March launched gh-observer; April cleaned up after the party — and then, just to keep things interesting, landed a pull request in Hugo itself. The same Hugo that builds the site you’re reading. Some months the recursion comes to you.

By the Numbers

Metric Count
Pull requests opened 123
… merged 123
… still open 0
… closed without merging 0
Issues opened (per my GitHub profile) 33
Commits (per my GitHub profile) 153
PR reviews given (per my GitHub profile) 84
Repos touched 24
Releases published 12
Bot-filed data updates 0

Second perfect merge month of the year — January went 130 for 130, April went 123 for 123. At this rate I’m either getting better at planning PRs or the reviewers are getting soft, and since most of the reviewers are robots I’m going to say it’s the planning. The busiest repo was gh-observer with 38 PRs in its post-launch polish phase.

Elsewhere

The upstream PR of the month goes to Hugo itself: fixing broken links in the top-level README of the static site generator that has been building https://www.chicks.net since the beginning. Merged, with a lively five-comment review thread of actual humans. There is something deeply satisfying about fixing a paper cut in a tool you use every single day, in public, for free, and having the maintainers say yes.

Releases Shipped

gh-observer v1.8 through v2.1

Four releases in six days — v1.8 and v1.9 on April 24, v2.0 on the 25th, and v2.1 on the 29th. The big-ticket items: cosign release signing with provenance bundles, least-privilege GitHub Actions permissions, and an OpenSSF best practices badge campaign that dragged the whole README up to level 3. The security work that would later flow into ctm got its rehearsal here.

gh-amp is born

gh-amp v0.1 shipped April 19 — the gh extension for wrangling Copilot review suggestions, born directly out of the copilot-review workflow that has been growing in template-repo since January. When your problem is “the robot reviewer is too efficient,” you build a tool, not a complaint.

The fleet

chicks-home v42.2, presentation-notes v2026.03, www v2026.03, fini-projects v1.2, data-curated v1.1, macaw v0.2, and homebrew-chicks v0.3 — twelve releases in one month, and the robot didn’t have to file any of them. April was old-school that way.

Notable Pull Requests

www-chicks-net #318: my first 8 months with AI

2,160 additions and 685 deletions — the biggest blog PR of the year to date — for My first 8 months with AI: the good, the bad, and the ugly. An honest accounting of what the AI-assisted workflow actually changed: what it’s good at, what it’s terrible at, and why I keep the robots around anyway. The follow-up PR cross-posting it to LinkedIn and noting the hugo update felt thematically appropriate — the post about AI tooling, promoted using AI tooling, on a site built by Hugo, days before merging a PR into Hugo. It all runs together after a while.

chicks-home #68: daily_desktop_cleanup learns launchd

588 additions turning the desktop cleanup scripts into proper launchd jobs with just recipes to manage them, plus the month’s most-discussed thread (ten comments, spread across the debug follow-ups — logs in the proper place, timestamps, the whole debugging journey, in public). Personal automation with production engineering standards, because if it’s worth automating it’s worth automating embarrassingly well.

gh-observer #222: release signing and provenance

76 careful additions wiring up cosign signing and provenance bundles for the release artifacts. Small PR, big deal: it’s the difference between “download this binary and trust me” and “verify this binary and trust math.”

www-chicks-net #315: the weather refund form

The weather refund form post landed April 1 with a straight face and a perfectly absurd premise. Eighteen additions, nine deletions, and one deadpan joke sustained across an entire blog post. Comedy is just engineering with worse acceptance criteria.

Embedded Metadata

These are times where it seems good to be explicit about our choices about and usage of AI (Artificial Intelligence). Plus any links to cross-posts I do.

  • This post was drafted by GLM-5.3 in OpenCode using the gh-recap skill; the numbers came straight from GitHub’s API via the gh CLI on 2026-10-30. The review-cycle counts come from the actual PR comment threads, robots and all.
  • The PR carrying this series was code-reviewed in CI by Claude Code — robots reviewing robots, which felt thematically correct.
  • The draft was reviewed and edited by me before publishing. The opinions, jokes, and any errors are mine.