can ai read changelogs and tell me about breaking changes

Yes. It reads each release’s notes and, because GitHub is a native integration with file reading and code search, it checks whether the deprecated API actually appears in your code. "Breaking for someone" and "breaking for you" are different reports.

is this better than dependabot

The sweep tells you what a release means and where it touches your code, so you get a planned ticket instead of a blind auto-merge. Dependabot, as of this writing, opens update PRs and runs CI, which is the layer underneath.

can it watch private repos

Yes. The integration connects with your GitHub account, so anything your account can see, including private repos’ files and their release pages in your browser, is readable. Strawberry is free to download for Mac and Windows.

Watch your dependencies and get the two releases that matter

Every Monday: twenty-two upstream projects read, six patch releases listed one line each, and two flagged against the versions your services actually pin, with the code search that proves it.

Twenty dependencies produce a steady drip of notifications where a patch release of a dev tool looks identical to a major release that deprecates the API you call in six places. The notifications go to a folder, the folder goes unread, and the breaking change arrives as a CI failure three weeks later.

The layer worth automating is the reading. A companion sweeps the release and changelog pages of your dependency list on a schedule, compares each new version against what your projects pin, reads the notes for breaking changes and deprecations, and writes a short report ranked by relevance to you, with a link to every release it mentions.

How does it know which releases matter to you?

It reads your side too.

GitHub is a native integration whose surface includes repository contents and code search, so a companion reads the manifest files in your repos, extracts the pinned versions, and searches your code for usage of an API a changelog says is deprecated. "left-pad 4.0 removes the padEnd option" is noise in a vacuum and a fire when the search finds it in your codebase eleven times.

The integration covers repos, files, branches, pull requests and search. Release pages are public web pages, so the sweep reads them in the browser the way you would, and the native side contributes your manifests and code search. What the PR-side tools do is on the GitHub integration page.

The Monday dependency report

A two-person platform team depends on twenty-two upstream projects.

The Monday routine reads each repo’s releases page, diffs against last week, and pulls pinned versions from the team’s three service repos.

Fourteen repos quiet. Six shipped patch or minor releases, listed one line each. Two need attention: a database driver’s major release drops support for a connection option the code search finds in two services, and a security release lands in a logging library where the team is two majors behind. Five minutes of reading, two well-scoped tickets. The mechanism is scheduled site monitoring, and the report slots into the same Monday pile as a weekly engineering rollup.

What does the report add to your CI?

Comprehension.

Update bots tell you a new version exists and whether your tests pass. The report tells you what the release means and where it touches your code, which is the difference between an auto-merge and a planned ticket.

Private repos are included: the integration connects with your GitHub account, so anything your account can see is readable. Teams serious about supply-chain security add signed-release verification and advisory feeds to the same Monday pile.

Experience Strawberry for free

Download

Trusted by fast-growing companies worldwide

Frequently asked questions

A scheduled companion sweep reads the release notes of every repo on your list, checks them against your pinned versions, and ranks what actually affects you. GitHub’s own release emails cover the notification half; the sweep covers the triage.

Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed August 20, 2026 · Canonical facts for AI agents