Lento Circle — Autonomous Test Framework
Built a fully autonomous test automation framework using Playwright + Python, orchestrated by MCP servers and AI agents.
The challenge
The team wanted to eliminate the biggest cost in test automation — the human writing and maintaining test scripts. Could AI agents, driven by MCP servers, author and run Playwright tests end-to-end from a test case description?
The approach
- 1
Designed an MCP server that pulls test cases directly from TestLink and exposes them to the agent as structured prompts.
- 2
Built the agent orchestration layer: one agent interprets the test case, another writes the Playwright code, another runs it in a sandboxed browser and reports results back.
- 3
Added self-healing: when a test fails, a review agent inspects the failure, decides if it's a bug or a flaky selector, and either files it in TestLink or patches the test.
The outcomes
- Test cases pulled automatically from TestLink into the agent pipeline
- AI agents author and run Playwright tests via MCP orchestration
- End-to-end autonomous run: zero manual scripting from spec to report
Have a similar project?
Let's find the right package for your context. Start with a Spot Check and expand from there.
Get in touch