There is a lot of cool things you can do with CI/CD, such as:
- Build complex apps on many OSes
- Build on simple tag change
- Run tests and security checks
- Deploy builds to releases that are consumed by users via update automatically
- Build and ship apps to Apple TestFlight/App Store
That covers the shipping part. However, when you ship a new version or some fix, you should always add information regarding what has changed, and it should be written in such a way that it highlights what users care about. It should not include the nitty gritty code changes.
Historically, writing changelogs has been done by humans by hand, as understanding what constitutes "what the user cares about" versus what actually changed technically wasn't possible to synthesize based on code changes alone. Some more revisionist solutions, like xyz, GitHub now offer writing verbose changelog in their web markdown based on diff between tags. This is manual and requires separate GitHub Copilot subscription.
The Solution: By using omni --release directly in CI you get:
- Changelog that highlights what users care about
- Process data from the 6 core context sources (setup connectors via prompt)
- Your choice of AI BYOK: OpenAI, Anthropic, Gemini or local by/code subscriptions (local 64mm)
- Use the release.md slug or inject user prompt to format the frequency to run the way you prefer
- Release works with assets, drafts, beta, release etc.
Read full setup instruction for CI/CD here. Or point your agent to this markdown file and it can setup automatic changelog customized to your project.
The outcome: Razor sharp changelogs based on the knowledge graph of your project. Perfect for projects that move fast and change a lot.