Pricing
Licensed directly, sized to your team.
There is no public price list. Tell us about your team and the projects you run, and we’ll quote a license that fits.
Every license includes
- Your project and each developer’s profile, carried into every AI session
- Every commit measured against the project, with drift named to the AI and to the person
- Specialist reviews of your code and product, every finding cited to its file and line
- The dashboard for every project: measures, findings, sessions and decisions
- Organization model policy: which models your projects may use
- An independent second check on the measuring itself
- The record of every decision and ruling, kept in your own repositories
Every protection is in every license, and some are not features at all: your code goes to a new AI company only when your team says yes, and the record of every decision stays yours. Nothing that keeps you safe is held back for a higher tier.
Regulated and government programs
For programs under ATO and compliance mandates, a license adds:
- ATO evidence as a side effect of the work: SSP draft, NIST 800-53 mapping, POA&M
- Tamper-evident audit-trail anchoring
- Licensed per system boundary
Questions about licensing
How is a license priced?
Directly, sized to your team and the projects you run. There is no public price list: we talk first, then quote.
What does a license include?
Everything x-Arbiter does, for every project: the guidance in each AI session, every commit measured against the project, the specialist reviews, the dashboard, model policy, the independent check on the measuring, and the record in your repositories. No protection is held back for a higher tier.
What data leaves our machines?
The dashboard receives results — grades, findings and session summaries — never your code or file contents. Code travels in two places, never to Ghost Works: the commit judge sends it to the model company you choose, and the specialist reviews run inside your AI tool, on its model. A model chosen on the dashboard that would send your code to a new company waits until someone on your team agrees.
Which AI tools and models does it work with?
Claude Code and Cursor run x-Arbiter’s hooks; Copilot and Windsurf get its rules file and tools, with write-time enforcement there not yet verified. For the judge that measures each commit you choose the model: Claude, OpenAI, Gemini, xAI, or a local one through Ollama. Each is tested on your own work before its verdicts count, and one that fails is reported as failing.
Can we run it without the dashboard?
Yes. The framework runs in your repositories on its own. The dashboard adds the view across projects, the trends and the shared record.
Is it released?
Not publicly. It is licensed directly to businesses today and installed from our private registry; the public release comes later.
Talk to us about a license.
x-Arbiter is licensed directly to businesses. Tell us about your team and what you’re building, and we’ll show you what it measures on one of your projects.
Not yet publicly released; licensed directly.