Pre-engagement triage.
Before the migration, SOW, or audit.

Every Ghost Open™ mode plus teams, projects, Ghost Partner™ profiles and automatic checks on every commit, running entirely on hardware you own against a model server you run. Ghost Open™ is free for one person and sends your code to Anthropic's API with your own key; Local sends nothing anywhere, needs no API key, and is a monthly licence per seat band. Choose Local when the code cannot leave the building, or when a per-token bill is the wrong shape for your team.

No. Local runs against your own model server: Ollama, vLLM, LM Studio, Jan.ai or any OpenAI-compatible endpoint. The model, the weights and every prompt stay on your network.

Request the download, install it, and the trial starts inside the product on first use, registered against that install; the product tells you the day it ends. No card. Convert during the trial and your first month is half price.

Agency ($1,000 a month, 15 seats) and Enterprise ($2,000 a month, unlimited seats). Zero-egress mode refuses, at the code level, any model endpoint that would send source code off the machine, and ships with the egress inventory that proves what the product can and cannot reach. Solo and Small Team run against your own model server too; they do not carry the enforced refusal.

Directly from us, not through a package registry. Buy a licence and your key and a download link arrive by email the moment you pay; for the evaluation trial, EJ sends the tarball or gives your GitHub account access to the delivery repository. Every tarball has a published SHA-256 to check before you unpack it. On Linux the installer provisions a service account and two systemd units (the portal and the scheduler). You upgrade by re-running the installer on the new tarball; your configuration, scan history and licence persist. The changelog lists every release.

Whatever runs your model server. Ghost Architect™ Local measures its own pass times on your machine and estimates from them, so every time estimate you see is yours and not a benchmark from ours. Larger models read code more carefully and take longer per pass; the product's LLM Status page shows your machine's speed, memory and context window so you can see the trade-off before a long scan.

Ghost Open™ is source-available under the Business Source License 1.1. Its Additional Use Grant lets one person use it in production free of charge, on their own machines, for their own work or their employer's or clients' work. Every single-user mode is on, with no quotas, no trial clock and no license key. An organization using it through a shared, multi-user or hosted deployment needs a commercial license; for teams, that is Ghost Architect™ Local.

Yes. Ghost Open™ is BYOK (Bring Your Own Key). This is by design: your code goes directly from your machine to Anthropic's API. We never see it. A typical scan costs $0.23 in Anthropic API usage. You can verify this yourself: the cost of every scan is printed in the output.

Anything that needs more than one person or a server: teams, projects and project tracking, the Project Dashboard, Ghost Partner™ profiles and white-label reports, and automatic checks on every commit. Those are in Ghost Architect™ Local, along with your own model on your own hardware.

Yes. Email [email protected] or connect on LinkedIn and we'll schedule a 15-minute Teams call with a live scan of a sample codebase.

Ghost Architect works on any codebase, any language, any platform, including PHP, JavaScript, TypeScript, Python, Ruby, Java, and more. It's framework-aware and analyzes code structure, dependency relationships, configuration files, and integration patterns regardless of what stack you're running.

Yes. Ghost Architect supports private GitHub, GitLab, and Bitbucket repositories. Authenticate with a personal access token, Ghost clones the repo locally, runs the scan, and removes the local clone when done. Your client's code never touches any third-party infrastructure beyond your own Anthropic API key.