ci: skip the desktop build on documentation-only changes

[ashutoshpw]

Summary

Skips the desktop build's heavy steps on documentation-only changes, inside the existing Check job.

A new API-only detection step runs at the top of Check and computes changed from the PR/compare file list. Only documentation and repository-metadata paths (docs/**, *.md, .github/ISSUE_TEMPLATE/**, .github/pr-assets/**, LICENSE, .gitignore) skip the desktop work; anything else — source, config, lockfile, workflow, script, native, or an empty/unresolvable list — runs it. Detection is fail-open.

Guarded steps (all otherwise byte-identical): setup-apt-mirrors, Install browser secret helper build libraries, Build desktop pipeline, Verify preload bundle output. Guard: success() && steps.detect.outputs.changed != 'false'; detection pre-writes changed=true and uses continue-on-error, so a hard detection failure still runs the build.

Why reworked (v1 → v2)

v1 split the desktop build into its own job. Same-repo measurements killed that: on this fork Test is ~424s and Check ~267s (main@99908f59f, run 35655900272), so Check is not the bottleneck, and the extra job added a runner, checkout, and full install for zero merge-wait benefit. v2 keeps the useful part — skipping heavy work when nothing relevant changed — at zero extra cost on code PRs.

Evidence

• Fork baseline (same commit as this branch's base): Check 267s (apt 13s + desktop build 57s at the tail), Test 424s.
• Synthetic path tests: docs-only / README / ISSUE_TEMPLATE skip; workflow, package.json, lockfile, script, native, empty list → run.
• This layer does not reduce merge wait; it removes wasted runner work on documentation-only PRs.

Stack

#9 (ci: batch short jobs) ← this PR: ci: skip the desktop build on documentation-only changes

Merge in order; after #9 merges, this PR is retargeted to main.

Manual Testing Guide

1. On a code-changing PR run: gh run view <id> --repo ashutoshpw/t2code --json jobs — confirm Check still runs checkout, PR-assets rejection, Setup Vite+, Electron setup, knip, vp check, typecheck, apt setup, browser-secret libs, desktop build, and preload verification.
2. On a docs-only PR (change docs/anything.md): confirm Detect desktop build changes logs Only documentation or repository metadata changed. and the four guarded steps are skipped, while the job stays green.
3. Confirm a code file (package.json, pnpm-lock.yaml, .github/workflows/ci.yml, scripts/*.ts, native/**) makes all four steps run.
4. Fail-open: in a log where detection errors, confirm changed=true is written and the build runs rather than being skipped.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
🔒 Security Review	✅ Completed 2026-09-22T04:54:49.893007Z	04e4be3	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	13.5 KiB	—	15.1 KiB	✅
Codex	Thread snapshot wire	—	7.0 KiB	—	7.3 KiB	✅
Codex	Live turn WebSocket wire	—	6.5 KiB	—	7.8 KiB	✅
Codex	Live turn WebSocket decoded	—	56.3 KiB	—	66.4 KiB	✅
Codex	Live turn messages	—	10	—	21	✅
Claude	Total thread wire	—	13.5 KiB	—	15.1 KiB	✅
Claude	Thread snapshot wire	—	7.1 KiB	—	7.3 KiB	✅
Claude	Live turn WebSocket wire	—	6.5 KiB	—	7.8 KiB	✅
Claude	Live turn WebSocket decoded	—	57.0 KiB	—	66.4 KiB	✅
Claude	Live turn messages	—	9	—	21	✅

Baseline: unavailable · PR result: 0ec7827 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 113.9 KiB
• Claude decoded thread snapshot: 114.6 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.