Getting Started

Skiritai is an AI-driven browser test automation framework. It uses LLMs to explore web applications and generate replayable Playwright scripts.

How It Works

Skiritai operates in a two-phase loop:

1. Explore — An AI agent analyzes the page, decides what actions to take, and executes them via Playwright. It records every step.
2. Replay — The recorded steps are compiled into a standalone Python script. Replaying it is ~30x faster because no LLM is involved.

# example.py
from skiritai import BaseCase, step, run_case

class MyTest(BaseCase):
    """Search for Skiritai on GitHub."""

    @step
    async def search_github(self, ai):
        await ai.action("导航到 GitHub 并搜索 Skiritai AI testing")

    @step
    async def verify_results(self, ai):
        await ai.action("验证搜索结果包含 Skiritai")

if __name__ == "__main__":
    run_case(MyTest)

Run it:

skiritai run . --case example.py

On the first run, the AI explores and figures out the right selectors. On subsequent runs, the generated replay script runs directly at full speed.

Alternative: Flow API (No Subclass)

Prefer a functional style? Use the Flow API — no BaseCase, no decorators:

from skiritai import flow

async with flow() as ai:
    await ai.action("Navigate to GitHub and search for Skiritai")
    await ai.verify("Search results include Skiritai")

Alternative: YAML Cases (No Code)

No Python at all? Define your test in YAML:

# case.yaml
steps:
  - action: Navigate to GitHub and search for Skiritai
  - verify: Search results include Skiritai

skiritai run .