AI use
dotMARC is built with the help of AI tools. This page says where they're used, which ones, and what stands between AI-assisted work and a release.
Where AI helps
- UI and UX changes. Laying out pages, tightening wording, making screens consistent with each other, and checking how changes look in a browser.
- Planning. Turning a feature request into a written design and a step-by-step implementation plan before any code is written, so the shape of a change is agreed up front.
- Development. Writing code and tests against those plans, tracking down bugs, and keeping the documentation in step with the app.
AI doesn't decide what dotMARC does. Features, scope and priorities are chosen by the maintainer, often from ideas on the public roadmap.
Providers and models
| Provider | Tool | Used for |
|---|---|---|
| Anthropic | Claude Code, with Claude models (mainly Opus) | Planning and design, implementing planned changes, reviewing a whole branch before it's merged, and checking UI changes in a browser |
| GitHub | GitHub Copilot in the editor, mainly with Anthropic's Claude Sonnet and OpenAI's GPT Sol | Inline suggestions and quick questions while writing code by hand |
Model versions change as the providers release new ones; the table names the families in use rather than pinning exact versions.
Human review
Every change is reviewed by a human developer familiar with .NET before it's released. AI output is treated like a contribution from anyone else: it's read, questioned, and changed or thrown away where it isn't right.
Alongside that review:
- Tests. Changes come with automated tests, and new behaviour is written test first where it can be. The full test suite runs on every push and again in the release workflow, which won't publish a release if any test fails.
- Browser checks. UI changes are checked in a real browser before they're committed.
- Releases. A release is only cut when the maintainer decides to, and the release process is the same whoever, or whatever, wrote the code.
Data
dotMARC itself doesn't send your DMARC data, domains or settings to any AI service. AI tools are used to build the app, not inside it.