AI workflow runtime

Team OS for Codex and Claude

A local operating layer that keeps AI product work moving through strategy, planning, design, implementation, finish review, and QA without losing context between sessions or assistant providers.

Start with the runtimeRead the quickstart
~/project · team-os
primary workflow
Assistant
Handoffcodex>
state trace locked · next role has context

Why it exists

The runtime is the product.

Role prompts are useful, but the durable advantage is the low-level runtime that makes AI work legible, resumable, and checkable.

Runtime that preserves context

Deterministic commands own ids, validation, handoffs, stale-ref gates, sprint sync, and archive behavior so the AI does not have to remember the operating state from chat.

Explore runtime

Sprint artifacts that stay useful

UI, FE, and QA write detailed evidence into the active sprint, then Team OS archives completed sprint workspaces so future sessions can load only the context that matters.

See artifacts

PIP ledger for self-improvement

Recurring misses become role-level signals. Future role runs can see patterns like weak handoffs, missing tests, or repeated responsive issues before repeating them.

Understand PIP

QA closes the loop

QA creates matrices, links automation or E2E coverage when required, files precise bugs, reopens failed work, and records ship confidence after the latest delivery change.

Read QA role

Clear division of labor

Scripts handle correctness. AI handles judgment.

Team OS keeps deterministic state work out of the model and keeps creative product decisions out of scripts. That split is what lets context survive across handoffs.

Programmatic Runtime

Create state, assign stable ids, validate references, archive sprint workspaces, block invalid completions, and print the next handoff.

AI Role Judgment

Choose the product bet, plan slices, design the interface, implement code, polish the experience, and decide the right regression depth.

Sprint memory

Artifacts stay detailed while the active context stays small.

Every sprint gets its own UI, FE, and QA workspace. When the sprint is done, the runtime archives that workspace instead of dragging every old screen, audit, and test matrix into the next prompt.

Open sprint artifact docs
01UI

Design tokens, screens, flows, components

02FE

Surface audits, finish reviews, QA watch points

03QA

Matrices, bugs, ship-confidence decisions

04Archive

Completed sprint context moves out of the active window

Role skills

Each role leaves something the next role can use.

The skills are not just personas. They are evidence producers that write through the same runtime contract, from UI tokens to QA ship confidence.

CEO

Frames the opportunity and decides whether it deserves work.

Product

Turns validated direction into scoped sprint tasks and acceptance criteria.

UI

Tokenizes the design system and writes refs Code, FE, and QA can trust.

Code

Implements against artifacts while preserving architecture and evidence.

FE

Makes the interface feel finished across states, breakpoints, and input modes.

QA

Generates durable tests, files bugs, and records defensible release confidence.

Start simple

Install it, initialize state, and let the next handoff tell you where to go.

Install Team OSCommand reference

Pricing

Simple, transparent pricing

Start free, keep your workflow local, and upgrade when your AI delivery process needs more structure. No hidden fees, cancel anytime.

Free

Everything you need to install Team OS locally and validate the workflow on a real project.

$0forever

 

Start for free

What is included

  • Local Codex or Claude Code setup
  • Shared .team-os runtime state
  • Core CEO, Product, UI, Code, FE, and QA roles
  • Validation, report, and next-handoff commands
  • Sprint artifact archive basics
Most popular

Pro

More structure for builders running Team OS across multiple products, clients, or active workstreams.

$16/ mo

billed annually

Start Pro trial

Everything in Free, plus

  • Unlimited local product workspaces
  • Provider-aware Codex and Claude workflows
  • Advanced PIP learning and role-quality loops
  • Durable sprint closeout and reopen evidence
  • Priority implementation guidance
  • Future hosted collaboration features

Enterprise

Custom workflow enablement, security review, and onboarding for teams standardizing AI delivery.

Custom
Contact sales

Everything in Pro, plus

  • Team rollout and workflow mapping
  • Custom role and runtime policies
  • Security and governance review
  • Dedicated onboarding support
  • Custom SLAs and procurement support

Prices in USD. Have questions about the right plan? Talk to the team.