Build Outcome Verifier

Stop trusting "build succeeded." Start verifying.
A SKILL.md for AI coding agents that physically installs and launches APKs after every build.
Android APK/AAB Claude Code Codex CLI Cursor Gemini CLI OpenClaw $7.99 on Agensi 0.002 $SKILL on SkillMarket

Demo

See Build Outcome Verifier in action — 30 seconds:

Direct link: demo-30s.mp4

How to Get It

Build Outcome Verifier is a premium agent skill available exclusively through:

Compatible with Claude Code, Codex CLI, Cursor, Gemini CLI, OpenClaw, Windsurf, GitHub Copilot, and 15+ other agent platforms.

Why This Exists

Agents say "build succeeded" when:

This skill forces observable proof that the artifact is functional. No proof, no "done."

Platform Support

PlatformStatusTools Required
Android (APK/AAB)✅ Primaryadb, device/emulator
iOS (IPA)🔜 Plannedxcrun, ios-deploy
Web (static/PWA)🔜 Plannedhttp-server, Playwright
macOS/Windows🔜 Plannedhdiutil / Selenium

Use Cases

AI Agents That Lie About Builds

Claude Code says "Build successful." You install it — crash on launch. This skill catches that before you waste time.

CI/CD Pipelines That Need Real Smoke Tests

Your GitHub Actions build passes but the APK is 4KB (it's a stub). This skill runs as the last CI step: install, launch, verify. If it fails, the pipeline fails — not production.

Solo / Indie Devs Shipping at 2am

No QA team. Pushing a build at 2am. You need to know the APK actually installs before submitting to Google Play. One command loads this skill.

Multi-Agent Build Pipelines

Agent A builds. This skill verifies. Agent C deploys. Quality gate between stages — no broken artifacts progressing downstream.

Regression Detection

A dependency update breaks launch but compiles fine. Only a runtime smoke test catches it. Run this on every PR merge.

CI/CD Integration

Build Outcome Verifier works with any CI runner that supports Android emulators — GitHub Actions, GitLab CI, Jenkins, CircleCI. After purchasing, the skill includes a ready-to-use GitHub Actions workflow snippet.

The skill's verification commands run as the final step in your pipeline. If the app doesn't install or crashes on launch, the pipeline fails — not production.

Safety & Permissions

PermissionWhat It DoesWhy Needed
adbInstall APKs, launch apps, read device logsVerify installs and detect crashes
Local shellRun adb commands on the user's machineAll verification is local
Logcat (read-only)Inspect crash/ANR outputDetect runtime failures

🔒 No network calls. No telemetry. No API keys. No third-party services. Works fully offline.

Troubleshooting

ProblemCauseFix
adb: command not foundADB not installedsudo apt install adb / brew install android-platform-tools
no devices/emulators foundNo Android device connectedStart emulator or connect USB device
INSTALL_FAILED_UPDATE_INCOMPATIBLEDifferent signatureadb uninstall com.yourcompany.yourapp first
ANR in backgroundBackground thread deadlockIncrease sleep to 15s

FAQ

Do I need an emulator running?

Yes — adb needs a device or emulator. Most CI runners support Android emulators.

Will this slow my builds?

~15 seconds total. The time saved not debugging a broken deploy dwarfs this.

Does it support iOS?

Not yet — planned for a future release.

Can I add custom checks?

Yes — it's plain markdown. Add your own verification commands between the install step and the pass declaration.

What agents does it work with?

Any agent supporting SKILL.md — Claude Code, Codex CLI, Cursor, Gemini CLI, OpenClaw, Windsurf, Copilot, and 15+ others.

What if verification fails?

The agent is instructed to record the failure logs and NOT claim "task complete." It must fix the issue and rebuild before the verifier will pass.