Release New Version logo

Release New Version

Community
niki914
release-new-version

Use when the user wants to release a new Zafiro version — drafting bilingual release notes, deciding the next version number, bumping app/build.gradle.kts, tagging, and publishing to GitHub (main repo, optionally the Xposed repo).

Overview

Publisherniki914
Repositoryzafiro
Skill namerelease-new-version
Stars
305
Forks
50
Bundled files
2
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.

  • 2 bundled files

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

  • Open source

    Published by niki914 on GitHub. Read the source before you install it.

Installation

Install the Release New Version 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/niki914/zafiro/tree/main/.agents/skills/release-new-version
  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/niki914/zafiro.git /tmp/zafiro
mkdir -p .claude/skills
cp -r /tmp/zafiro/.agents/skills/release-new-version .claude/skills/release-new-version
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Release New Version 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 Release New Version 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 Release New Version 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.

Release New Version

Release workflow for Zafiro.

Repos involved:

RepoTag formatRelease notes
niki914/zafiro (main, open source)v<code>-<name> e.g. v8-1.1.0Full bilingual notes
Xposed-Modules-Repo/com.niki914.zafiro (closed)<code>-<name> e.g. 8-1.1.0Same feature entries, closing line points back to the main repo. Optional — ask the user every time.

CI builds and signs one APK when a tag is pushed to the main repo, stamps channel github into its APK Signing Block, and uploads that file as the GitHub Release asset. The Xposed repo has no CI. Its release is a manual gh operation: download that asset, stamp channel xposed over it, and upload the result. The signature and certificate stay the same. Never build locally.

Phase 1 — Gather facts

Collect all of this before asking the user anything:

  1. Local version: read versionCode / versionName from app/build.gradle.kts.
  2. Latest release baseline: gh release view -R niki914/zafiro (returns the latest non-draft, non-prerelease release by default). Parse its tag v<code>-<name>.
  3. Version string locations: find every place the current version literal lives:
    bash
    rg -n '<current versionName>' -g '*.{md,kt,kts,txt,py}' .
    Only scan these extensions: md, kt, kts, txt, py. Nothing else. rg respects .gitignore by default — do not disable that.
  4. Changes since last release: commits and merged PRs from the last release tag to origin/main:
    bash
    git log v<code>-<name>..origin/main --oneline --no-merges
    gh pr list -R niki914/zafiro --state merged --limit 50 --json number,title --jq '.[] | "\(.number) \(.title)"'
  5. If local versionCode/versionName do not match the latest release tag: stop and ask the user how to proceed. Do not guess.

Summarize the changes for the user in plain language before drafting.

Phase 2 — Draft release notes (iterate until explicit approval)

Write bilingual notes (Chinese first, ---, then English) from the change list. Apply these rules strictly — they are derived from every historical release:

  1. Only user-perceivable changes. New features, visible improvements, refactors with user-facing impact. Never list chores, CI fixes, string cleanups, dependency bumps, internal renames.
  2. Bug fixes are always vague. Collapse all fixes into one line: "修复了一些问题" / "Fixed several issues". Optionally append one phrase naming the improved area (e.g. "优化了 Phone Use 执行效率"). Never enumerate specific bugs.
  3. 4–8 numbered entries. Feature names are brand terms (Phone Use, Skills System) — keep them consistent with past usage.
  4. Breaking changes get a blockquote before the list, in both languages, telling users what to do (e.g. the 1.1.0 package-name rename note told users to set up the new version as a fresh start).
  5. Milestone items (e.g. "仓库已开源,欢迎 Star、提 Issue、参与贡献") go as the final entry when applicable.

Present the draft. The user will revise it — redraft and repeat until the user explicitly approves. Approval is an explicit statement from the user, not silence.

Few-shot (real entries from past releases):

  • Good feature entry: 1. 新增 Phone Use:支持自动点按、滑动、滚动、输入文字、打开 App
  • Good vague fix entry: 7. 修复了一些问题
  • Bad entry (do not write): 修复了 composer 光标在深色主题下的偏移 — too specific; fold into "修复了一些问题"
  • Bad entry (do not write): chore: remove 13 unreferenced UI strings — not user-perceivable, drop entirely

Phase 3 — Propose version number (iterate until explicit approval)

Semver x.y.z:

  • Pump X — proud version: milestone, rebranding, headline feature, open-source moment.
  • Pump Y — normal update: regular features.
  • Pump Z — shamed version: fixing a severe/P0 bug.

Propose one number with a one-line justification referencing the change list. Also report every location found in Phase 1 step 3 that needs the version string updated. Iterate with the user until explicit approval of both the number and the file list.

Phase 4 — Execute

Confirm once more before anything irreversible (exact version number, tag, target commit, and whether to publish to Xposed).

Save the approved release notes into a temporary file (e.g. /tmp/zafiro-release-notes.md).

Automated Execution (Preferred)

Run the release execution script from repo root:

bash
# Main repo only:
bash .agents/skills/release-new-version/scripts/release-exec.sh <new code> <new name> /tmp/zafiro-release-notes.md

# Main repo + Xposed repo:
bash .agents/skills/release-new-version/scripts/release-exec.sh <new code> <new name> /tmp/zafiro-release-notes.md --xposed

The script automates:

  1. Re-counting code lines badge (update-code-badge.sh).
  2. Updating versionCode and versionName in app/build.gradle.kts.
  3. Committing release: bump to <new name> and pushing to origin main.
  4. Creating tag v<code>-<name> and pushing to origin.
  5. Waiting for GitHub Actions release build CI to succeed (gh run watch).
  6. Updating the GitHub Release title to Release - <name> and replacing CI notes with approved bilingual notes.
  7. If --xposed is given: downloading the GitHub Release APK, stamping channel xposed (tools/apk_channel.py stamp, which checks read-back and apksigner verify), generating Xposed notes (with open-source links), and creating the Xposed release with --latest on Xposed-Modules-Repo/com.niki914.zafiro.

Manual Fallback Steps

If running manually without the script:

  1. Update the code-line badge so it lands in the release commit:
    bash
    bash .agents/skills/release-new-version/scripts/update-code-badge.sh
  2. Update versionCode = <new code>, versionName = "<new name>" in app/build.gradle.kts.
  3. Update module structure in both READMEs (README.md and README_CN.md) and AGENTS.md if applicable.
  4. Audit and complete multilingual string parity across all modules.
  5. Commit: release: bump to <new name>, push to origin main.
  6. Tag and push:
    bash
    git tag v<code>-<name>
    git push origin v<code>-<name>
  7. Wait for CI:
    bash
    gh run watch <run-id> -R niki914/zafiro --exit-status
  8. Replace CI notes:
    bash
    gh release edit v<code>-<name> -R niki914/zafiro --title "Release - <name>" --notes "<approved notes>"
  9. (Optional) Xposed repo. Stamp xposed on the downloaded asset before uploading. stamp exits non-zero unless the channel reads back and apksigner verify passes:
    bash
    mkdir -p /tmp/zafiro-xposed
    gh release download v<code>-<name> -R niki914/zafiro -p "*.apk" --dir /tmp/zafiro-xposed/
    APK="$(echo /tmp/zafiro-xposed/*.apk)"
    python3 tools/apk_channel.py stamp "$APK" "$APK" xposed
    gh release create <code>-<name> "$APK" -R Xposed-Modules-Repo/com.niki914.zafiro --title "Release - <name>" --notes "<approved notes>" --latest

Release notes format

Historical releases follow this format strictly (no deviations):

  • English first, Chinese after, separated by a single --- line. Never reverse the order.
  • No language headers (no **English** / **中文** labels) — the notes start directly with the numbered list. Blank line separates the --- from the lists.
  • The --title "Release - <name>" is part of the format, not optional. Set the title via --title, never inside the release note body.

Phase 5 — Xposed repo (optional, ask first)

Ask: "这次要发 Xposed 仓库吗?" If yes, pass --xposed to release-exec.sh or run step 9 manually. The file uploaded there is the xposed stamp, not the GitHub asset as-is. If no, run without --xposed and state clearly "本次未发 Xposed 仓库". The GitHub Release asset still carries channel github.

Known pitfalls

  • CI build failure: Aliyun Maven mirror 502. Overseas GitHub Actions runners intermittently fail on the Aliyun mirror if it is ordered before mavenCentral() / gradlePluginPortal() in settings.gradle.kts. Standard sources go first, Aliyun after. If CI fails on dependency resolution, check the mirror order.
  • Channel stamp rejected. tools/apk_channel.py stamp exits non-zero when apksigner is missing, the APK has no v2/v3 signing block, apksigner verify fails, the signer certificate changes, or the channel does not read back. apksigner is build-tools/<version>/apksigner under ANDROID_HOME, or the APKSIGNER path. CI has it after the release build. The Xposed stamp runs on the machine that runs this skill, so that machine needs it too. Do not upload a rejected file. Re-run the release CI; leave the keystore alone.
  • Tag already exists / CI failed and needs a re-trigger. The same tag cannot be pushed twice:
    bash
    git push origin :v<code>-<name>   # delete remote tag
    git tag -d v<code>-<name>         # delete local tag
    git tag v<code>-<name>            # re-tag at latest commit
    git push origin v<code>-<name>

Hard rules

  • Never build locally. CI builds everything.
  • Stamp a distribution channel only with tools/apk_channel.py stamp. CI stamps github on the GitHub Release asset. The Xposed publish path stamps xposed. Names match [a-z0-9_]{1,32}. The stamp does not use the keystore.
  • Never modify versionName without explicit user approval.
  • Never push a tag without showing the user the exact tag and target commit first.
  • If any state is ambiguous (local ≠ remote, tag exists, CI red), stop and ask.

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 Release New Version AI skill do?

Use when the user wants to release a new Zafiro version — drafting bilingual release notes, deciding the next version number, bumping app/build.gradle.kts, tagging, and publishing to GitHub (main repo, optionally the Xposed repo).

Why use Release New Version on TypingMind?

Because you install it once and use it with any model. Release New Version 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 Release New Version in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/niki914/zafiro/tree/main/.agents/skills/release-new-version. 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 Release New Version?

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 Release New Version?

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

Is the Release New Version AI skill free?

Yes. It is published on GitHub by niki914 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 👇