Mini Program Release Governance for AI Agents

Where this fits

A developer asks their agent: "Prepare v1.4.2 for WeChat mini program release." The agent runs a build, commits a version bump, uploads to the platform, and reports success. Later the developer discovers the uploaded artifact came from the wrong branch and the privacy declaration is outdated. This skill solves that by forcing agents to separate evidence from action: it runs a read-only release preflight, checks every version and artifact fact, and withholds judgment until the chain from source to build to platform is auditable.

Why agents benefit

• Read-only by default: The skill never treats a Git push, successful build, platform upload, or review submission as interchangeable. Each external state change requires explicit separate authorization.
• Evidence correlation: It verifies that the current branch, commit, version facts, source code, and build artifacts are all bound to the same release fingerprint. If any evidence mismatches, the agent stops at not-ready.
• Rules-aware checks: For WeChat permissions, privacy declarations, and review requirements, the skill queries current official documentation rather than relying on stale assumptions; uncertain results stay unknown.
• Auditable release readiness: The output is a precise list of what has and has not shipped, not a vague summary, which gives the developer a verifiable preflight report before any destructive or external action.
• Release lifecycle coverage: It handles the full path: version bump, packaging, upload readiness, submission readiness, production launch, and rollback preparation.

Practical scenarios

Pre-submission audit for WeChat review

Before submitting a mini program for review, the agent runs the skill against the target branch. It catches that the VERSION file was bumped but the README badge still points to the previous release, preventing an inconsistent submission.

Post-release rollback preparation

After a production launch, the developer asks for rollback options. The skill verifies which staging version is fully uploaded and approved, checks the release history, and reports the exact previous production artifact and the steps needed to restore it.

Multi-branch feature merge check

A feature branch has been merged, but the build artifact was generated from an older commit. The skill detects the fingerprint mismatch and returns not-ready with evidence, prompting a clean rebuild rather than a blind upload.

Add it to your agent workflow

Install it with:

npx skills add NocodeMrLi/mini-program-engineering-skill-suite --skill mini-program-release-skill

Then ask your agent something like: "Run the mini-program release preflight for v1.4.2 targeting the main branch. Do not upload or submit anything; just give me a release-ready verdict with the evidence you found." The skill will gather the facts and stop before any external action.

Compare before adopting

Before adopting this skill, compare it against alternatives by looking at:

• Quality signals: Does the README clearly define authorization boundaries and evidence rules? Is the release logic explicit about read-only defaults?
• Maintenance freshness: Are the WeChat platform checks updated regularly, given that policy changes frequently?
• Alternatives: Generic release scripts or simple version-bump prompts often lack multi-stage governance. Weigh that tradeoff.
• Workflow fit: Does your agent already have a release pipeline? This skill works best when you need an independent preflight audit layer, not when you want a single command to do everything.

Why it is worth tracking

With 22 stars and a narrow, well-defined scope, this skill is early but purposeful. Its value comes from enforcing a strict release governance model that most agents lack by default. Track it if you build agents that touch WeChat mini program releases or similar platform submission workflows; as the author adds more policy checks and evidence types, it could become a reliable reference point. For a living index of agent skills including this one, see https://www.openagentskill.com/skills/nocodemrli-mini-program-release-skill when evaluating it against other release-governance options.