Quality engineer
Nimali Jayawardena
Colombo
I want agents writing tests I would still sign. Grounded locators, real browsers, suites that live in the repo — not a dashboard I cannot audit.
OpenEvident Foundation
Make good ideas matter
Open-source community building local-first test agents for E2E quality governance. Run it on your machine, no cloud account.
Local firstPlaywright MCPAgent-driven testsCross-service proofNo telemetryQuality engineeringOpen sourceApache 2.0Local firstPlaywright MCPAgent-driven testsCross-service proofNo telemetryQuality engineeringOpen sourceApache 2.0
01 Principles
What we stand for.
01
Local first
Code and data stay on your machine. No cloud account, no remote job runner, no telemetry you cannot disable.
02
Agent native
Cursor, Copilot, and Claude Code call the tools directly. Quality engineering that agents can actually run.
03
Evidence over dashboards
Prove what happened from the structured logs you already produce. No tracing agent embedded in your services.
04
Open by default
Apache 2.0. No login wall. Community-wide code of conduct and security disclosure, in the open.
02 How we work
How quality becomes evident.
Grounded in the product. Generated by agents. Proven in your logs.
01
Ground
Capture real flows in the browser. Selectors from the live page, not guessed from a screenshot.
02
Generate
Agents write Playwright tests from that evidence. Page objects, specs, and fixtures that a human can still read.
03
Prove
Fire a real request against your services. Read the structured logs downstream. That is the assertion.
04
Heal
Keep suites alive as the product moves. Recordings, grounded locators, and CI that does not leave your account.
03 Building blocks
From code to confidence.
Click a block to see which GitHub project ships that function — and what is still on the open-source map.
- Live
- Early
- Coming
- Flow of work
- Collaboration & feedback
Inputs
Design docs
Architecture, APIs
Testing pyramid · HITL
AcceptanceUAT Agent
Product Outside - InSystem Test Agent
Dev Inside - OutUnit & Integration Agent
Validation · Outcomes
05 Integration & automation foundation
04 Community
Decentralized Agentic innovation. Built on trust.
The OpenEvident Foundation brings developers and organizations together in a neutral, trusted community to build and scale open technology projects and agentic ecosystems, shaping the future of AI-powered Quality Engineering—together, today
1/3
44
Stars on Vindicate
4
Public repositories
0
Required cloud accounts
05 About
Small org. Zero layers between you and the work.
OpenEvident is an open-source organization in Sri Lanka building tools for trustworthy AI, software quality, and developer engineering. We kept the circle small on purpose. You open an issue. Maintainers answer.
Values
- Precision
- Evidence
- Community
06 FAQ
Things people ask.
Where do I start?
Star Vindicate, clone it, and follow the README. The MCP server runs on your machine. A 15-minute setup is enough to know if it fits your stack.
What is the license?
Apache License 2.0 on every public repository. There is no feature gated behind a login, and no telemetry you cannot disable.
Do I need a cloud account?
No. Vindicate is a local-first stack. The agent talks to a process on 127.0.0.1. CI uses composite GitHub Actions in your own runners.
I do not have a test plan yet.
Good. Ground a real flow first. Most teams arrive with a flaky suite or none at all. You leave with evidence the agent can write from.
07 Star
If this is useful, star Vindicate.
Stars help other quality engineers find a local-first toolkit. Takes two clicks.
What happens next
01
Open the repo.
Vindicate on GitHub. No form inside a form.
02
Click Star.
That is the whole conversion. It tells the next QE this exists.
03
Clone and run locally.
MCP on 127.0.0.1. Your agent, your machine.
04
Open an issue if you want to contribute.
Discussions for questions. Issues for bugs and features.