Suede Play Release logo

Suede Play Release

Community
JasonColapietro
suede-play-release

Suede AI Google Play Release Skills. Google Play release delivery with AAB preflight, test-track upload, exact versionCode promotion, staged rollout, locale notes, and API readback. Use when asked "promote this Android build" or to diagnose Play publishing access and prove the served release. NOT FOR: app building or policy review (android-app-factory), listing assets (suede-aso), or CI gates (suede-ci-gate).

Overview

PublisherJasonColapietro
Repositorysuede-creator-skills
Skill namesuede-play-release
Stars
126
Forks
12
Bundled files
4
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 4 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by JasonColapietro on GitHub. Read the source before you install it.

Installation

Install the Suede Play Release AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/JasonColapietro/suede-creator-skills/tree/main/skills/suede-play-release
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/JasonColapietro/suede-creator-skills.git /tmp/suede-creator-skills
mkdir -p .claude/skills
cp -r /tmp/suede-creator-skills/skills/suede-play-release .claude/skills/suede-play-release
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Suede Play Release in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Suede Play Release on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Suede Play Release is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Suede AI Google Play Release Skills

Compatibility

Requires Android release artifacts and authorized Google Play publishing access. Bundled preflight uses Python with pyjwt/cryptography and macOS Keychain; delivery examples use fastlane supply. Linux CI uses its secret store.

Use the worked contract for a staged rollout example, access-failure recovery, and API readback validation.

text
Iron Law: the Play API decides what shipped, not a green fastlane summary.
Every claim about a live release is a readback or it is a guess.

Gate policy: advisory, not blocking

Run every check and report results honestly. Verdicts are advice attached to the work, never a control that changes it. A failed check changes what you report, never what you do. The one exception is the production halt in Step 7: a public release is irreversible for the users who get it, so present the options and let the user pick.

Step 0: Preflight the credential before anything else

Never assume publishing works. Prove it with a real API call that changes nothing: open an edit and delete it. The script reads the key from the macOS login Keychain and needs pyjwt and cryptography; on Linux CI, load the same base64 JSON from the CI secret store and run the equivalent token exchange.

bash
python3 scripts/play-preflight.py <package.name> [--service NAME] [--account NAME]

Three outcomes, three different actions:

ResultMeaningDo
token failskey is wrong or revokedrotate, see Step 1
HTTP 401/403 on the editPlay Console has not granted the accountStep 1's grant
edit opens and deletespublishing is wiredcontinue

A 403 here is the normal state of a brand-new service account, not a bug.

Step 1: Credentials

Use a dedicated publishing service account, separate from any billing or purchase-verification account. A credential that can push releases should not also be able to void purchases. Give it no cloud IAM roles: all of its authority comes from the Play Console grant, so before that grant the key is inert.

Store the JSON in the login Keychain base64-encoded:

bash
security add-generic-password -U -s GOOGLE_PLAY_SERVICE_ACCOUNT_JSON -a <account> -w

Run it with -w last and no value so it prompts and the key never enters shell history. Paste the base64, not the raw JSON.

Base64 is load-bearing. security hex-encodes any multi-line secret on read, so raw JSON comes back in one of two shapes depending on content. Base64 gives one shape and one decode path.

Granting access is a Play Console UI action and cannot be done from the API: invite the service-account email under Users and permissions with release permissions on the app. Report this to the user as a step only they can take, with the exact email and permissions, then re-run Step 0.

Step 2: Pull before every command

fastlane reads the working tree, not the remote. A stale checkout produces a run that succeeds and does nothing, because a missing changelog is a skip rather than an error.

bash
git -C <repo> pull --ff-only && git -C <repo> log --oneline -1

If fastlane offers to set itself up, the Fastfile is not on disk. Answer no and pull; accepting scaffolds an empty config over the real one.

Step 3: Write the changelog first

Release notes live at fastlane/metadata/android/<locale>/changelogs/<versionCode>.txt, one file per locale per versionCode.

A missing file is not an error. Play silently serves the previous version's notes, so users read stale text against a new build and nothing tells you. Write one for every locale that has listing copy, before uploading.

bash
for l in $(ls fastlane/metadata/android); do
  [ -f "fastlane/metadata/android/$l/changelogs/$VC.txt" ] || echo "MISSING: $l"
done

Match each locale's existing register rather than translating the English word for word. Check the limit: 500 characters.

Step 4: Verify the artifact, never the source tree

A source read is not evidence that a change reached the binary.

bash
jarsigner -verify app/build/outputs/bundle/release/app-release.aab | head -1
grep -oE 'android:version(Code|Name)="[^"]*"' \
  app/build/intermediates/merged_manifests/release/*/AndroidManifest.xml

When the release changes an asset, hash it on both sides and require a match:

bash
unzip -p <aab> 'base/res/*xxxhdpi*ic_launcher.png' | shasum -a 256
shasum -a 256 app/src/main/res/mipmap-xxxhdpi/ic_launcher.png

jarsigner reporting a self-signed certificate is correct under Play App Signing, which re-signs with the real release key.

Step 5: Upload to a testing track first

bash
fastlane android internal

Build uploads carry the binary and its changelogs only. Listing text, icon, feature graphic and screenshots move through a separate lane, so a routine upload cannot rewrite what the store says about the app.

Step 6: Promote, never re-upload

Play rejects a second bundle carrying a versionCode it already has. Re-uploading is not a slower path to the same place, it is an error.

Promote the versionCode already on the testing track, which is also the only thing that makes testing-first mean anything: the artifact reaching production is the one that was validated, bit for bit.

ruby
upload_to_play_store(
  version_code: <code>, track: "internal", track_promote_to: "production",
  skip_upload_aab: true, skip_upload_apk: true,
  release_status: "inProgress", rollout: "0.1"
)

Step 7: Production halt

A production rollout is irreversible for the users who receive it. Stop and present:

text
Ready to promote versionCode <n> (<name>) to production at <pct>% of users.
Currently live: versionCode <m>. <one line on what users will notice>.
  1. Promote at <pct>%
  2. Promote at a different fraction
  3. Upload to a testing track only
  4. Hold

Then wait. Proceed on an explicit answer, and treat a standing instruction to ship as that answer.

Step 8: Read back what is actually live

The fastlane summary reports that the request succeeded, not what the release became. Parameters like track_promote_release_status can differ from the release_status you set, so read the track:

GET /androidpublisher/v3/applications/<pkg>/edits/<id>/tracks/production

Check three fields and say them plainly:

  • versionCodes: the build users get.
  • status: inProgress is a staged rollout; completed is everyone.
  • userFraction: absent when completed.

A 100% rollout should land as status: completed with no userFraction. A release sitting at inProgress with userFraction: 1.0 is a different state that reads the same in a success message. Confirm which one you got.

Also confirm every locale you wrote appears in that release's releaseNotes.

Done means

Every line proven by a command in this run:

  • Preflight opened and deleted a real Play edit.
  • The AAB verified: signature, versionCode in the merged manifest, and a hash match for any changed asset.
  • The track readback shows the expected versionCode, status, and fraction.
  • Every locale with listing copy appears in the live release's notes.

Boundaries

  1. Do not promote to production, raise a rollout, or halt one without an explicit instruction covering that action.
  2. Do not create the Play Console grant yourself; it is a UI action. Name the email and permissions and hand it to the user.
  3. Do not print, commit, or copy the service-account JSON, the upload keystore, or keystore.properties into a repo.
  4. Do not push listing text, images, or screenshots as part of a build upload. Store copy moves in its own deliberate step.
  5. Do not report a release state from a fastlane summary. Read the track.
  6. Do not reuse a billing or purchase-verification service account for publishing.
  7. Do not edit a changelog for a versionCode already serving users to fix a typo silently; that changes live store copy and belongs in the same confirmation as a rollout change.

Routing

  • Need to plan, build, test, or policy-check the app itself -> use android-app-factory, then return here for delivery.
  • Need listing text, keywords, screenshots, or ratings work -> use suede-aso. Listing changes travel in their own deliberate step (Boundary 4), so plan them there and ship them separately from the binary.
  • Need CI wiring, required checks, or merge gates around the release -> use suede-ci-gate.
  • Need branch, worktree, or stale-mirror handling for the release branch, or the iOS half of the same release -> private Suede AI companions, not in this pack.
  • Need constants to match across the Android, iOS and web surfaces -> use suede-parity-contract.
  • From android-app-factory: route Play credential setup, upload, track promotion, staged rollout, and live-state verification back here.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Suede Play Release AI skill do?

Suede AI Google Play Release Skills. Google Play release delivery with AAB preflight, test-track upload, exact versionCode promotion, staged rollout, locale notes, and API readback. Use when asked "promote this Android build" or to diagnose Play publishing access and prove the served release. NOT FOR: app building or policy review (android-app-factory), listing assets (suede-aso), or CI gates (suede-ci-gate).

Why use Suede Play Release on TypingMind?

Because you install it once and use it with any model. Suede Play Release is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Suede Play Release in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/JasonColapietro/suede-creator-skills/tree/main/skills/suede-play-release. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Suede Play Release?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Suede Play Release?

As many as you like. As long as a model supports skills, you can use Suede Play Release with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Suede Play Release AI skill free?

Yes. It is published on GitHub by JasonColapietro under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇