agentleFS
SentryKnown publishergetsentry/sentry-godot/.agents/skills/build-ci-review/SKILL.md

build-ci-review

Skill260 starsChanged 5 days agoMITView on GitHub

Finds bugs in sentry-godot build, packaging, export, release, and CI changes, including GitHub Actions workflows, test runner scripts, SCons, Gradle, Craft, editor export plugins, and dependency version pins. Use for pull request review of .github/, scripts/, SConstruct, site_scons/, modules/ build files, Gradle files, src/editor/, and the .gdextension manifest.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Test runner scripts live in scripts/.

    line 17
  2. Very specific:

    When a PR bumps sentry-native (modules/sentry-native submodule), @sentry/browser (src/sentry/javascript/bridge/package.json), …

    line 39

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

You review build and CI changes for the Sentry SDK for Godot Engine. Report only changes that would break a build, a release, a user's project export, or a CI job, or make a job pass without testing what it claims to test.

Repo Layout

  • .github/workflows/ci.yml calls reusable workflows: build_gdextension.yml, unit_tests.yml, test_*.yml, package.yml, static_checks.yml. release.yml and .craft.yml drive releases; update-deps.yml bumps dependencies with scripts/*-version.ps1.
  • Test runner scripts live in scripts/. They run on Linux, macOS, and Windows runners, so paths, quoting, and process handling must work on all three.
  • SConstruct and site_scons/ build the GDExtension per platform and arch. modules/*.SConscript builds sentry-native and fetches sentry-cocoa; modules/*.properties pin downloaded versions.
  • android_lib/ (Gradle) builds the Android plugin AAR.
  • src/manifest.gdextension is the template SConstruct copies into the addon. It maps each platform, arch, and target to a library path under res://addons/sentry/bin/.
  • src/editor/ holds the editor plugin and per-platform export plugins (sentry_editor_export_plugin_*.cpp) that add libraries, dependencies, and settings when users export their projects. Changes here often go with changes to src/manifest.gdextension or SConstruct.

What To Report

Values that must stay in sync

A change updates one copy of a value and misses another, such as a version, a matrix entry, or an artifact name. Grep for the old value, and follow NOTE: comments that name other files to update.

Dependency bumps

When a PR bumps sentry-native (modules/sentry-native submodule), @sentry/browser (src/sentry/javascript/bridge/package.json), sentry-android, sentry-cocoa (modules/sentry-cocoa.properties), or sentry-dotnet (Sentry.Godot.Versions.props) across a major version, grep this repo's calls into that SDK (src/sentry/native/, sentry-bridge.ts, SentryAndroidGodotPlugin.kt, src/sentry/cocoa/, src/sentry/dotnet/managed/) for options or methods the new version removed or renamed. Report only removals that the PR, its changelog link, or code in this repo shows. Past real case: the v11 Web SDK bump left sendDefaultPii and scope.clear() calls in the bridge.

Script runtime bugs

  • PowerShell: unquoted arguments to Start-Process -ArgumentList or native commands when paths can contain spaces; Kill() or WaitForExit races that throw and abort a loop; parsing tool output (adb devices, xcrun) that also matches header or blank lines; exit codes that are dropped so a failed run reports success.
  • Bash: missing set -e or pipefail where a later step depends on success, unquoted variables, and checks that read the wrong process or component.

Workflow logic

  • Conditions, needs, or if: that skip a job on the event it is meant for, or let a required check pass while its work is skipped.
  • Outputs, artifact names, or paths referenced but never produced.
  • New third-party actions not pinned to a full commit SHA with a version comment, which is this repo's convention.

Build scripts

  • SCons options, defines, or source globs that leave a platform or arch without a backend, drop files from the package, or change output paths that package.yml, src/manifest.gdextension, or export plugins expect.
  • src/manifest.gdextension entries that don't agree with SConstruct. Some platforms and architectures are unsupported on purpose, so removing an entry is fine when the PR or commit explains it.

What Not To Report

  • Claims that a version, tag, or release does not exist, or that an action or tool option is unsupported. Releases often ship the same day as the bump, and you cannot check. Every past finding of this kind here was wrong.
  • Claims that a pattern syntax is wrong (regex vs glob, Sauce Labs descriptors) unless the consuming code in this repo shows which syntax it expects.
  • Security issues in workflows (handled by security-review).
  • Logic in SDK source or tests (handled by sdk-review and test-review), except calls broken by a dependency bump in this PR.

Severity

  • high: release or package job broken, shipped artifact missing a platform or file, or a required check that passes without running.
  • medium: a CI job or test runner fails or hangs in a reachable case, or one copy of a synced value drifts.
  • low: narrow failure on one runner or an uncommon local setup.

Finding Format

  • Title: the broken job or script and the trigger.
  • Description: one or two sentences on impact and the fix.
  • Verification: the file and line that consumes the changed value, and what happens on the failing runner or event.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.