JC Meteor钧宸流星
中文

MeteorTest · MVP, active development

JC Meteor钧宸流星

Mobile engineering with release quality built in.

I build iOS and cross-platform apps, the API contracts they depend on, and the automated checks that keep each release dependable.

Engineering portfolio of Pengyu Qi (Connor).

See the mobile work Open MeteorTest Web Preview GitHub
MeteorTest

A testing control plane that runs the suite and keeps the evidence.

  • task queue
  • local executor
  • Allure reports
Sprout

One product, three repositories, one interface contract.

  • SwiftUI
  • uni-app x
  • FastAPI contract
TankQuest Academy

A 2.5D learning game with backend-authoritative progress.

  • Phaser
  • Tauri preview
  • parent report
Latest validated mobile suite
Task
phase9-api-smoke-001
Suite
iOS-Automation-Framework · api_smoke
Result
6 passed · exit 0
Artifacts
output.log + Allure
Runtime
private Local Agent
Stack in daily use
SwiftUIuni-app xHarmonyOS NEXTFastAPILangGraphAppiumpytestAllurePhaserTauriNext.jsTypeScriptSwiftUIuni-app xHarmonyOS NEXTFastAPILangGraphAppiumpytestAllurePhaserTauriNext.jsTypeScript
Screens

What the products actually look like.

Captured from the running builds, not mockups.

MeteorTest · workspace
MeteorTest · workspaceProjects, suites, tasks, and artifacts in one console.
MeteorTest · reports
MeteorTest · reportsRun status, logs, and Allure artifacts per task.
TankQuest · missions
TankQuest · missionsTwelve missions across math, English, and direction drills.
TankQuest · battle
TankQuest · battleArcade physics with an in-battle math challenge.
By the numbers

What is actually in the repositories.

Counted from the public GitHub repositories on 31 October 2026. Private repositories are not included.

626commits
10public repositories
3.0 MBtracked source
6smoke tests passing

Where the code actually lives

  • TypeScript65%
  • Swift13%
  • Python9.5%
  • JavaScript4.6%
  • CSS3.9%
  • Other4%
How it is built

Three scenes, one product.

Scroll to move through the build.

01 · The app

A SwiftUI client that keeps working offline

Recording, playback, local storage and offline practice flows, with mock services marked explicitly in Debug builds so nothing pretends to be real data.

  • iPhone + iPad
  • SwiftUI
  • offline practice
02 · The contract

One interface contract, three repositories

EC-API-0.1 is versioned and hash-distributed to both clients, so the iOS app, the uni-app x app and the FastAPI backend move together instead of drifting apart.

  • EC-API-0.1
  • uni-app x
  • FastAPI + LangGraph
03 · The release check

The suite that decides it is safe to ship

Appium and pytest suites produce real pass/fail results and Allure artifacts before a build leaves the machine, and the same run is recorded on this page.

  • 6 passed
  • exit 0
  • Allure artifacts
System

How the three Sprout repositories fit together

One interface contract, versioned and hash-distributed, keeps the iOS client, the cross-platform client and the backend in step.

iOS clientSproutIOSSwiftUI · iPhone + iPad
Cross-platformSproutCrossPlatformuni-app x · mini program, Web, Android, HarmonyOS NEXT
BackendSproutApiFastAPI + LangGraph · practice, check-ins, rewards
Release quality

MeteorTest

Automation testing platform for projects, suites, local executors, reports, and AI-assisted operations.

MVP / active development

A testing control plane that connects project contracts, AI-assisted operations, local execution, and report context.

Open MeteorTest Open Web Preview Try demo GitHub
MeteorTest Web dashboard preview screenshot
Projectyunlu-ios
Suiteapi_smoke
Executorphase9-local-agent
Reportoutput.log + Allure
AI Operation Flow
  1. ImportProject contract and suites
  2. PrepareTask, environment, build
  3. ExecutePrivate Local Agent
  4. AnalyzeReport context and next action
Validation and preview status
Local validationSucceeded
Smoke suite6 passed
Web previewOnline
ExecutionPrivate Agent
Preview online
Project substance

Why MeteorTest exists

MeteorTest is built around one practical loop: let a platform or AI assistant import test projects, register suites, create tasks, trigger local execution, and bring logs, reports, and failure context back into one place.

Problem

  • Automation scripts often live in separate repositories without a shared entry point.
  • Task execution is easy to start manually, but hard to trace later.
  • Reports, logs, app artifacts, environments, and task context are usually disconnected.
  • AI assistants are less useful when they can only comment on logs instead of operating with project, suite, task, and report context.

What it does

  • Imports projects and suites from a test project's meteortest.yml contract.
  • Lets the AI assistant help create tasks, select suites, inspect reports, and summarize failures from platform context.
  • Lets a Python Local Agent claim tasks and execute pytest, Appium, or Locust commands.
  • Collects status, logs, Allure artifacts, and AI-assisted operation history.

Current status

  • MVP and active development.
  • The platform-to-agent-to-test-repository execution path has been validated with iOS-Automation-Framework.
  • The API smoke suite can now produce real pass/fail results against the iOS-Automation-Framework local mock API.
  • The MeteorTest Web preview is online at meteortest.jcmeteor.com; Local Agent execution remains private.

Validation progress

  • The public site keeps the demo interactive and mock-data based.
  • The separate MeteorTest Web preview is online for console surface validation.
  • Use the local mock API as public-safe validation results for API smoke assertions.
  • MeteorTest Local Agent has run the same mock-backed smoke suite and collected task-specific logs/report artifacts.
Validated local run

MeteorTest ran the mock-backed API smoke suite locally

A MeteorTest Local Agent task executed iOS-Automation-Framework `api_smoke` against the deterministic local mock API and collected task-specific logs plus Allure result artifacts.

Taskphase9-api-smoke-001
ResultSucceeded
Pytest6 passed
Exit code0

Run summary

  • Suite: api_smoke
  • Environment: local-mock-api
  • Runtime: iOS-Automation-Framework/.venv on Python 3.13
  • Selection: 6 smoke cases selected, 16 non-smoke cases deselected
  • Artifacts: output.log and Allure results collected under task-specific report paths

The public website still uses a browser-side mock demo, and the live Web preview keeps execution private. Public connected execution is a later design topic after authentication, data isolation, and executor safety are designed.

The test code behind it

iOS-Automation-Framework

The first practical test-code carrier and platform integration sample for MeteorTest.

GitHub
project: yunlu-ios
suites:
- id: api_smoke
- command: python -m pytest
report: allure
Work

Projects, with the state they are actually in.

Each entry links to public code and states how far it has been validated. Several repositories behind these systems are still private.

ActiveSwiftUI / uni-app x / FastAPI

Sprout

An English speaking practice app built as three cooperating repositories: a SwiftUI iOS client, a uni-app x cross-platform client, and a FastAPI + LangGraph backend.

View details
01 / 12
Engineering focus

What this site is about

Automation testing platforms

Build control planes that connect projects, suites, tasks, local executors, logs, reports, and follow-up actions.

Project contractsTask queuesReport context

iOS automation infrastructure

Keep UI automation and API smoke tests maintainable through clear fixtures, runtime isolation, and platform integration contracts.

pytest/AppiumAllure outputRuntime ownership

AI-assisted engineering workflows

Use AI as an operation partner that can import projects, prepare tasks, inspect results, guide voice practice, and suggest concrete next actions.

Agent guardrailsVoice feedbackHuman review

Voice-first learning products

Build spoken interaction loops where ASR, endpointing, TTS, session state, and feedback controls stay observable across web and mobile.

ASR provider layerTTS routingMobile session control

Repeatable local execution

Prefer small, observable execution loops that can run locally first, then become website validation results or platform workflows later.

Local-first validationEnvironment checksDemo readiness
Contact & collaboration

Open project channels first

JC Meteor is currently organized around public engineering work. GitHub is the best place to understand the projects, follow progress, and discuss project-specific questions.

Good topics to discuss
  • Automation testing platforms and local execution loops.
  • Voice AI practice products, ASR/TTS provider routing, and mobile session control.
  • iOS automation, pytest/Appium integration, and test infrastructure.
  • AI-assisted development workflows that can operate on real project context.
GitHub ProfileStart here for the public project hub, repositories, and profile updates.Open link MeteorTest IssuesUse this for platform questions, bugs, roadmap feedback, and testing workflow ideas.Open link MeteorVoice IssuesUse this for voice practice, ASR/TTS routing, mobile session, and speech feedback questions.Open link iOS Automation IssuesUse this for test project integration, pytest/Appium, and sample execution questions.Open link

For private details such as credentials, internal URLs, device identifiers, or test accounts, use a private channel when one is published.