Create Release logo

Create Release

Community
richkuo
create-release

Use when the user asks to "create a release", "cut a release", "tag a version", "publish release notes", or ship a new versioned GitHub release for the current repo.

Overview

Publisherrichkuo
Repositoryrk-skills
Skill namecreate-release
Stars
50
Forks
8
Bundled files
Instructions only
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.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

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

Installation

Install the Create 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/richkuo/rk-skills/tree/main/skills/create-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/richkuo/rk-skills.git /tmp/rk-skills
mkdir -p .claude/skills
cp -r /tmp/rk-skills/skills/create-release .claude/skills/create-release
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Create 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 Create 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 Create 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.

create-release

Tag a SHA on origin/<default> (annotated semver) and publish a GitHub release with generated notes. The local branch never decides what ships. Never push to the default branch: the version bump and notes entry land through a pull request (PR).

Preconditions: verify ALL before doing anything

bash
gh auth status
git status --porcelain -- . ':(exclude).claude/worktrees'
DEFAULT=$(gh repo view --json defaultBranchRef -q .defaultBranchRef.name)
git fetch origin --tags

gh auth status must succeed. git status --porcelain must print nothing; if the tree is dirty, stop and surface the diff. Every later read uses origin/$DEFAULT. Any failure: stop and tell the user.

Tag-target invariant. The tagged commit must pass git merge-base --is-ancestor <sha> "origin/$DEFAULT" right after a fresh git fetch origin. No confirmation, flag, or caller instruction overrides this. On failure, stop and report the SHA and origin/$DEFAULT.

Steps

  1. Inspect actual tags. Never trust memory or CHANGELOG: git tag --sort=-v:refname | head -10.

  2. Review commits since the last tag: git log <last-tag>..origin/$DEFAULT --oneline; with no semver tag (first release), git log origin/$DEFAULT --oneline.

  3. Determine the semver bump and state the rationale. Breaking = major (never patch), feature = minor, fixes and polish = patch. Proceed without confirmation: the only question is the bump PR merge. Ask about the bump only when its type is genuinely ambiguous (for example, unclear whether commits are breaking). Then git ls-remote --tags origin refs/tags/vX.Y.Z must print nothing. Never force-overwrite a tag that exists locally or on origin: ask the user.

    Previous version. The newest semver tag without its v. With no semver tag, the value all table manifests below agree on; if they disagree or none exists, ask the user (AskUserQuestion) for the first version and treat each manifest version field's value as its previous value. A first release takes X.Y.Z from the user's version, else from that agreed value.

  4. Land the version bump and the release-notes entry through a PR.

    • Resume an interrupted release first. List bump PRs: gh pr list --state all --search 'in:title "chore(release): bump version to"' --json number,title,state,isCrossRepository,headRefName,baseRefName,headRefOid,mergeCommit,files --limit 50. A candidate meets all of: its title names a version V above the previous version (or, with no semver tag, it is merged and V equals the manifests' agreed value: an interrupted first release), and git ls-remote --tags origin refs/tags/vV prints nothing; open or merged, never closed unmerged; isCrossRepository false; headRefName is <harness-prefix>/release-vV (cc/, cursor/, or codex/), optionally suffixed -<n>; baseRefName is $DEFAULT; every path in files is a table manifest or the notes file, and gh pr diff <n> only moves those version fields from their pre-PR values to V and adds at most one notes entry. A title match that fails any check is never resumed or merged: report it and continue with the version-fields bullet. No candidate: continue with that bullet. More than one: stop and report them. One: if V differs from the version the user asked for or step 3 chose, ask the user (AskUserQuestion); never override their version yourself. X.Y.Z is then V, and step 3's tag-free check applies to V. Open: reuse it (edit nothing, open no second PR), record its headRefOid as the head the checks read, and go to the Merge bullet, which always asks about a resumed PR: no caller plan authorization covers a PR this run did not open. Merged: go to the Resolve bullet, which checks the invariant and manifests before step 5. Either way, reuse the worktree named for its headRefName if it still exists; the Resolve bullet removes it.

    • Version fields. Change only manifest version fields whose value equals the previous version: grep it (for example git grep -n -F "<previous>" origin/$DEFAULT -- <manifest files>) and check each hit against the table. Libraries, CLIs, and installers count: a package.json "version" must be bumped. Schema and config versions stay (for example .github/dependabot.yml version: 2), and so do native build numbers (buildNumber, versionCode, CURRENT_PROJECT_VERSION) unless the user asks. With a semver tag, a manifest field holding another version, with no bump PR kept by the resume bullet, is a stop: report the file and value.

      FileField
      package.json"version"
      .claude-plugin/plugin.json"version"
      app.json / app.config.js / app.config.tsexpo.version
      pubspec.yamlversion
      Cargo.toml[package] version
      pyproject.toml[project] version or [tool.poetry] version
      iOS *.xcodeproj/project.pbxprojMARKETING_VERSION
      Android build.gradleversionName
    • Release-notes entry. If the repo has a MEMORY.md or CHANGELOG.md with a Releases section, the same PR appends **vX.Y.Z** — YYYY-MM-DD: <one-line summary>, unless that file on origin/$DEFAULT already holds a **vX.Y.Z** entry.

    • Publish-on-release check (every time). Find .github/workflows/ jobs on release: published (for example npm publish): they usually guard on the manifest version and silently skip while package.json holds the previous version, so a tag and manifest mismatch is a bug.

    • No version field needs a change and no notes entry is due (no manifest, or a first release whose manifests already hold X.Y.Z; and no notes file, or one already holding **vX.Y.Z**): skip the PR, git fetch origin, set SHA=$(git rev-parse "origin/$DEFAULT"), and go to step 5.

    • Otherwise open the PR from a worktree off origin/$DEFAULT, on a branch with the harness prefix (cc/ Claude Code, cursor/ Cursor, codex/ Codex): git worktree add .claude/worktrees/cc+release-vX.Y.Z -b cc/release-vX.Y.Z "origin/$DEFAULT". If that branch exists locally or on origin, never delete or reuse it: add the next free suffix to branch and worktree name (cc/release-vX.Y.Z-2, then -3). Edit there, run every git command with -C <worktree>, and stage changed files by name. Commit chore(release): bump version to X.Y.Z with the repo's commit footer (its CLAUDE.md or AGENTS.md attribution rule), push, and run gh pr create --base "$DEFAULT", titled chore(release): bump version to X.Y.Z plus any suffix the repo's PR convention requires. The body lists the bumped files and bump rationale and follows the repo's PR body rules.

    • Merge. Wait with gh pr checks <n> --watch. A failing or pending check never merges: stop and leave the PR open. When gh pr checks still reports no checks about a minute after the PR opened, read the required checks on $DEFAULT: gh api repos/{owner}/{repo}/branches/$DEFAULT --jq .protection.required_status_checks.contexts and any required_status_checks rule in gh api repos/{owner}/{repo}/rules/branches/$DEFAULT. With none required, there is no failing check: go on to the merge question. A required check that has not reported is pending. Ask the user once to merge (AskUserQuestion), unless the caller's approved plan already authorizes this bump PR and its merge (as the milestone-workflow close-out does). A decline leaves the PR open and stops with no tag. Merge with a method the repo allows and pin the head: read gh repo view --json squashMergeAllowed,mergeCommitAllowed,rebaseMergeAllowed, take --squash, else --merge, else --rebase, and run gh pr merge <n> <method> --match-head-commit <head-sha>; a resumed PR's <head-sha> is the headRefOid its checks read. If branch protection or a required review blocks the merge, stop, leave the PR open, and report what it waits on.

    • Resolve the tag target. Read gh pr view <n> --json state,mergeCommit. If state is not MERGED (a merge queue holds it, or gh pr merge only enabled auto-merge), stop, report it queued or waiting with its state, and tag nothing; a rerun resumes it through the resume bullet once merged. Else SHA is mergeCommit.oid, which must be a non-empty commit SHA (empty is a stop naming the PR state); never tag the bump branch head or the checkout's HEAD. Then git fetch origin, check the invariant on $SHA, confirm git show "$SHA":<manifest> holds X.Y.Z for every bumped manifest, and remove the worktree if it exists (git worktree remove <path>).

  5. Create an annotated tag on the resolved SHA. Recheck the invariant first:

    bash
    git fetch origin
    git merge-base --is-ancestor "$SHA" "origin/$DEFAULT"
    git tag -a vX.Y.Z "$SHA" -m "vX.Y.Z"
  6. Push the tag: git push origin vX.Y.Z.

  7. Create the release with auto-generated notes: gh release create vX.Y.Z --verify-tag --generate-notes --title "vX.Y.Z", and print the URL from its output. Never hand-write notes, and never let gh release create create the tag.

Frequently asked questions

What does the Create Release AI skill do?

Use when the user asks to "create a release", "cut a release", "tag a version", "publish release notes", or ship a new versioned GitHub release for the current repo.

Why use Create Release on TypingMind?

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

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

Which AI models can use Create 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 Create Release?

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

Is the Create Release AI skill free?

Yes. It is published on GitHub by richkuo 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 👇