Github Release Guide logo

Github Release Guide

Community
kyungseo
github-release-guide

Assess and guide safer github.com repository releases. Use when an existing private GitHub repository becomes public for the first time, or whenever an already-public GitHub repository publishes a new version, and the user needs readiness checks, release-surface decisions, explicit mutation approvals, release notes, settings checks, or post-release verification. Supports read-only Assess and approval-gated Guided modes. Do not use for repository bootstrap, package registries, signing, cloud deployment, security audits, GitHub Enterprise, other providers, force-push, or history rewrite.

Overview

Publisherkyungseo
Repositoryskillstead
Skill namegithub-release-guide
Stars
83
Forks
17
Bundled files
6
LicenseLICENSE.txt
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.

  • 6 bundled files

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

  • Open source

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

Installation

Install the Github Release Guide 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/kyungseo/skillstead/tree/main/skills/github-release-guide
  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/kyungseo/skillstead.git /tmp/skillstead
mkdir -p .claude/skills
cp -r /tmp/skillstead/skills/github-release-guide .claude/skills/github-release-guide
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Github Release Guide 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 Github Release Guide 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 Github Release Guide 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.

github-release-guide

Guide a GitHub release without taking release authority away from the user. Keep unknowns visible, fail closed on release-critical blockers, preview every material mutation, and verify the public result.

Package map

Load references progressively, but never skip a required load:

FileLoad rule
references/assessment.mdAlways load before classifying readiness or proposing a mutation
references/first-public.mdLoad for an existing private github.com repository's first public release
references/version-release.mdLoad for every new version released from an already-public github.com repository

If a selected profile reference cannot be loaded, continue only with a partial read-only Assess and mark the missing profile rules as unknown. Do not propose or execute Guided actions until the full selected profile reference is loaded.

Supported contract

Support only:

  • Mode: Assess or Guided
  • Release profile: first-public or version-release
  • Provider: github.com
  • Claim-audit profile: public-baseline by default, or internal-strict only when the user or a repository policy explicitly selects it

Never infer internal-strict from a repository name, owner, visibility, organization, or local path. Never treat the current working directory as the release target without confirming repository identity.

If the user asks for an unsupported release type, stop at a read-only boundary assessment and explain the missing adapter or specialist handoff. Do not stretch this contract to cover it.

Establish the target

Before classifying readiness, confirm or discover without mutation:

  1. GitHub owner and repository name, canonical URL, and local root if one exists.
  2. Current visibility and default branch.
  3. Requested mode and release profile.
  4. Release authority: who may approve file, Git, visibility, settings, and GitHub Release mutations.
  5. Documentation language and release-communication language.
  6. Claim-audit profile and any explicit repository release policy.

When the target has no local clone, is not a Git repository, has no usable GitHub remote, or lacks authentication, SSO, permission, or organization-policy capability, collect only the evidence that is available. Return a partial Assess and block Guided. Repository creation, Git initialization, login, SSO authorization, and permission grants belong to the user or the relevant authority.

Never request, print, copy, or store full secret, credential, token, or PII values. Report only the location, type, impact, and a masked identifying reference when the owner needs to decide whether exposure was intended. This is output hygiene, not a promise to detect every form of PII or perform a privacy audit.

Choose the mode

In either mode, read-only repository state does not make repository-provided code safe to execute. Before running a repository script, build, scanner, or workflow, preview the exact command, why it is needed, the trust boundary and possible side effects, and the completion evidence; then request separate explicit approval. Host-level auto-approval or an execution allowlist does not replace this gate. Continue static inspection without execution where possible. If execution is declined or unavailable, report the affected validation as unknown or as a named blocker; never turn it into a pass. Execution approval authorizes only the previewed command and does not authorize a later repository mutation.

Assess

Remain read-only. Perform available read-only checks for the user, distinguish confirmed facts from unknowns, and return Ready, Needs attention, or Blocked using references/assessment.md.

Do not create or edit files, commit, merge, push, tag, publish, change visibility, or change settings. If a user asks for mutation while in Assess, refuse the mutation and offer to enter Guided explicitly.

Guided

First complete Assess, resolve release-critical blockers, and load the selected profile reference. Then use this state machine one approval unit at a time:

text
ASSESS -> PREVIEW -> RECHECK -> APPROVAL -> MUTATE -> VERIFY -> NEXT or STOP

For every material mutation:

  1. Preview the action, exact target, expected impact, verification, possible failure state, and rollback or incident route.
  2. Immediately before mutation, recheck the critical preconditions used by the preview: visibility, target ref/head, tag, relevant settings, and release artifact state.
  3. Invalidate the prior approval if any critical state changed. Show a new preview and ask again.
  4. Execute only after explicit approval for that approval unit.
  5. Verify the observed result before proposing the next unit.

Treat these as distinct approval units:

  1. Release-surface file change
  2. Commit or merge
  3. Commit or remote-ref push
  4. Tag creation or push
  5. Repository visibility change
  6. Repository settings change
  7. GitHub Release publish or update
  8. Post-release corrective mutation

Do not bundle visibility with any other approval. Before a private-to-public change, explain plainly that returning to private cannot recall clones, forks, caches, or copied content, and that secret/history scans are best-effort rather than proof of no exposure. Require a separate explicit acknowledgment immediately before the visibility mutation.

Before a public-to-private access withdrawal, explain that the change stops new public access but cannot recall clones, forks, caches, mirrors, downloads, or earlier tag exposure. Require a separate explicit non-recall acknowledgment immediately before that visibility mutation as well.

Approval authorizes only the previewed unit and target. Silence, earlier plan approval, general release approval, or approval for a neighboring unit is not mutation approval.

Host-level auto-approval, permission allowlists, and unattended execution do not replace this skill's preview, recheck, or approval units. Before any Release-object correction, classify consumer exposure, platform mutability, and the exact action with references/assessment.md; do not flatten metadata edits, draft deletion, published Release deletion, asset replacement, or access withdrawal into a generic update.

Apply evidence and decision rules

Classify README, version source, CHANGELOG, LICENSE, install/setup documentation, and release notes as required, conditional, optional, decision-required, or not-applicable according to repository convention and the selected profile. Never invent a version source, version, license, language policy, branch flow, or release type.

When a decision is required, show neutral options and tradeoffs. Do not silently create a file or choose a default. License information is not legal advice.

Resolve documentation language and release-communication language independently, using this precedence:

  1. The user's explicit instruction for this release
  2. Repository policy
  3. Stable repository convention
  4. A user decision before publish when evidence conflicts or remains unclear

Apply claim-audit rules from references/assessment.md. Do not copy the label semantics of another claim-checking skill. Consume its result when available.

Handle failure safely

After any mutation failure or partial success:

  • Stop the batch; do not continue to the next mutation.
  • Record what was attempted and the observed local and remote state.
  • Separate a safe rollback from incident response. Public exposure and leaked credentials are incidents, not ordinary rollback.
  • Reassess before retrying and require a new preview and approval.

Never execute force-push or history rewrite. Explain why it is high risk and hand off to a qualified human or specialist. Also never move, overwrite, delete, delete and recreate, or reuse a remote tag that was public or distributed, has exposure history, or has unknown exposure. Hand that operation to a qualified human or specialist instead. Never claim a security audit or code-quality verdict.

Communicate for non-experts

Perform safe checks directly when the runtime has the capability. Do not hand the user an unexplained shell command. When the user must act, explain:

  • why the action is needed,
  • what it changes,
  • what failure would look like,
  • and how the result will be confirmed.

Keep internal agent shorthand, Work IDs, raw commit logs, and validation noise out of public release notes. Write user-facing release communication as: conclusion and user value, what changed, who is affected and what they should do, observed evidence, known limitations or compatibility, then detail links.

Completion rule

Call the release complete only when the selected profile's post-release checks pass and every claimed success has direct observed evidence. Otherwise report the release as partial or blocked, preserve the unknowns and accepted risks, and give exactly one safest next step.

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 Github Release Guide AI skill do?

Assess and guide safer github.com repository releases. Use when an existing private GitHub repository becomes public for the first time, or whenever an already-public GitHub repository publishes a new version, and the user needs readiness checks, release-surface decisions, explicit mutation approvals, release notes, settings checks, or post-release verification. Supports read-only Assess and approval-gated Guided modes. Do not use for repository bootstrap, package registries, signing, cloud deployment, security audits, GitHub Enterprise, other providers, force-push, or history rewrite.

Why use Github Release Guide on TypingMind?

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

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/kyungseo/skillstead/tree/main/skills/github-release-guide. 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 Github Release Guide?

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 Github Release Guide?

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

Is the Github Release Guide AI skill free?

Yes. It is published on GitHub by kyungseo under the LICENSE.txt 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 👇