OpenEvidentOpenEvident home

01 Agentic quality engineering

The agent decides what to test next. A human still reviews it.

Agentic quality engineering means AI agents plan, generate, execute, and analyze quality work across the software development lifecycle — not just replay a fixed script.

The agent decides what to cover next, writes the check, runs it, and reports back. A human reviews the output instead of writing every line by hand — that is the shift from test automation to agentic QA.

Several open-source projects are exploring this space. Agentic QE Fleet, for example, packages many specialized skills into a single coding-agent session. OpenEvident takes a different shape: each layer of the lifecycle — the Playwright layer today (Vindicate), cross-service proof next (Evident) — ships as its own growing, independently star-able GitHub project, on the same local-first, Apache-2.0 bar.

In practice, that means a Playwright test generation AI agent already handles the build-and-execute layer, while planning, design, and non-functional layers are open modules still landing. Follow the org to see each piece go from “coming” to “live.”

  • 01

    Plan & analyze

    A Strategy Agent turns requirements and risk into a test plan another agent can follow. Not shipped yet — tracked on the same Apache-2.0 bar as everything else.

  • 02

    Design & create

    A Test Architect Agent drafts suites and data maps before a single script exists. Also an open module still landing.

  • 03

    Build & implement

    Vindicate's Scripting Agent writes and maintains the suite locally over MCP. This layer is live today.

  • 04

    Execute & orchestrate

    Vindicate's Execution Agent runs and recovers tests on your machine — no third-party job runner. Also live.