ci: match scoped release PR titles in auto-merge

[jdx]

https://entire.io/gh/jdx/tak/trails/19

The release PR was never merged automatically. auto-merge-release.yml looked for an open PR whose title starts with chore: release. But when release-plz releases a single crate, it titles the PR with the crate as a scope, for example chore(tak-cli): release v0.0.12 (#133) or chore(tak-cli): release v0.0.11 (#100). Neither the daily schedule nor a manual gh workflow run auto-merge-release.yml (the early-release path in RELEASING.md) found them. On 2026-09-24, run 36034389411 logged No open release-plz release PR to main, and #133 had to be merged by hand.

Changes

• auto-merge-release.yml: the title filter is now test("^chore(\\([^)]+\\))?: release( |$)"). It accepts both chore(<crate>): release v<version> and the unscoped chore: release that release-plz uses when it releases several crates at once (chore: release #27). The other checks are unchanged: head branch prefix release-plz-, head owner jdx, base main, and a non-empty body.
• cliff.toml: the rule that keeps release commits out of the changelog now matches the same optional scope and requires a space or end of subject after release, like the PR-title filter (^chore(\([^)]+\))?: (release|Release)( |$)). A commit such as chore(ci): release-plz config therefore still appears in the changelog.

Verification

• I ran the new jq filter against every PR in the repo. It selects all past release PRs (chore(tak-cli): release v0.0.3 #14 through chore(tak-cli): release v0.0.12 #133, including chore: release #27's chore: release).
• It rejects near-misses such as chore: release-plz config, chore(ci): releases cleanup and fix: release foo.

AI-assisted — Tool: Claude Code; model: anthropic/claude-opus-5-5; version: unavailable.

🤖 Generated with Claude Code

---

Note

Low Risk
CI and changelog regex tweaks only; no runtime or auth behavior changes.

Overview
Fixes release auto-merge failing when release-plz opens a single-crate PR titled chore(<crate>): release v<version> instead of the unscoped chore: release.

The auto-merge-release.yml jq title filter now uses a regex that accepts an optional scope before release, while keeping the existing guards (branch prefix, head owner, non-empty body). cliff.toml uses the same pattern so scoped release commits stay excluded from the changelog.

^{Reviewed by Cursor Bugbot for commit 4a6b591. Bugbot is set up for automated code reviews on this repo. Configure here.}

[coderabbitai]

Warning

Review limit reached

Next included review available in 36 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Central YAML (base), Organization UI (inherited)

Review profile: CHILL

Plan: Advanced

Run ID: b63bd4c4-d8d4-4c12-8e9e-9af8f672d89a

📥 Commits

Reviewing files that changed from the base of the PR and between e065738 and 4a6b591.

📒 Files selected for processing (2)

• .github/workflows/auto-merge-release.yml
• cliff.toml

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[greptile-apps]

Confidence Score: 5/5

The PR appears safe to merge; no outstanding finding remains.

Summary

The PR updates release-PR selection to recognize scoped and unscoped release titles, and aligns the changelog skip rule with those titles.

• The change since the previous review adds a word boundary to the changelog rule, preserving maintenance commits such as chore(ci): release-plz config.

_{Reviews (2) · Last reviewed commit: "ci: require a word boundary after releas..."}

[github-actions]

Instruction counts

Nothing was compared, and so nothing was gated. No series appears on both sides: either the base has no measurements recorded, or the two were measured on different runner classes, which are deliberately not comparable — counts shift between machine types by more than a real regression does.

New, nothing to compare against: help on jdx-perf-v1-ubuntu24.04-x64-tak-gha20260920.314.1-rust1.98.1-img5263c143, version on jdx-perf-v1-ubuntu24.04-x64-tak-gha20260920.314.1-rust1.98.1-img5263c143

_{Only instruction counts gate. Wall clock is shown for context — on identical hardware it moves 4-20% run to run.}

_{Measured by tak — instruction-counted CLI benchmarks, stored in this repository's git notes.}

_{4a6b59192990 vs e065738428a3 · measured on the runner, not pushed to the history.}