-repo.sh: one-command onboarding for any repo, creating a thin AGENTS.md and symlink.
  • secret.sh: encrypted-at-rest secrets with auto-detected backend (macOS Keychain, libsecret, age, or chmod-600 plaintext).
  • tokens.py: live token-burn dashboard across zcode, claude, codex, and gate, reading local usage data read-only.
  • constitution.md: the written laws (builder ≠ reviewer, agents never merge, worktrees for parallel writers, secrets encrypted, state on disk).

Use cases

  • Enforcing cross-model review in AI-assisted development: a builder from one model family, a reviewer from a different family.
  • Gating GitHub PRs with a fail-closed mechanism that never merges without an approved verdict.
  • Managing parallel writers using isolated worktrees.
  • Onboarding new repositories with a constitution and git-versioned memory.
  • Monitoring token usage across multiple AI coding tools.

Getting started

The README provides a ten-minute install section with a one-liner command: bash # one-liner (stages ./rig-lite from a pinned, checksum-verified rel … The tf command is installed automatically in Codespaces and by init-repo.sh anywhere else.

License and pricing

License: MIT. Pricing type: open source (free).

Based on the project's public README and repository metadata.

Add this badge to your README

Featured on YunoDB Badge styles & help

Maintain this project? Show that it is featured on YunoDB.

Markdown[![Featured on YunoDB](https://yunodb.com/badge/marcuspatturbo-flow.svg)](https://yunodb.com/tools/marcuspatturbo-flow?utm_source=github&utm_medium=badge)
HTML<a href="https://yunodb.com/tools/marcuspatturbo-flow?utm_source=github&utm_medium=badge"><img src="https://yunodb.com/badge/marcuspatturbo-flow.svg" alt="Featured on YunoDB" height="20"></a>