AI coding agent. Yours to own.
Bring your own OpenAI or Anthropic key.
No subscriptions, no caps. AST-validated edits,
visual verification, tier-bounded cost.
$ git clone https://github.com/BedreddinErkan/zone.git
$ cd zone && npm install
$ cp .env.example .env # add your API key
$ npm run build && npm run serveThen open http://localhost:3000
BYOK pricing
Bring your own OpenAI or Anthropic key. Pay your provider directly — no markup, no monthly subscription, no per-seat fees. Anthropic Sonnet 4.6, Haiku 4.5, Opus 4.7 + OpenAI GPT-4o, GPT-4-Turbo all supported.
No caps, no waiting
Unlike $20-200/mo plans with message quotas and 5-hour rolling windows, Zone has no usage limits. Run as many agent loops as your wallet allows — no fast-request budgets, no cooldowns.
AST-validated edits
TypeScript AST checks syntax integrity, scope discipline, and structural smells before applying any change. Per-file failure tracking forces strategy switch after 3 bad iterations — agent can't loop forever on a broken patch.
How it works
Pick repo
Select your project
Describe task
Tell the agent what to do
Agent iterates
With AST validation
Review & apply
Check diff and merge
Compare with subscription-based agents
Pro tiers cap at $20–200 per month with usage limits and rolling windows. Zone is BYOK — you pay your provider directly, no markup.
Zone BYOK | Cursor Pro $20–200/mo | Claude Code Max $100–200/mo | Codex Pro $20–200/mo | |
|---|---|---|---|---|
| Pricing model | Pay your provider directly | Subscription + usage | Tiered subscription | Tiered subscription |
| Usage limits | None | Quota-based | 5-hour rolling windows | Quota-based |
| Provider support | OpenAI + Anthropic | Multiple cloud | Anthropic only | OpenAI only |
| AST validation | ||||
| Self-hostable | ||||
| Open source | AGPL-3.0 | |||
| Visual verification | Playwright auto-screenshot | |||
| Cost transparency | Per-tier breakdown |
Pricing as of May 2026. With Anthropic prompt caching enabled by default, a typical multi-iter Zone run uses ~$0.05–0.50 BYOK Sonnet 4.6 depending on task complexity.
What's inside Zone
Two built-in control surfaces for visibility and cost governance.
Visual Verification
Configure dev server URL + viewport. Auto-verify takes a Playwright screenshot after every UI patch and embeds it inline.
Tier-Bounded Execution
Per-tier token caps (simple 400K, medium 600K, complex 800K) and iteration limits. Override per-task with forceTier API.
Auto-verify in action
Patch UI. Get the screenshot. See your work without leaving the chat.
1. You type a UI change
You: "Change body background to #1a1a2e"
Zone reads file, generates patch
2. Patch applies + typecheck passes
AST validation → typecheck → build
Running auto-verify...
3. Screenshot inline
Playwright captures dev server
Thumbnail renders inline. Click to expand.
Set DEV_SERVER_URL and VIEWPORT in Zone settings to enable. Requires Playwright installed in your project.
Real numbers, not just “BYOK pricing”
Solo developer running ~50 dispatches/month:
Numbers are approximate per-task averages — your variance dashboard shows real history from your runs.
Already have a Claude Pro or ChatGPT Plus subscription? Use Zone alongside it — your existing API key works.
Pricing as of May 2026. Zone costs scale with your dispatch volume — more dispatches cost more, but never more than what your provider charges.
Run Zone locally
Self-host
$ git clone https://github.com/BedreddinErkan/zone.git
$ cd zone && npm install
$ cp .env.example .env # add your API key
$ npm run build && npm run serveRun Zone on your own infrastructure with full control.
Get in touch
Have a question, feature request, or just want to say hi? Drop us a message and we'll get back to you.