Free · Open source · AppEsteem Agent Skill

Check your consumer app for compliance

The AppEsteem ACR Compliance Skill teaches your AI coding tool (Claude Code, Codex, GitHub Copilot, Gemini CLI, Cursor, or any tool that supports the Agent Skills standard) to review your installer, offer screens, monetization and uninstall flow against the Application Compliance Requirements: the standard AppEsteem certifies against, and the one security vendors draw on.

Already flagged? If your app is being detected as PUA or PUP by security software, if your installer is being blocked, or if an app store rejected your submission as deceptive, the skill finds what is triggering it and tells you how to fix it.

Install it in one command:

npx skills add AppEsteem/skills

What the skill does

It turns the ACRs into a code-review checklist your agent can actually run. All 131 current requirements, including the 57 Deceptor-level ones, with each requirement's intent, compliance guidance, applicability gates, and the notes AppEsteem gives its own testers.

Build compliant by construction

Writing an installer, offer screen, trial or cancellation flow? The agent applies the requirements while the code is being written, not after the fact.

Review before you ship

Every finding cites the ACR id, states the severity, points at the specific code or copy, and gives a concrete fix.

Diagnose a detection

Already flagged, blocked, or listed? Point the agent at your code and it identifies the behaviors that draw AV attention, ordered worst-first.

What it reviews

  • Install Installer projects (NSIS, Inno Setup, WiX, MSI, pkg), setup UI, first-run code, what gets written to disk and registry.
  • Bundled offers Bundled offer requirements: disclosure, consent, pre-checked boxes, decline paths, and what a carrier app may install alongside itself.
  • Software Background processes, tray behavior, notifications, data collection, updates, free-scan logic, monetization.
  • Landing page & funnel Download buttons, claims, offer screens, cart, trial and subscription terms.
  • Uninstall Cleanup routines, settings reversion, retention prompts, and what survives removal.
  • Docs EULA, Terms of Service, Privacy Policy, Returns and Cancellation Policy.
  • Ads Ads about the app, ads inside the app, and anything injected into browsers or other apps.

Four prompts to start with

Install the skill, open your project in your AI coding tool, and paste one of these.

1. Build it compliant from the start

I'm adding a bundled offer screen to my installer. Using the AppEsteem ACR compliance skill, tell me the bundled offer requirements that apply, then implement the screen so it's compliant by construction.

2. Review before I ship

Review this installer and its landing page against the AppEsteem ACRs. Report every Deceptor-level violation first, then certification-level ones, and mark anything you can't judge from the code as needing human verification.

3. Check my monetization for dark patterns

Check my trial, subscription, upsell and cancellation flows against the AppEsteem ACRs for dark patterns: scare tactics, fake urgency, trick buttons, unclear recurring billing terms, or cancellation that's harder than signup.

4. Something is flagging my app

My installer is being detected as PUA by security software. Use the AppEsteem ACR compliance skill to review this project and tell me exactly which behaviors are likely triggering the detection, worst first, and how to fix each one.

Why a compliance bug is worse than a security bug

  1. AV and security vendors actively hunt deceptive behaviors. Apps get flagged as PUA/PUP or blocked outright.
  2. AppEsteem's Deceptor list feeds those vendors. A listed app can be blocked or removed without the consumer ever being prompted; installs silently fail, existing installs get cleaned off machines.
  3. Detections cascade. One vendor's flag propagates through shared intelligence; stores reject or delist, payment processors and ad networks drop the app, download portals refuse it.
  4. Reputation damage is sticky. Getting de-flagged takes months of vendor-by-vendor dispute work. Building clean from the start costs almost nothing.

Consumers deserve software that says what it does, does what it says, and leaves when asked. That is the point of the ACRs, and the commercial argument happens to run the same direction.

What's inside, and where the data comes from

  • SKILL.md: the review workflow, covering app profiling, scenario-by-scenario checks, the severity model, report format, and honest limits.
  • references/acrs.md: every current ACR with intent, compliance guidance, applicability gates and evaluation notes, generated from the live requirements API.
  • references/report-template.md: the structure findings are reported in.

The requirements are living, so the bundled reference is regenerated weekly by CI from the source of truth: AppEsteem's requirements API. The skill also tells agents to prefer fetching that API live when they have network access, so a review reflects the current standard even between regenerations.

The skill is instructions and reference data only. There is no executable code for your agent to run. It's published by AppEsteem Corporation, which writes the requirements; the tooling is MIT-licensed and the skill content is CC BY 4.0, so it can ship inside other tools' catalogs.

Honest about limits: a code review can't see everything AppEsteem tests: live server behavior, what an ad network actually serves, call-center scripts, or how the shipped binary behaves on real machines. Passing this review is not AppEsteem certification, confers no seal, and doesn't guarantee that AV vendors won't flag your app. It is a strong preventive practice, not a verdict.

Frequently asked questions

Is this a Claude Code skill, or does it work with other AI coding tools?

It's built on the open Agent Skills standard, so it works with Claude Code, Codex, GitHub Copilot, Gemini CLI, Cursor and any other tool that supports it. Install it with npx skills add AppEsteem/skills, or copy the skills/appesteem-acr-compliance/ folder into your agent's skills directory. It's instructions and reference data only; there's no executable code for your agent to run.

How do I check my software for dark patterns and compliance issues?

Dark patterns are what the ACRs have covered for years under names like scare tactics, forced action, and misdirection. Install the skill and ask your coding agent to review your offer screens, upsells, trial-to-paid conversion and cancellation flow. It classifies every applicable requirement as pass, violation, needs human verification, or not applicable, so you get an honest picture rather than a clean bill of health it can't actually give from code alone.

What are installer best practices to avoid antivirus detection?

Disclose everything you install before you install it, use one consistent app and company name everywhere including the code signature, never pre-check an offer or make declining harder than accepting, don't claim problems you can't substantiate, and make uninstall remove what you added and revert what you changed. That's the summary; the ACRs are the specifics, and they're free to read. The skill applies them to your installer project directly, so compliance gets checked at build time rather than after a detection.

My installer is being flagged or blocked by security software. How do I fix it?

Security products flag installers for specific behaviors, not for being unknown: undisclosed bundled software, alarming or unsubstantiated system-health claims, offers that are hard to decline, changes that survive uninstall. Because vendors share intelligence, one product's block tends to spread, so disputing it product by product rarely holds while the behavior is still there. Install the skill and point your AI coding tool at your installer project; it maps the code and setup UI against the ACRs and tells you which behaviors are Deceptor-level, the ones vendors act on. Fix those, rebuild, and resubmit through the vendor's false-positive channel; if your app is already on AppEsteem's Deceptor list, the fix has to happen before any dispute will hold.

My app was detected as PUA. Is it a false positive?

Sometimes, but far less often than developers assume. PUA (potentially unwanted application) detections usually point at a real behavior in the app: a pre-checked offer, a free scan that reports problems it can't substantiate, a subscription that's hard to cancel, leftovers after uninstall. Run the skill over your code and copy first. If it finds nothing and every requirement passes, you have a much stronger false-positive case to make to the vendor, and a concrete list of what you checked.

PUP detection: how do I fix it?

PUP (potentially unwanted program) is the same class of detection as PUA under a different vendor's name, and the fix is the same: identify the behavior, remove it, ship a clean build, then dispute with evidence. The 57 Deceptor-level ACRs are the shortlist of behaviors most likely to be behind it. The skill checks all of them against your actual code and copy instead of leaving you to guess.

My app store submission was rejected as deceptive. What now?

Store policies on deceptive design overlap heavily with the ACRs: misleading claims, fake urgency, trick buttons, unclear subscription terms, and cancellation flows designed to be difficult. Ask your AI coding tool to review your store listing copy, screenshots, purchase flow and cancellation flow with the skill. Each finding cites the requirement and gives a concrete fix, which usually translates directly into the change the reviewer is asking for.

Read the standard yourself

All free, no account needed. The skill points developers here too, so a review can always be checked against the source.

If violations remain that you can't resolve, if your app is already listed and you need it resolved, or if you need to prove compliance to a third party, AppEsteem offers paid app reviews and certification.