Context-aware completion
Suggestions draw on files around the cursor, so completed code fits your project's naming, structure, and conventions.
CodeCraft AI helps engineering teams write, refactor, and understand code across real projects. It reads context from your repository so suggestions match the way your project actually looks — not a generic template. This page describes what the platform does, who it is designed for, and how the team behind it operates.
Each capability is focused on the parts of a working day that repeat — context switching, boilerplate, and code review.
Suggestions draw on files around the cursor, so completed code fits your project's naming, structure, and conventions.
Rename symbols, extract functions, and update call sites across a repository — with a diff you can review before anything is written.
Ask for a change and get a failing test first. The assistant iterates until the suite passes, keeping regressions visible in the diff.
Ask the assistant to walk through a function or a subsystem. Explanations reference the actual files in your project, not generic docs.
Longer tasks become structured proposals: a plan, a list of files, and a diff. Nothing is applied until you approve it.
Source code stays inside your workspace. Configurable retention, no training on customer data by default, and clear audit logs for teams.
The platform scales from a single developer working on a side project to enterprise engineering organizations with dozens of repositories.
For engineers working on personal or small-team projects who want smarter completions and faster refactors.
For teams of 5–50 engineers who share style rules, review processes, and a common set of repositories.
For larger organizations with strict data governance, audit requirements, and multiple business units.
Rather than treating every request as a fresh prompt, the platform maintains a working picture of your project.
CodeCraft AI is a coding assistant designed for engineers who work in large, opinionated codebases. It maintains a working picture of the project — file structure, dependencies, tests, and per-repository conventions — and uses that context to make suggestions that fit in with everything else. When a task takes more than one file, the assistant produces a structured proposal that a human can review, edit, and accept before anything is written to disk.
The platform is language-agnostic and integrates with the editors and terminals people already use. Teams typically start with completion and light refactoring, then move on to test generation and larger, multi-file tasks as they get comfortable with reviewing machine-generated diffs.
CodeCraft AI is designed to slot into the tools your team already uses. The categories below reflect the current supported surface area; specific tools within each category are listed in the documentation.
Free for individual use on personal projects, with paid team and organization plans for shared configuration, audit logging, and administrative controls. Full plan comparisons and current prices are on the pricing page.
Team and Business plans are billed monthly per user with an annual option. Enterprise pricing is quoted after a scoping conversation.
We publish a full trust page with our current certifications and subprocessors. The summary below is intended to answer the questions we hear most often.
Source code processed by the assistant is not used to train foundation models on customer data by default. Retention is configurable per workspace.
We maintain a SOC 2 Type II report and align with common enterprise security frameworks. Current reports are available under NDA on request.
Enterprise customers can select regional processing for workspaces in the European Economic Area, the United Kingdom, and North America.
If your question is not answered here, please reach out through the contact page.
Yes. Individual users can use the platform on personal projects at no cost. Paid plans exist for teams that need shared configuration, administrative controls, and audit logging. See the pricing page for current details.
The platform integrates with widely used code editors and terminal environments through a standard plugin protocol, and supports more than twenty programming languages, including TypeScript, JavaScript, Python, Go, Rust, Java, C#, C++, Ruby, and PHP. The documentation lists specific integrations.
Code you share with the assistant is processed in secured cloud infrastructure. Enterprise customers can select a processing region. Source code is not used to train foundation models on customer data by default, and retention is configurable.
We currently offer a hosted service. A private deployment option for eligible enterprise customers is described on the trust page and can be discussed as part of a scoping conversation.
We publish a Data Processing Addendum aligned with the General Data Protection Regulation and the UK Data Protection Act. For customers subject to the EU consent framework, our marketing site uses Consent Mode where required. Full details are in the Privacy Policy.
You do. Suggestions surfaced by the assistant become part of your project when you accept them. Our terms describe intellectual property allocation and the boundaries of our license to processed content.
Lumark Software Inc. is a privately held software company incorporated in the State of Delaware, United States, in 2021. We build developer tools that help engineering teams focus on the parts of the work that matter most — designing systems, reviewing changes, and shipping software that behaves the way it is supposed to.
CodeCraft AI is the company's core product. The team is remote-first, with contributors in North America and Europe, and operates on a small number of principles: publish clear documentation, ship software that fails safely, and treat customer data as a responsibility rather than an asset.
For editorial inquiries, security reports, or partnership questions, please use the contact page. Security researchers can find our responsible disclosure policy on the security page.