Describe what to test in plain English. Vigil launches a browser, reasons through your UI with an LLM, and produces video recordings with structured pass/fail reports.
No selectors to maintain. No brittle page objects. Describe the task, point at a URL, and let the agent figure it out.
Write a natural language task like "log in and add a high-priority todo" or define YAML steps for repeatable scenarios.
Vigil launches a browser, reads the DOM and accessibility tree, and uses an LLM to decide what to click, type, and verify.
Every run produces video recordings, Playwright traces, screenshots, and structured JSON/Markdown reports with pass/fail status.
Describe what you want in plain English. The agent figures out the steps, navigates pages, fills forms, and validates results on its own.
Define explicit YAML step-by-step workflows for repeatable demos. Ideal for CI pipelines, regression testing, and stakeholder reviews.
Claude, GPT-4, Gemini, Groq, OpenRouter, Azure, Ollama, or any OpenAI-compatible endpoint. Use what you already pay for.
Video recordings, Playwright traces, per-step screenshots, and animated cursor effects. Show stakeholders exactly what happened.
JSON and Markdown reports with step-by-step results, agent reasoning, timings, and error details. Machine-readable and human-friendly.
Ships with a GitHub Actions workflow. Run demos on every PR, upload artifacts, and post summary comments automatically.
The agent reads the DOM, accessibility tree, and page context before acting. Surgical, reasoning-driven interaction — not blind clicking.
Built-in web interface to browse scenarios, run demos, watch real-time progress via SSE, and inspect results — no terminal needed.
Install and set up
Set your LLM API key
Run the built-in demo
Or just describe what to test
Vigil is free, open-source, and works with the LLM you already use.