🐛 Fix translation review never triggering on auto-labelled PRs

[9larsons]

no ref

Problem

The Translation Review workflow (translation-review.yml) triggered only on pull_request_target labeled / synchronize, gated on the affects:i18n label.

That label is applied automatically by the label-actions workflow using the default GITHUB_TOKEN. GitHub deliberately does not start new workflow runs from events triggered by GITHUB_TOKEN, so the labeled path has produced zero runs since the workflow shipped.

Reviews still appeared because i18n PRs almost always get a follow-up push (a translator iterating, or a "Merge branch 'main'" commit) which raises a synchronize event — and that's what actually ran the job. The bug stayed hidden until a PR was opened, auto-labelled, and never pushed to again (e.g. #28462), which got no review at all.

Evidence (PR #28434)

• 10:23:49 affects:i18n added by github-actions[bot] → no run
• 13:32:07 push "Merge branch 'main'" → synchronize → run 13:32:14 → review 13:32:37
• 14:24:38 push "Merge branch 'main'" → synchronize → run 14:24:45 → review 14:25:05

Every run that produced a review was synchronize; none fired at label-time. Same pattern across every i18n PR sampled.

Fix

Trigger on real PR-author events (opened / reopened / synchronize) gated by a paths filter on ghost/i18n/locales/**, instead of relying on the bot-applied label. These events come from the PR author (not GITHUB_TOKEN), so they aren't suppressed. labeled is kept only as a manual maintainer re-trigger (remove + re-add the label).

The review script remains the precise final gate — it only reviews ghost/i18n/locales/<locale>/<namespace>.json (excluding en/) and exits cheaply otherwise, so the broader paths glob costs at most a no-op checkout.

No change to the security posture: still pull_request_target checking out main (never the PR head), reading the diff via API only, with the existing file/line/timeout abuse guards intact.

SETUP.md updated to document the new triggers and the GITHUB_TOKEN rationale.

[coderabbitai]

Walkthrough

The PR updates the Translation Review workflow to shift from label-driven triggering (affects:i18n label) to path-based triggering on changes to ghost/i18n/locales/** files. The workflow now triggers on PR author events (opened, reopened, synchronize) and manual re-runs via labeled, with a refined job condition that allows labeled events only when the label is exactly affects:i18n. Supporting documentation in SETUP.md explains the rationale for avoiding label-only triggering, clarifies the script's file-level filtering behavior, and updates guidance for retroactively reviewing existing labeled PRs.

Possibly related PRs

• TryGhost/Ghost#27966: Added the translation review workflow and scripts that this PR now refines by switching from label-driven to path-based triggering.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	Title clearly identifies the bug being fixed and is directly related to the changeset, which updates the translation review workflow trigger mechanism.
Description check	✅ Passed	Description comprehensively explains the problem, the root cause involving GITHUB_TOKEN suppression, the evidence of the bug, and the proposed fix with security considerations.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/i18n-review-trigger

---

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 and usage tips.}

[coderabbitai]

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

.github/scripts/i18n-review/SETUP.md (1)

4-5: ⚠️ Potential issue | 🟡 Minor | ⚡ Quick win

Update stale label-driven wording in setup docs.

Line 4 and Line 144 still describe label-based behavior, but the workflow now primarily keys off paths + PR-author events. Renaming affects:i18n no longer disables normal runs, and the opening description should reflect path-based triggering.

Suggested doc patch

- that posts an advisory (non-approving) GitHub review on PRs labelled `affects:i18n`.
+ that posts an advisory (non-approving) GitHub review on PRs that touch locale files under `ghost/i18n/locales/**`.

- - Or simply rename the `affects:i18n` label — the conditional won't match
+ - Or comment out/disable the workflow trigger block in `translation-review.yml` (label renaming only affects the manual `labeled` re-trigger path)

Also applies to: 144-145

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In @.github/scripts/i18n-review/SETUP.md around lines 4 - 5, Update the stale
label-driven wording that claims runs are gated by the `affects:i18n` label and
that renaming it disables normal runs; instead state that the workflow is
triggered primarily by changed paths and PR-author events (e.g., path-based
triggers) and no longer relies on label presence to allow runs. Search for
occurrences of the literal label `affects:i18n` and the opening
description/top-level summary and the later paragraph that mentions disabling
runs, then replace those sentences to explain path-based triggering and that
renaming the label does not stop workflows from running. Ensure the new wording
briefly describes "paths" + PR-author events as the trigger mechanism and
removes any implication that removing/renaming the label disables workflow
execution.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Outside diff comments:
In @.github/scripts/i18n-review/SETUP.md:
- Around line 4-5: Update the stale label-driven wording that claims runs are
gated by the `affects:i18n` label and that renaming it disables normal runs;
instead state that the workflow is triggered primarily by changed paths and
PR-author events (e.g., path-based triggers) and no longer relies on label
presence to allow runs. Search for occurrences of the literal label
`affects:i18n` and the opening description/top-level summary and the later
paragraph that mentions disabling runs, then replace those sentences to explain
path-based triggering and that renaming the label does not stop workflows from
running. Ensure the new wording briefly describes "paths" + PR-author events as
the trigger mechanism and removes any implication that removing/renaming the
label disables workflow execution.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 31b03982-61e7-474a-b851-3ea01f4d607b

📥 Commits

Reviewing files that changed from the base of the PR and between 31c1252 and 8bd25f2.

📒 Files selected for processing (2)

• .github/scripts/i18n-review/SETUP.md
• .github/workflows/translation-review.yml

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 73.74%. Comparing base (31c1252) to head (8bd25f2).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main   #28465   +/-   ##
=======================================
  Coverage   73.74%   73.74%           
=======================================
  Files        1541     1541           
  Lines      132184   132184           
  Branches    15784    15784           
=======================================
  Hits        97484    97484           
  Misses      33733    33733           
  Partials      967      967           

Flag	Coverage Δ
admin-tests	54.67% <ø> (ø)
e2e-tests	73.74% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.