跳到正文
cccskills
免费在 GitHub 公开

app-release-readiness

Use when preparing a desktop or web app update for commit, packaging, installer validation, GitHub publication, release asset upload, or old-release pruning. Protects user data and verifies that published artifacts match the tested build.

中文简介正在准备中。当前显示原文说明。

集成工具GitHub
查看安装方法

包含的文件(5)

  • SKILL.md3.8 KB
  • agents/openai.yaml287 B
  • LICENSE1.0 KB
  • references/app-quality-principles.md5.2 KB
  • references/living-to-tell-casebook.md4.7 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

App Release Readiness

Use this skill when a user asks to submit, push, package, upload an installer, or confirm that the latest app update is actually available.

Workflow

  1. Preflight the repository.

    • Run git status --short and identify unrelated changes.
    • Confirm branch, remote, version/tag policy, and whether this is a source commit, installer upload, or both.
    • Do not commit build artifacts unless the repo explicitly tracks them.
    • If the user asks to limit old GitHub Releases for a maintained desktop app and gives no other number, keep the newest 15 release records and do not delete git tags.
  2. Verify before packaging.

    • Run the smallest meaningful backend/frontend tests for the change.
    • Run full build checks when UI, packaging, provider, or app lifecycle code changed.
    • For desktop apps, include Rust/Tauri or native shell checks when present.
    • Scan for key-shaped secrets and private data when provider credentials, release notes, screenshots, or docs changed.
  3. Package with process safety.

    • Close or handle app/sidecar processes before overwriting binaries.
    • Prefer installers that fail on locked files over half-updated installs.
    • Ensure uninstall/update behavior does not delete user data by default.
  4. Inspect build outputs.

    • Record installer paths, sizes, and SHA256 hashes.
    • Compare expected sidecar/app binaries when stale-process or half-update bugs were part of the release.
    • Smoke-test the installed app when feasible.
  5. Commit and push deliberately.

    • Stage only intended source/docs changes.
    • Use a direct commit message that names the user-visible change.
    • Push the branch and verify the remote commit SHA before publishing source claims or release notes that point at the new version.
  6. Upload release assets when requested.

    • Use the repository's current release/tag policy.
    • Upload with replace/clobber only when the user asked for the latest asset to replace the old one.
    • Verify remote asset names and sizes after upload.
    • Download uploaded assets back from the release URL and verify SHA256 against the local artifacts before saying the release is current.
  7. Prune releases only when requested.

    • Delete release records, not tags, unless the user explicitly asks for tag deletion.
    • Re-list releases after pruning and report the final count.
    • If keeping 15 records, preserve the newest 15 by published release order.
  8. Report with evidence.

    • Include commit SHA, pushed branch, tests run, package paths, hashes, and release URL when applicable.
    • Say explicitly if installation or provider smoke tests were not run.

Reference Use

Read references/app-quality-principles.md before releasing changes that touch data paths, provider integrations, startup behavior, user-facing errors, or installer behavior.

Read references/living-to-tell-casebook.md when the release includes fixes for stale sidecars, startup black screens, raw Not Found/Failed to fetch errors, AI provider transport, or uninstall/data-location safety.

Release Red Lines

  • Do not say "uploaded" until the release page or API confirms the asset.
  • Do not say "latest" until the remote asset hash or commit matches the current build.
  • Do not force-push, delete tags, or replace release assets unless the user asked for that release behavior.
  • Do not package or publish secrets, local databases, private writing content, logs, or screenshots containing private data.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Use when diagnosing or fixing user-reported application bugs. Guides evidence-backed root-cause analysis from symptoms, screenshots, logs, provider failures, background tasks, storage, or desktop lifecycle behavior through a scoped patch and regression tests.

中文简介正在准备中。当前显示原文说明。

sidiangongyuan/codex-skills-library242026年9月21日 更新

Use when implementing or improving a user-facing app workflow. Guides product-grade UX, frontend, backend/API behavior, AI/provider or long-running task states, tests, and end-to-end verification from rough requirements to a shippable feature.

中文简介正在准备中。当前显示原文说明。

sidiangongyuan/codex-skills-library242026年9月21日 更新

Use when Codex Desktop sidebar sessions disappear, fail with systemError, or cannot continue after a model_provider switch. Diagnoses and repairs active, non-archived sessions without changing authentication or provider configuration.

中文简介正在准备中。当前显示原文说明。

sidiangongyuan/codex-skills-library242026年9月21日 更新

Anti-slop frontend skill for landing pages, portfolios, and redesigns. The agent reads the brief, infers the right design direction, and ships interfaces that do not look templated. Real design systems when applicable, audit-first on redesigns, strict pre-flight check.

中文简介正在准备中。当前显示原文说明。

sidiangongyuan/codex-skills-library242026年9月21日 更新

Describe what this skill does and the concrete situations in which an agent should use it.

中文简介正在准备中。当前显示原文说明。

sidiangongyuan/codex-skills-library242026年9月21日 更新

Use when exploring, continuing, or accepting a deep-learning or computer-science research idea. Converts claims into pilot-first experiment matrices covering diagnostics, failure analysis, authorized execution, evaluation, and paper-story viability.

中文简介正在准备中。当前显示原文说明。

sidiangongyuan/codex-skills-library242026年9月21日 更新

查看 sidiangongyuan 的全部技能

报告此技能的问题