expect
Expect tests your agent's code in a real browser
Expect is an agent skill that enables AI coding assistants to automatically test code changes in a real browser environment. It analyzes git diffs, generates test plans, and executes them using Playwright to identify performance issues, security vulnerabilities, design regressions, and functional bugs. The skill integrates with popular coding agents including Claude Code, Codex, GitHub Copilot, Cursor, and others, running entirely on your local machine. Instead of manually writing test scripts, developers can invoke Expect via a slash command to have their AI agent spawn subagents that simulate real users and automatically verify code quality.
Key Features
Use Cases
- 01Automated regression testing before committing code changes
- 02Performance profiling of web applications to identify slow interactions and render times
- 03Security auditing of npm dependencies and common web vulnerabilities
- 04Testing authenticated user flows by reusing real browser cookies
- 05Continuous integration testing via GitHub Actions on every pull request
- 06Design QA to catch broken UI states and missing metadata before deployment
Related Skills
View moresuperpowers
An agentic skills framework & software development methodology that works.
skills
Skills for Real Engineers. Straight from my .agents directory.
skills
Public repository for Agent Skills
ponytail
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
expect — FAQ
What is Expect and how does it work?+
Expect is an agent skill that reads your git changes, automatically generates a test plan, and executes it in a real browser using Playwright. It integrates with coding agents like Claude Code and Cursor, running locally on your machine to test for performance, security, design, and functionality issues.
How do I install Expect?+
Install Expect globally via npm with 'npm install -g expect-cli'. After installation, ask your coding agent to fetch the init script from expect.dev, then invoke the skill by running '/expect' inside your agent.
Which coding agents does Expect support?+
Expect works with Claude Code, Codex, GitHub Copilot CLI, Gemini CLI, Cursor, OpenCode, and Factory Droid. It auto-detects which agents are installed on your PATH, or you can specify one using the '-a' flag.
Does Expect require API keys or external services?+
No, Expect runs entirely on your local machine and does not require API keys. It uses your system's Chrome browser profile and cookies to test authenticated flows, and integrates with whichever coding agent you have installed.
Is Expect free to use?+
Yes, Expect is open-source software licensed under FSL-1.1-MIT and available for free via npm. Hosted and enterprise versions are planned for the future.
Can Expect run in CI/CD pipelines?+
Yes, Expect supports CI mode with the '--ci' flag or via the 'add github-action' command. In CI, it runs headless, skips cookie extraction, auto-approves test plans, and enforces a 30-minute timeout.
How do I install expect?+
Open the source repository on GitHub and follow its README. expect is a skill — MCP Agents Market links you directly to the official repo.
Is expect free?+
expect is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.