需要 Node.js 24+；init 后还要按 quickstart 连接模型才能跑 demo；项目处于早期阶段，可能遇到 bug
agent-qa
vostride活跃维护用自然语言为 Web 和移动端编写端到端测试的自改进 QA 智能体，能从每次运行中学习并自适应 UI 变化，在发布前捕获回归问题。
Open-source self-improving QA agent for software teams. A test harness with memory. Write tests in natural language for web and mobile. agent-qa learns from every run, adapts to UI changes, and catches regressions before you ship.
v0.3.3发布于 2026-11-07 12:36 UTC852star24fork同步于 2026-11-10 17:55 UTC
npm install -D agent-qa
npx agent-qa init 硬件要求按 README 说明
README 未说明
用自然语言写 Web 和移动端测试，带记忆的自改进 QA agent，能适应 UI 变化并抓回归
用自然语言为 Web 和移动端编写端到端测试的自改进 QA 智能体，能从每次运行中学习并自适应 UI 变化，在发布前捕获回归问题。
核心能力
- 用自然语言描述用户旅程，在 Web、Android、iOS 上执行测试
- memidex 把事实和操作流程存成仓库内可审查的文件
- 从每次运行中验证经验、淘汰过期知识并按反馈纠错
- 保留失败步骤与截图证据，可在 dashboard 中对照查看
- 提供 CLI、dashboard 和 MCP/skills 三种接入方式
容易踩的坑
内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README
- v0.3.3agent-qa v0.3.3最新正式版2026-11-07 · 2026-11-07 12:36 UTC
What's Changed
- Clean refusals and exit codes by @pranshuchittora in https://github.com/vostride/agent-qa/pull/27
Full Changelog: https://github.com/vostride/agent-qa/compare/v0.3.2...v0.3.3
npm package: https://www.npmjs.com/package/agent-qa Docker images: https://hub.docker.com/u/vostride Docs: https://vostride.com/docs/agent-qa Update: update the agent-qa dev dependency in your project with your package manager.
- v0.3.2agent-qa v0.3.22026-11-07 · 2026-11-07 08:28 UTC
查看更新说明
What's Changed
- Additional plugins for remote execution infra by @pranshuchittora in https://github.com/vostride/agent-qa/pull/26
Full Changelog: https://github.com/vostride/agent-qa/compare/v0.3.1...v0.3.2
npm package: https://www.npmjs.com/package/agent-qa Docker images: https://hub.docker.com/u/vostride Docs: https://vostride.com/docs/agent-qa Update: update the agent-qa dev dependency in your project with your package manager.
- v0.3.1agent-qa v0.3.12026-11-06 · 2026-11-06 17:45 UTC
查看更新说明
Full Changelog: https://github.com/vostride/agent-qa/compare/v0.3.0...v0.3.1
npm package: https://www.npmjs.com/package/agent-qa Docker images: https://hub.docker.com/u/vostride Docs: https://vostride.com/docs/agent-qa Update: update the agent-qa dev dependency in your project with your package manager.
- v0.3.0agent-qa v0.3.02026-11-06 · 2026-11-06 16:19 UTC
查看更新说明
What's Changed
- Remote runner and first party plugins support by @pranshuchittora in https://github.com/vostride/agent-qa/pull/25
Full Changelog: https://github.com/vostride/agent-qa/compare/v0.2.6...v0.3.0
npm package: https://www.npmjs.com/package/agent-qa Docker images: https://hub.docker.com/u/vostride Docs: https://vostride.com/docs/agent-qa Update: update the agent-qa dev dependency in your project with your package manager.
- v0.2.6agent-qa v0.2.62026-11-03 · 2026-11-03 08:00 UTC
查看更新说明
What's Changed
- Major dependencies upgrade by @pranshuchittora in https://github.com/vostride/agent-qa/pull/24
Full Changelog: https://github.com/vostride/agent-qa/compare/v0.2.5...v0.2.6
npm package: https://www.npmjs.com/package/agent-qa Docker images: https://hub.docker.com/u/vostride Docs: https://vostride.com/docs/agent-qa Update: update the agent-qa dev dependency in your project with your package manager.
用户点评
登录用户的真实使用记录，可按版本和硬件筛选共 0 条
还没有人点评。第一个说出真实体验的人最有价值。
Your AI QA engineer, with memory. Describe a user journey in plain language. agent-qa tests it on web, Android or iOS and keeps the evidence. Use the CLI, the dashboard, or your coding agent through MCP and skills.
- memidex: remembers facts and working procedures as reviewable files in your repository.
- Self-improvement: validates lessons from runs, retires stale knowledge, and uses your feedback to correct mistakes.
| Benchmark | Rank | Published result |
|---|---|---|
| AndroidWorld | #1 | 100% pass@1 (116/116 tasks) |
| AppControlBench | #1 | 99.6% completion (114.5/115 points) |
Read the full benchmark reports, with run evidence and methodology, on our blog. Both used gpt-6-sol with memory and self-improvement off; AppControlBench gives half credit for partial completion.
See it in action
Inspect a failure. The demo catches a cart showing $29 instead of $53, with the failed step and its screenshot side by side.
See what memory contributed. Inspect the knowledge supplied to a run and trace it back to the steps that used it.
Run from your terminal. Real output from the demo suite: checkout passes; the deliberate cart-total bug fails.
Get started
Requires Node.js 24+.
npm install -D agent-qa
npx agent-qa init
Follow the quickstart to connect a model and run the demo, or set up with your coding agent.
[!NOTE] Early stage. Expect bugs. Report an issue when something breaks.