cy-ai: Use Natural Language to Automatically Generate End-to-End Test Code for Cypress

cy-ai is an open-source Cypress plugin developed by the ai-action organization. It integrates large language models (LLM) to let developers generate end-to-end (E2E) test code using natural language, reducing the barrier to writing Cypress tests. Key benefits include lowering cognitive load (no need to memorize Cypress APIs), improving efficiency, and enabling non-technical team members to participate in testing. The project is available on GitHub under the MIT License.

Writing Cypress E2E tests requires familiarity with its API and syntax, which can be a barrier for many. Complex interaction scenarios lead to冗长 and hard-to-maintain code. cy-ai addresses this by leveraging LLMs to translate natural language descriptions into valid Cypress code. The project was created in May 2025 and is actively updated as of June 2026.

cy-ai's workflow has three steps:

1. Intent Capture: Developers use cy.ai() with natural language (e.g., cy.ai('see heading "Example Domain"')).
2. Context Building: It captures current page HTML to provide context to the LLM.
3. Code Generation: The LLM produces Cypress code following best practices (no redundant describe/it). Tech stack: Cypress (base framework), TypeScript (language), LangChain (LLM interaction), Ollama (local model support like qwen2.5-coder, codellama).

Pre-requisites: Cypress installed, Ollama running locally. Installation:

• NPM: npm install cy-ai --save-dev
• Yarn: yarn add cy-ai --dev Configuration:

1. Import command into Cypress support file (TypeScript: import 'cy-ai'; JS: require('cy-ai')).
2. Start Ollama: ollama serve and pull model: ollama pull qwen2.5-coder.
3. Chrome: Disable chromeWebSecurity in cypress.config.js to avoid CORS issues.

cy-ai supports customization:

• Switch Models: Use other Ollama models (e.g., codellama) via LangChain.
• Custom Prompts: Define your own prompt templates to control code generation.
• Global Config: Use cy.aiConfig() to set options like custom LLM chain, log toggle, regenerate flag, and timeout. Example code snippets for custom models and prompts are available in the project docs.

cy-ai is useful in:

1. Quick Prototype Validation: Fast test generation for rapidly changing prototypes.
2. Non-Technical Team Participation: Product managers/QA can write tests using natural language.
3. Legacy Project Test Coverage: Quickly generate basic tests for projects with no existing test suite. These scenarios help teams save time and reduce communication gaps.

As of June 2026:

• GitHub stars:7.
• Tech stack: TypeScript.
• CI: Configured build/test workflows.
• Docs: Comprehensive README and Wiki.
• Examples: cypress-ai-demo repository.
• License: MIT. The project is actively developed. Community members can contribute via GitHub Issues or Discussions.

cy-ai is a valuable tool for AI-assisted E2E testing. Its core values are lowering entry barriers, improving efficiency, and enhancing code maintainability. Future prospects include AI integration in more dev areas (unit tests, code refactoring). Who should try: Front-end teams wanting easier E2E testing, AI tool explorers, agile teams needing quick test coverage. Tips: Start with simple tests, review generated code, and follow project updates for new features.