agentleFS
wanggang316wanggang316/you-skill-ios/.agents/skills/release/SKILL.md

release

Skill2 starsChanged 15 days agoMITView on GitHub

Prepare, resume, submit, and verify a YouSkill iOS App Store release. Use for builds, store metadata or screenshots, review resubmission, and release tagging after successful review submission.

194 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use create-agent-adapter, changelog-style and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • MicrosoftSkill

    create-agent-adapter

    Technical guide for creating a new Paperclip agent adapter. Use when building a new adapter package, adding support for a new AI coding tool (e.g. a new CLI agent, API-based agent, or custom process), or when modifying the adapter system. Covers the required interfaces, module structure, registration points, and conventions derived from the existing claude-local and codex-local adapters.

    Also in 5 projects that use this

  • GrinvSkill

    changelog-style

    Turn a commit range into CHANGELOG.md entries following common-changelog style. Use when finishing a PR or cutting a release, to write/fix the [Unreleased] section's entries — moving [Unreleased] under a new dated version heading is a separate, automated step (see the release skill / scripts/sync-version.ts), not this skill's job.

    Also in 4 projects that use this

  • GrinvSkill

    docs-consistency-check

    Check README/manifest.json/server.json/CHANGELOG.md/AGENTS.md and docs/*.md for drift against the actual registered tools and source. Use after adding, renaming, or removing a tool, or as part of a live-audit pass.

    Also in 4 projects that use this

  • Microsoft AzureSkill

    component-release

    Prepares a verifiable Agent Landing Zone UI release. Use for release branches, VERSION, changelog entries, tags, release notes, and compatibility evidence; never publish without explicit human approval.

    Also in 4 projects that use this

  • Microsoft AzureSkill

    documentation-consistency

    Keeps AI Landing Zone contributor, user, and operator documentation aligned with shipped Bicep behavior. Use for parameters, defaults, outputs, topology, deployment flow, operations, or releases.

    Also in 8 projects that use this

  • rivet-devSkill

    launch-graphics

    Generate agentOS marketing/launch graphics (launch & social heroes, code-snippet cards) from the committed generators in docs-internal/graphics. Use when asked to make, update, or render an agentOS launch image, social image, "agentOS Apps" graphic, or technical code-snippet image.

    Also in 3 projects that use this

  • +3 more from Grinv

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 183 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before a long operation, inspect:

    line 14
  2. Distinctive:

    Every supported locale and intended device group is present.

    line 52
  3. Distinctive:

    After submission, read App Store Connect again and verify:

    line 79
  4. Distinctive:

    All intended locales and device groups, exact ordered filenames/counts, COMPLETE image states, and source checksums matching local images.

    line 82

“Common” is measured: kept by at least 6 in 10 of the 183 edited versions of this file we could read. “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

Release YouSkill iOS

Complete the requested release through the existing Fastlane workflow. Reuse verified work and report the actual App Store state. Upload, review submission, approval, public availability, and Git tagging are separate milestones. A tag does not publish this app; the current GitHub Actions workflows only run checks.

Establish the candidate and remote state

Read AGENTS.md, docs/references/ios-release.md, the relevant lanes in fastlane/Fastfile, and scripts/release-candidate.rb. These are the release contract. Keep reusable workflow logic there rather than creating a second release system. Use the existing execution plan for decisions, evidence, and pending work; do not treat ignored recovery scripts as maintained tooling.

Before a long operation, inspect:

  • Branch, HEAD, working tree, project.yml version/build, and existing tags.
  • build/release/manifest.json, its original source revision and hashes, and the IPA's embedded version/build.
  • Acceptance records, existing source captures, and fastlane/screenshot-selection.json.
  • App Store Connect version/build processing, selected build, localized metadata, screenshots and processing errors, review submission, and release mode.
  • Running uploaders and simulator suites. Do not start a duplicate process.

Use the repository Ruby version, Bundler, selected Xcode, and existing local configuration. A transferred checkout may contain artifacts but lack gems; restore the locked dependencies before running Fastlane. Required API variables are ASC_KEY_ID, ASC_ISSUER_ID, and ASC_KEY_PATH. Never print credentials or store the private key in Git.

Resume the next unfinished stage

Observed stateNext action
Prepared candidate, clean matching HEAD, valid assetsInspect selected images, then run bundle exec fastlane ios release.
Prepared candidate, later non-artifact commitsUse release reuse_prepared:true only when the existing gate accepts ancestry, unchanged app inputs, original clean acceptance, version/build, hashes, and screenshot inventory.
Metadata or images partly uploadedRead remote state and reconcile only missing or incorrect items through installed Fastlane/Spaceship dependencies.
IPA upload accepted or build processingCheck the exact version/build before any retry. Wait for processing; do not upload the same build again.
Build VALID, assets ready, no review requestedSelect the exact existing build and submit without binary or asset upload.
Review requested, no requested changesVerify selection and release mode, then create the submission tag if missing. Do not submit again.
Review requested, user asks to change screenshotsCancel that submission, wait until the version is editable, update the requested assets, then resubmit the same valid build.
No prepared candidateRun ./scripts/bump-build.sh (next build number above every App Store Connect build of the version), commit release inputs, and use prepare_release; inspect its output before submission.
Public release confirmedRecord public availability separately from the existing submission tag.

prepare_release runs acceptance, capture, and archive; it is not a resume command and rejects an existing capture directory. Preserve valid outputs. Repair only the failed stage, with a narrow tested lane change if required.

The release lane requires a clean checkout. By default HEAD must match the manifest. The explicit reuse path permits only the changes accepted by scripts/release-candidate.rb; it does not assert that newer tests ran. Preserve the original manifest revision and record the submission revision separately. Never rewrite provenance or check out an old commit to defeat the gate. Later documentation commits also advance HEAD.

Select screenshots before uploading

The capture inventory is evidence, not the upload selection. Do not upload every captured image just because it exists or passes its hash check.

Use 3–5 distinct core screens per language and device display group, with the same content and order across groups. Do not interpret this as a combined limit across all device groups. The current selection is categories, popular repositories, skills, and reader for both ordinary iPhone and expanded Duo, in each supported locale. Read the selection file for the actual filenames and groups rather than relying on a remembered total.

Use the largest supported ordinary iPhone screenshot size and Apple's scaling for smaller sizes where applicable. Preserve a distinct Duo group with equivalent screen coverage; one Duo image does not replace a full matching set. The regression device matrix does not require marketing images for every tested model or posture. Consult current Apple specifications if display-size acceptance is uncertain.

Validate and inspect the complete proposed selection before any remote replacement:

  • Every supported locale and intended device group is present.
  • Screen identities, count, and order agree across the matrix.
  • Images have accepted dimensions, opaque pixels, correct locale/UI, live content, and no blank or clipped content. Inspect Duo expanded layouts.
  • Selected files match original manifest hashes. Use the selection helper to create a separate upload directory; preserve unselected captures and the original manifest.

fastlane/screenshot-selection.json is the maintained selection. ReleaseCandidate.select_screenshots! validates it and produces a content-addressed upload directory. For a missing or bad image, repair that item rather than recapturing the whole matrix. Preserve supported locales and accurate accepted text unless the release requires a change.

Upload and recover without repeating successful work

The release lane uploads metadata and IPA, replaces selected screenshot sets, requests review, and sets automatic release. Understand all of those effects before invoking it. Use it for a fresh submission, not as a blind retry after partial success.

For the pinned Fastlane, ipa and build_number are mutually exclusive. For a new upload, pass the IPA and validate its embedded version/build; Deliver derives the build to wait for from the IPA. For an existing build, pass the exact build number, omit the IPA, and skip binary upload. Validate options before a remote mutation when changing invocation code.

Read remote state before a partial retry. Compare selected filenames, display groups, source checksums, processing states, and order. Retain matching COMPLETE images and upload only missing items. Remove only images outside the authorized selection. Do not delete accepted images merely because processing has not yet supplied a checksum. Treat UPLOAD_COMPLETE as pending processing, not an upload failure or completion of acceptance.

Concurrent screenshot requests on this host produced SSL_read / tls_retry_write_records failures. Check the partial result and retry the affected operation once with DELIVER_NUMBER_OF_THREADS=1. A TLS error is not proof of invalid screenshot dimensions. Do not restart preparation or repeat metadata/binary upload to recover a screenshot request.

Poll processing with a bounded wait and keep logs. Inspect the installed helper's behavior: the pinned beta screenshot sync has an unbounded exponential processing wait and can replace images whose checksums are not yet available. Prefer targeted reconciliation for a partial upload. Image and build processing can proceed independently; end the previous uploader before starting a separate missing stage, then gate review on both being ready.

Cancellation is also asynchronous. A cancellation response does not mean screenshots are editable. Wait for the old submission to leave progress and the version to become editable (for example, DEVELOPER_REJECTED) before changing images. Preserve the valid build and verified declarations during resubmission.

Submit and verify the result

A release request authorizes upload and review submission for the agreed candidate. A skill-edit or preparation-only request does not. User-requested screenshot corrections during review authorize cancellation and resubmission needed for that correction. Do not infer permission to merge, push tags, or change unrelated services.

Verify version/build, signing, privacy/support links, metadata, selection, and provenance. Preserve review contact, content-rights, and encryption declarations unless a verified change requires an update. Fix factual precheck issues from evidence: for example, add the release year while retaining the verified copyright owner. Do not invent compliance answers or dismiss a real warning because Fastlane allows it.

After submission, read App Store Connect again and verify:

  • Exact version/build, VALID processing, selected build, review state, and automatic release setting.
  • All intended locales and device groups, exact ordered filenames/counts, COMPLETE image states, and source checksums matching local images.
  • Intended descriptions, release notes, names, subtitles, and privacy/support URLs.

A zero exit status or successful upload alone is not enough. Once WAITING_FOR_REVIEW is confirmed, report that review is pending and automatic release is configured. Do not wait indefinitely for Apple's decision or claim public availability from submission.

Tag after successful review submission

Create an annotated tag after App Store Connect confirms successful review submission for the exact version and selected build, normally WAITING_FOR_REVIEW. Do not wait for approval or public availability. An accepted binary upload alone is not enough. The tag records a submitted candidate; it does not claim that Apple approved or published it.

Follow the repository's existing naming convention. Point the tag at the source revision that produced the submitted IPA, taken from the verified manifest, rather than an arbitrary later documentation or tooling commit. Include the marketing version, build number, and confirmed submission evidence in the annotation. Record later submission-tool revisions separately. Check existing local tags and relevant remote tags before creation. Tag creation does not by itself authorize pushing it or publishing a GitHub Release.

If review fails and code changes are needed, create and validate a new build, submit it, then create a new annotated tag for that candidate. Preserve previous tags. Use the new version tag when the version changes; when only the build changes, use a distinct build-qualified tag such as <version>-build.<build>. Do not move or overwrite a tag from an earlier submission. Screenshot or metadata corrections that reuse the same IPA do not require a new source tag; record the resubmission evidence in the execution plan.

Record the tag name, source revision, version/build, and review state in the execution plan. Continue to report review and public availability separately. Do not imply that background monitoring is active unless it was actually configured.

Completion and cost control

Keep successful outputs and logs. If the same operation fails again without new evidence, stop that retry loop and report the precise blocker and smallest next action. Do not erase simulators, broaden captures, or alter app behavior for an unrelated environment problem. Do not rerun acceptance without source changes, failures, or missing evidence.

Record durable outcomes in the existing execution plan after accounting for the clean-HEAD gate. Report version/build, screenshot matrix, upload/review/public state, tag state, and any blocker. Label local checks, simulator acceptance, remote verification, and public availability separately.

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.