agentsclimarketplace

Release new version

Skill fastrepl/anarlog/.agents/skills/release-new-version

Release a new desktop stable version for Anarlog. Use when asked to cut, publish, or prepare a new stable desktop release after checking and merging the changelog.From its SKILL.md

Install
npx -y skills add fastrepl/anarlog --skill release-new-version

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

8.7 KB, as published. Nobody here has run it

Release a New Desktop Version

Use this for stable desktop releases. A stable release must come from main, after the changelog for the computed version is present, accurate, validated, and merged.

Core Rule

Do not trigger a stable release from an unmerged branch. First make the changelog up to date, merge that changelog change to main, then run the stable release from main.

QA Gate

After the changelog is merged to main, read and run ../qa-critical-ux/SKILL.md against that final commit. Require a recorded PASS for both:

  • native Dev QA pinned with ANARLOG_QA_GIT_SHA and a helper manifest with git_dirty=false
  • the run-scoped staging artifact whose Actions head SHA exactly matches the Dev manifest's git_head_sha

Before QA, check Linear. While ANLG-98 is not completed, AEC quality is explicitly non-blocking. While ANLG-222 is not completed, automatic speaker identification is explicitly non-blocking. Do not delay a release, patch the candidate, or run dedicated fixture/provider matrices for those deferred areas. Core recording, audio capture, transcript and note persistence, manual speaker correction, crash/hang safety, and the remaining critical checklist still require PASS. Each deferred gate resumes only when its issue is completed.

The manifest's git_head_sha is the exact final main commit, including the changelog, not a synthetic GitButler workspace HEAD. The report must include that Dev SHA, staging run URL and head SHA, staging artifact SHA-256, and the applicable critical checklist results. Any applicable failure, missing evidence, SHA mismatch, rebuild, or later source change blocks stable unless the user explicitly waives that specific gate. The temporary ANLG-98 and ANLG-222 exclusions above are already authorized policy, not missing evidence.

This release path approves macOS, Windows, and Linux. Mobile remains closed. The patched CloudSync vendor bundle is rebuilt from source and cancellation-tested on every desktop lane: rebuild-macos.sh for Apple Silicon and Intel, rebuild-windows.sh under UCRT64 in windows_ci, and rebuild-linux.sh in linux_ci for x86_64 and aarch64. Each lane then runs cargo test -p cloudsync and cargo test -p db-core cloudsync:: against that freshly built library, covering the stalled-network, logout, configuration cleanup/init, worker-drain, and immediate-local-write cancellation gates.

The rebuild steps run only on workflow_dispatch, so a routine pull-request run does not prove them. Dispatch desktop_ci.yaml against the candidate SHA and confirm the cloudsync-windows-* and cloudsync-linux-* artifacts before treating a desktop lane as approved. Do not treat macOS artifacts or Rust-only tests as cross-platform approval, and do not open the mobile lane until its bundle gets the same treatment.

Preflight

  1. Inspect the workflow before assuming release behavior:
sed -n '1,280p' .github/workflows/desktop_cd.yaml
  1. Validate the explicit stable version requested by the user:
VERSION=<version>
[[ "$VERSION" =~ ^[0-9]+\.[0-9]+\.[0-9]+$ ]]
test -f "packages/changelog/content/$VERSION.md"

Stable desktop releases never infer a version. The workflow requires the exact stable semantic version and a matching changelog file.

  1. Identify the latest stable desktop tag and the commits that will ship:
git fetch --tags --force
git tag -l 'desktop_v*' --sort=-v:refname | grep -E '^desktop_v[0-9]+\.[0-9]+\.[0-9]+$' | head -n1
git log --oneline <latest-desktop-tag>..HEAD

Use read-only git commands for inspection. If the workspace is on gitbutler/workspace, use the but skill for commits, pushes, PRs, merges, and other write operations.

Changelog Gate

The changelog is the release gate. Before releasing:

  1. Open packages/changelog/content/AGENTS.md and follow its instructions.
  2. Confirm packages/changelog/content/<version>.md exists.
  3. Compare the file against the desktop user-facing changes since the latest desktop_v* tag.
  4. If the changelog is missing or incomplete, update it before release.

Changelog entries should be worth reading for app users. Exclude internal-only refactors, CI changes, infra noise, and implementation details unless they explain a user-visible change.

Each changelog file must include:

---
date: "YYYY-MM-DD"
summary: "One concise, user-facing sentence for the changelog index preview."
---

After editing the changelog, run:

pnpm exec dprint fmt
pnpm -F @anlg/changelog typecheck

Merge to Main

Only after the changelog is accurate and validation passes:

  1. Commit the changelog change.
  2. Open or update the changelog PR.
  3. Wait for CI and required review state to be clear.
  4. Merge the changelog PR to main.
  5. Verify main contains packages/changelog/content/<version>.md.
  6. Record the resulting main SHA and complete the QA Gate against that exact commit before triggering stable.

If using GitButler, prefer:

but diff
but commit chore/release-changelog -c -m "Update desktop release changelog

Refresh the desktop changelog for the next stable release." --changes <ids>
but pr new <branch-id> -t

Use actual IDs from but diff / but status -fv; do not invent IDs.

Trigger Stable Release

After the changelog merge and QA Gate pass on the same main SHA, verify main has not moved, then build the stable candidate without publishing:

gh workflow run desktop_cd.yaml \
  --ref main \
  -f channel=stable \
  -f publish=false \
  -f version=<version>

Watch the dry-run build:

gh run list --workflow desktop_cd.yaml --branch main --limit 5
gh run view <run-id> --json headSha,url
gh run watch <run-id>

The run's headSha must equal the Dev/staging candidate SHA. A mismatch blocks acceptance even if the workflow succeeds.

Do not use GitHub's rerun button for a failed stable candidate or Linux audio gate. Dispatch a fresh run instead; publication only accepts first-attempt run IDs so evidence cannot be mixed across attempts.

The dry-run workflow must:

  • use the exact explicit stable version
  • build both Apple Silicon and Intel macOS artifacts
  • build the signed Windows and Linux artifacts for the same version and commit
  • upload a draft CrabNebula release without publishing it
  • upload desktop-release-provenance-<version>-<sha>, including the exact artifact hashes and pinned CrabNebula CLI version, asset ID, and SHA-256

Run the Linux virtual-audio gate against that exact dry-run candidate:

gh workflow run desktop_linux_audio_qa.yaml \
  --ref main \
  -f version=<version> \
  -f candidate_sha=<40-character-main-sha> \
  -f dry_run_id=<dry-run-id>

Watch the run and verify its head SHA is still the candidate SHA. It must produce passing linux-audio-qa-<version>-x64 and linux-audio-qa-<version>-arm64 evidence artifacts. This gate proves virtual microphone/system routing, capture, separation, and persistence with NO_AEC=1. Physical-device AEC validation remains non-blocking while ANLG-98 is open and returns to the release gate when that issue is completed.

After the exact dry-run artifacts and Linux audio QA run pass the required platform gates and main still points to the candidate SHA, publish only through the provenance workflow:

gh workflow run desktop_publish.yaml \
  --ref main \
  -f version=<version> \
  -f candidate_sha=<40-character-main-sha> \
  -f dry_run_id=<dry-run-id> \
  -f audio_qa_run_id=<linux-audio-qa-run-id>

Watch that workflow to completion. It must verify the dry-run run identity, artifact hashes, both Linux audio evidence artifacts, CrabNebula tool identity and hash, current main, and the immutable tag before publishing. It must also verify every file mirrored to GitHub against the provenance manifest.

Final Checks

Before reporting success, capture:

  • computed stable version
  • dry-run workflow URL and head SHA
  • Linux audio QA workflow URL, head SHA, and x64/ARM64 evidence artifacts
  • publish workflow URL and head SHA
  • desktop_v<version> tag
  • GitHub release URL
  • whether CrabNebula publish completed
  • changelog URL
  • stable DMG SHA-256
  • installed stable critical-QA PASS

If the workflow fails, inspect the failed job logs with:

gh run view <run-id> --log-failed

After the workflow succeeds, follow the QA skill's post-publish stable gate: download the matching DMG from the GitHub release, install it, and use Computer Use to repeat the core checks. Do not declare the release complete until both the stable workflow and installed stable QA succeed.

Keep looking

Skills are one crate of 613,373. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.