skills.rest
SmallAi-API
SmallAi-APICOMMUNITY
Explore all skills by this author →
WebsiteGitHub

agent-testing

Runs agentic end-to-end verification across CLI, web, and Electron surfaces with evidence-backed reports.

Updated Jan 27, 2026
Search Tags:#electron#cdp#end-to-end testing#acceptance testing#agent-browser#screenshot evidence#cli verification
One-click install
npx skills add https://github.com/SmallAi-API/smaihub --skill agent-testing-smallai-api
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-testing
Source: https://github.com/SmallAi-API/smaihub/tree/main/.agents/skills/agent-testing
Command: npx skills add https://github.com/SmallAi-API/smaihub --skill agent-testing-smallai-api

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser, ws, and includes scripts (resource) and references (resource) components.

What problem does it solve? Manual end-to-end verification of backend, web, and desktop changes is slow, flaky, and hard to audit. This Skill drives the real product surface, captures visually confirmed evidence, and publishes a structured acceptance report instead of a chat-only summary. ## Core Features & Use Cases - Multi-surface verification: Tests backend changes through the project CLI, web changes through agent-browser (CDP), and desktop changes through Electron with raw CDP screenshots. - Evidence discipline: Enforces screenshot/GIF/audio evidence per case, dual text artifacts for non-visual claims, and native structured visualizations for metrics and comparisons. - Structured acceptance reports: Scaffolds immutable report rounds grouped by acceptance subject and publishes them to a LobeHub Acceptance page. - Use Case: After changing an API and its consuming UI, ask the agent to verify the feature end-to-end: it confirms the environment and auth, runs CLI and browser checks, captures screenshots, and publishes a reviewable acceptance round. ## Quick Start Ask the agent to run an end-to-end test of your current feature change and publish the acceptance report.

Frequently Asked Questions about agent-testing

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run end-to-end tests on a local dev server with an AI agent?▼

Invoke the skill with a test target; it reads the project's PROJECT.md adapter for start commands, ports, and auth, confirms the environment, then executes cases on the appropriate surface. It ends by publishing a structured acceptance report with captured evidence.

How do I test an Electron app with browser automation?▼

Launch Electron with a remote debugging port and drive it through agent-browser over CDP, or capture screenshots with the included cdp-screenshot.sh script using raw Page.captureScreenshot. CDP capture works even when the window is minimized or the display is asleep.

Why are my macOS screenshots coming out completely black?▼

Black captures happen when Screen Recording permission is missing or the display is asleep, locked, or screensavered. Run check-screen-recording.sh to diagnose both layers, and keep the display awake with caffeinate -dimsu during the run.

Can unit tests or lint checks be used as acceptance test cases?▼

No. Programmatic gates like unit tests, coverage, type-check, lint, and builds are hard-blocked from acceptance cases and are dropped on ingest. Every case must be a delivery outcome a person can judge with real surface evidence.

What evidence is required for time-based UI behavior like streaming or animations?▼

Time-based behavior requires a GIF recorded with record-gif.sh, since a static screenshot cannot prove streaming output, timers, or animations. Attach the GIF to the case's evidence so the acceptance page renders it inline.

Related Skills

Explore More Software Engineering Skills→
pattern-stackpattern-stack
1

create-worktree-skill

Create isolated Git worktrees with auto-incremented ports and environment files.

Official
Intermediate
auerbachbauerbachb
3

wrap

Verify unresolved reviews, squash-merge pull requests, and clean up local worktrees.

Community
Advanced
HomericIntelligenceHomericIntelligence
6

verification

Run pytest, mypy, ruff, and pre-commit checks before declaring completion.

Official
Intermediate
cosmicstack-labscosmicstack-labs
364

e2e-testing

Write stable end-to-end tests for web apps using Playwright and Cypress.

Official
Intermediate