新设备 2FA 提示跳过
在消费型的fintech应用中，登录路由变更绕过了新设备的验证步骤。
一个安全检查站被悄悄跳过。
Mo是AI QA的工程师，测试你的应用结束。 校验临界流量，从一个提示中找到错误。
由主要工程团队信任。
创建的测试 77 813
已执行测试 70 607 819
经核实的变动 339 166
80 290份经核实的公关文件
自动治愈 8 932 104
虫子抓到117 010个
每天由运输团队进行生产。
1
页：1
5
代码移动更快。 核查必须跟上。
编码agents使得构建软件的速度大大加快。 瓶颈现在知道它的作用。 校验属于agent环。 Agents需要应用程序本身提供的证据，以验证其变化能传达预期的行为。
探索
写入您需要验证的用户流。 对你的应用进行测试重新使用成功的AI行动，并审查作为应用改变而提出的修复建议。
从编辑器到CI的验证。
用英语写步骤和断言
在您的仓库中审查并版本测试。
01 / 作者数
描述一个用户流及其以简单英语表达的预期结果。 保留 YAML 文件中的步骤和断言您的团队和编码 agent 可以审查。
标准用户购买。 test.yaml - 瞬间
存储前端
测试
模块
1 文件类型：瞬间/测试/v2
2id：标准用户检查退出
3url: https://shop.momentic.ai (中文(简体) ).
4description：端到端的检出流程
5个步骤：
6 - 模块：./登录用户名-密码.module.yaml
7 - 行为：把瞬间背包加进推车，然后去退房
8 - 模块：./填出个人-info.module.yaml
9 - 断言：项目总额与所列每个项目的数额相符
10 - 点击：完成按钮
瞬间运行
02 运行
用CLI来检查你正在更改的应用程序。 在下次合并或部署前在 CI 中运行相同的测试。
03次/细分
看哪个步骤失败了，检查它的截图和痕迹。 AI分解调查原因，提出测试修补方案供您审查。
瞬态 AI 分类 - 交互
询问关于此 runctrl- c 的后续信息以退出
github.com (英语).
测试/标准用户-购买。 test.yaml+1-1
- 3 -5步骤
- 模块：./填出个人-info.module.yaml
- 点击：完成按钮
+ - 点击：位置顺序按钮
暂时性治疗
04/治愈
当UI改变时，Momentic 发现再次控制。 永久愈合建议测试文件编辑，以供审查。 应用 diff 并重运行以检查结果。
描述您的产品术语、商业规则，Agents使用这个上下文来选择动作和检查结果。
向agents提供他们检查流量所需的产品规则。
测试运行时看到的地图屏幕和用户流量。
审查会议的建议，然后将其纳入知识库。
受到工程团队的喜爱。
Poe.com的AI回应即使不是决定性的
Momoko F. Quora产品业务主管
30分钟
从7小时起每日执行测试
500+
手动测试案例已替换
100%
在一个月内进行的重要测试
用SSO和自定义数据政策测试私人应用程序。 获得专注登机和优先支持。 运动是SOC 2型认证。
工程团队使用Momentic.
01 SOC 2 类型 2
独立审计。 信托中心的合规报告。
02 合同正常
服务范围和营业时间承诺由您的企业协议规定。
SSO和RBAC
SAML和OIDC SSO，行政和成员角色，以及审计日志。
04 优先支助
优先支持和企业账户专用账户管理人。
Momentic发现应用失败的例子，有受影响的流量及其影响。
生产前117 010
在 Momentic 管理的浏览器上运行您的测试。 知识库也普遍可用，分类，分化，AI测试选择也放弃了他们的β标签。
Mo在云中运行自主的QA会话，在网络上或托管Android和iOS设备，并配有您的Org的MCP服务器和会话所附带的秘密。
每一个v5定位器和断言查询，现在都根据它需要多少推理，将自身引向快速、中或延伸的努力，这也确定了它的信用用途。
生成的 ids 已经正常化，并且从缓存选择器中忽略了源文件属性，因此重建您的应用不再错过每一个步骤。
一个AI动作现在可以缓存一个分支，因此只有显示的饼干横幅将一半的时间从缓存重放，而不是再次唤醒agent.
一个单独的评论员在AI动作通过之前，查看应用程序的实际结局，在你的目标没有实现时，将步骤带回它的推理。
Triage能查出每个测试都失败的原因它负责修复你的分支，然后打开拉动请求对于其他的事业，并要求谁拥有它去审查。
在主机模拟器上运行 Android 测试，没有 Android Studio, SDK，或者安装在您的机器上的adb.
用普通的英语说，在跑步的某个时候，什么应该是真实的。 瞬间根据录像的画面来检查一下烤面包片能闪亮一秒钟
将您的 agents 的应用错误提交.momentic/ 技能，以及探索，分解和结果分类 agents 运行时读取。
importic doctor告诉你设置有什么问题：CLI和Node版本，签入，浏览器，项目配置，以及与Momentic的连接。 --fix 清除已贬值的选项。
以 API 键拉动完成的运行并运行 REST 上的组。 通过它们通过cursor，并按日期，状态，或平台过滤。
捕捉到解析的测试到可移植的快照中，然后用瞬间运行——从-snapshot——在孤立中重放。
imeic ai 探索现在支持更深入的规划控制，JSON输出，以及产品bug报告带有repro步骤。
运行现在暴露了直播进度，结构化的JSON输出，以及更清晰的运行查看器对缓存状态，复刻，条件，和循环的诊断。
V3 agent现在是各地的默认。 它计划了前方的流量，缓存解决的步骤，和自我愈合失败。
瞬息万变的ai 分解组失败成标签的桶，预览计划以干运行，并且可以隔离它无法修复的东西。
新的项目脚手架采用更清洁的v2文件格式，瞬间升级迁移现有项目结束。
Pass -- 并行自动和移动大小从您的主机的CPU计数中计算出更快的本地运行和CI运行。
在一个测试中提供带有电子邮件.create()的ephemer inbox 和从您的 Org 游泳池租来的号码。
恢复流量可以检查浏览器状态，运行步骤，重试失败步骤，并刷新 stale缓存，每步可见。
motiic AI 将运行从它的 ID 或 URL 分类，这样您就可以将每个失败类别引导到正确的动作。
Android AI行动现在在前方规划，缓存解步骤，以及自愈，匹配网络体验。
iOS和Android测试运行期间自动捕获屏幕录音，并随运行尝试一起存储。
从Claude Code或Cursor为iOS和Android创建测试和模块，运行步程，并驱动模拟器会话。
AI动作默认是启用的，因此在imeric.config.yaml中不再需要ai.aiAction开关。
移动测试获得专用设置和拆卸部分，从而共享先决条件和清理在一个地方进行。
作者和运行了针对iOS的Momentic测试，带有模块，AI动作，并和Android并肩主持模拟器。
启动和终止浏览器会话，切换步骤，运行步骤范围从您的编码 agent 到 stdio MCP 服务器。
新建请求倾听器，请求录制，以及模拟路由步骤类型让agents捕获和 stub网络流量。
Run Viewer现在绘制系统图，处理执行过程中捕获的CPU和内存，以帮助调试flaky运行。
故障回收现在扩大到移动测试、重试和调整iOS和Android的瞬态问题。
瞬间结果查看本地运行的查看器发布在您的机器上浏览结果而不上传任何东西。
在撰写测试时通过移动编辑器中的编辑后退一步。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
New-device 2FA prompt skipped
At a consumer fintech app, a login routing change bypassed the new-device verification step.
A security checkpoint was silently skipped.
Mo is an AI QA engineer that tests your app end to end. Verify critical flows and find bugs from a prompt.
Trusted by leading engineering teams.
Tests created 77,813
Test runs executed 70,607,819
Changes verified 339,166
PRs verified 80,290
Auto-heals 8,932,104
Bugs caught 117,010
Trusted in production by teams shipping every day.
1
of
5
Code moves faster. Verification has to keep up.
Coding agents have made it dramatically faster to build software. The bottleneck is now knowing it works. Verification belongs in the agent loop. Agents need evidence from the app itself to verify that their changes deliver the intended behavior.
Explore
Write the user flows you need to verify. Momentic runs the tests against your app. Reuse successful AI actions and review proposed repairs as the app changes.
Verification from your editor to CI.
Write steps and assertions in plain-English YAML.
Review and version the tests in your repository.
01 / author
Describe a user flow and its expected result in plain English. Keep the steps and assertions in YAML files your team and coding agent can review.
standard-user-purchases.test.yaml - momentic
storefront
tests
modules
1fileType: momentic/test/v2
2id: standard-user-checkout
3url: https://shop.momentic.ai
4description: End-to-end checkout flow
5steps:
6 - module: ./log-in-username-password.module.yaml
7 - act: add the momentic backpack to the cart and go to checkout
8 - module: ./fill-out-personal-info.module.yaml
9 - assert: the item total matches the amount listed for each item
10 - click: Finish button
momentic run
02 / run
Use the CLI to check the app you're changing. Run the same tests in CI before the next merge or deploy.
03 / triage
See which step failed and inspect its screenshots and traces. AI triage investigates the cause and proposes a test repair for you to review.
momentic ai classify --interactive
Ask a follow-up about this runctrl-c to exit
github.com
tests/standard-user-purchases.test.yaml+1−1
@@ checkout flow · steps 3-5 @@
- module: ./fill-out-personal-info.module.yaml
- - click: Finish button
+ - click: Place Order button
momentic ai heal
04 / heal
Momentic finds controls again when the UI changes. Permanent healing proposes test-file edits for review. Apply the diff and rerun to check the result.
Describe your product’s terminology, business rules, and expected behavior in the Knowledge base. Agents use that context to choose actions and check outcomes.
Give agents the product rules they need to check a flow.
Maps screens and user flows seen during test runs.
Review session suggestions before adding them to the Knowledge base.
Loved by engineering teams.
“Momentic gave us a fast and reliable way to validate Poe.com's AI responses, even when they weren't deterministic.”
Momoko F. Head of Product Operations, Quora
30 min
daily test execution, down from 7 hours
500+
manual test cases replaced
100%
critical tests created in one month
Test private apps with SSO and custom data policies. Get dedicated onboarding and priority support. Momentic is SOC 2 Type 2 certified.
Engineering teams using Momentic.
01 SOC 2 Type 2
Independently audited. Compliance reports in the Trust Center.
02 Contractual uptime SLA
Service scope and uptime commitments are set in your Enterprise agreement.
03 SSO & RBAC
SAML and OIDC SSO, Admin and Member roles, and audit logs.
04 Priority support
Priority support and a dedicated account manager for enterprise accounts.
Examples of application failures found by Momentic, with the affected flow and its impact.
117,010 before production
Run your tests on browsers Momentic manages. The knowledge base is generally available too, and classification, triage, and AI test selection dropped their beta labels.
Mo runs autonomous QA sessions in the cloud, on web or hosted Android and iOS devices, with your org's MCP servers and secrets attached to the session.
Each v5 locator and assertion query now routes itself to a fast, medium, or extended effort based on how much reasoning it needs, which also sets its credit usage.
Generated ids are normalized and source-file attributes are left out of the cached selector, so rebuilding your app no longer misses every step.
An AI action can cache a branch now, so the cookie banner that only shows up half the time replays from cache instead of waking the agent again.
A separate reviewer looks at where the app actually ended up before an AI action passes, and hands the step back with its reasoning when your goal did not happen.
Triage works out what broke each test. It commits the fixes your branch caused, then opens a pull request for every other cause and asks whoever owns it to review.
Run Android tests on hosted emulators without Android Studio, the SDK, or adb installed on your machine.
Say in plain English what should have been true at some point in a run. Momentic checks it against frames from the run video, so a toast that flashes for a second still counts.
Commit what your agents keep getting wrong about your app to .momentic/skills, and the explore, triage, and result-classification agents read it while they run.
momentic doctor tells you what is wrong with your setup: CLI and Node versions, sign-in, browsers, project config, and the connection to Momentic. --fix clears out deprecated options.
Pull your finished runs and run groups over REST with an API key. Page through them by cursor and filter by date, status, or platform.
Capture resolved tests into portable snapshots and replay them in isolation with momentic run --from-snapshot.
momentic ai explore now supports deeper planning controls, JSON output, and product bug reports with repro steps.
Runs now expose live progress, structured JSON output, and clearer Run Viewer diagnostics for cache status, retries, conditionals, and loops.
The V3 agent is now the default everywhere. It plans the flow up front, caches the resolved steps, and self-heals on failure.
momentic ai triage groups failures into labeled buckets, previews the plan with a dry run, and can quarantine what it cannot fix.
New projects scaffold in the cleaner v2 file format, and momentic upgrade migrates existing projects end to end.
Pass --parallel auto and Momentic sizes the worker pool from your host's CPU count for faster local and CI runs.
Provision ephemeral inboxes with email.create() and lease numbers from your org pool with sms.lease() right inside a test.
The recovery flow can inspect browser state, run steps, retry the failed step, and refresh stale caches, with per-step visibility.
momentic ai classify categorizes a run from its ID or URL so you can route each failure category to the right action.
Android AI actions now plan up front, cache resolved steps, and self-heal, matching the web experience.
Screen recordings are captured automatically during iOS and Android test runs and stored with the run attempt.
Create tests and modules, run step ranges, and drive emulator sessions for iOS and Android from Claude Code or Cursor.
AI actions are enabled by default, so the ai.aiAction switch is no longer needed in momentic.config.yaml.
Mobile tests gain dedicated setup and teardown sections so shared preconditions and cleanup live in one place.
Author and run Momentic tests against iOS, with modules, AI actions, and hosted simulators alongside Android.
Start and terminate browser sessions, splice steps, and run step ranges from your coding agent over a stdio MCP server.
New request listener, request recording, and mock route step types let agents capture and stub network traffic.
The Run Viewer now charts system and process CPU and memory captured during execution to help debug flaky runs.
Failure recovery now extends to mobile tests, retrying and adapting around transient issues on iOS and Android.
momentic results view launches a local run viewer to browse results on your machine without uploading anything.
Step back and forward through edits in the mobile editor while authoring tests.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.