by Echoxiawan
Natural-language mobile UI testing with AI execution on real Android/iOS devices, generating standalone pytest + Appium scripts. 用自然语言写移动端测试用例,AI 在 Android / iOS 真机上执行,跑通后生成可独立运行的 pytest + Appium 脚本。
Available after the catalog security scan passes.
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM lists this project for discovery and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
# Add to your Claude Code skills
git clone https://github.com/Echoxiawan/AstraGuides for using ai agents skills like Astra.
See how Astra compares with popular alternatives.
AI-powered UI automation testing for real Android and iOS devices
English · 简体中文 · User Guide · Contributing · Security
Astra is an open-source UI automation platform for real Android and iOS devices. Testers describe preconditions, actions, and assertions in natural language. Astra combines multimodal AI with Appium to understand the screen, locate elements, and perform actions. Stable cases can be exported as standalone pytest + Appium scripts for repeatable regression without further AI calls.
Astra manages devices, app versions, cases, suites, plans, executions, and reports. Every successful AI run becomes a downloadable, auditable script. Devices, model endpoints, object storage, and runtime data stay under the control of the deployment.
See the user guide for the full UI and workflow.
flowchart LR
C["Natural-language case"] --> A["AI screen understanding"]
A --> E["Appium actions on device"]
E --> R["Step logs / screenshots / assertions"]
E -->|successful run| S["Standalone pytest + Appium script"]
S --> P["Script regression"]
R --> P
P --> D["Execution records and reports"]
Astra has two execution modes:
| Mode | Uses AI | Best for | Script behavior |
|---|---|---|---|
| AI execution | Yes | New cases, exploration, initial script generation | Generates a script after a manual, non-debug single-case success |
| Script execution | No | Stable daily and batch regression | Never generates or overwrites scripts |
app-steps.py and AI case CSV exports into Astra cases and platform scripts.utf8mb4.TEST_USERS is for local development only.git clone <repository-url>
cd Astra
cp backend/.env.example backend/.env
# Fill in MySQL, Redis, AI, and OSS settings
docker compose up -d --build
# Apply migrations when upgrading an existing database
docker compose exec backend alembic upgrade head
Verify the API:
curl http://localhost:8000/api/health
# {"status":"ok","service":"Astra"}
Open http://localhost:3011. Compose starts frontend, backend, celery-worker, and celery-beat; MySQL, Redis, Appium, OSS, and the AI endpoint are external dependencies.
Every host connected to real devices must run Appium and the device agent. Both live in runtime/ at the repository root:
| Entry | Purpose | Port |
|---|---|---|
runtime/appium.sh |
Install, start, stop, and diagnose Appium 2 | 4723 |
runtime/device_agent.py |
Device discovery, ADB/iOS operations, WDA, and Keychain | 5070 |
runtime/celery.sh |
Worker and Beat for non-Docker deployments | None |
runtime/kill_celery.sh |
Force-stop Celery Worker and Beat | None |
First Appium start:
./runtime/appium.sh install
./runtime/appium.sh start
Start the device agent in another terminal:
source .venv/bin/activate
python runtime/device_agent.py
Verify both services:
curl http://127.0.0.1:4723/status
curl http://127.0.0.1:5070/health
In the UI, open Device Center → Appium Servers → Add Server and enter an Appium URL reachable by both the Astra API and Worker, such as http://192.0.2.10:4723. Astra derives the device agent on port 5070 from the same hostname; there is no separate agent URL field.
Test Connection checks only Appium
/status. The device agent may still be down. Run Manual Sync on the device page and confirm that real devices appear. See the runtime guide.
python3 -m venv .venv
source .venv/bin/activate
pip install -r backend/requirements.txt
cp backend/.env.example backend/.env
Initialize the database and start the API:
cd backend
alembic upgrade head
python run.py
Start the frontend in another terminal:
cd frontend
npm install
npm run serve
For non-Docker deployments, start Celery Worker and Beat:
./runtime/celery.sh start all
./runtime/celery.sh status
Astra imports two artifact types exported by Echoxiawan/echobench:
| File | Echobench export | Purpose |
|---|---|---|
*-test-case.csv |
"AI Case" export | Create and bind the target case |
app-steps.py |
"Export Script → Python (Appium)" | Import an executable platform script |
The CSV columns are exactly 用例标题,序号,操作步骤,断言 (case title, step number, operation, assertion). Each row is one step; assertions may be empty. The PY file contains PLAN = json.loads("..."); Astra only extracts that constant and never executes code from the file.
Open Test Cases → Import Script.
app-steps.py and link an existing case.app-steps.py, choose "Create case from CSV and bind", then upload the CSV and fill in the app, module, preconditions, and postconditions. Astra creates the case first and then binds the script.The UI previews each step's locator and assertion. Coordinate locators, missing element locators, and unsupported actions are reported explicitly. Re-importing the same platform and language for a case replaces the previous imported script; AI-generated scripts are unaffected.
See backend/.env.example for all documented defaults. Common variables:
| Variable | Default | Purpose |
|---|---|---|
DATABASE_URL |
mysql+pymysql://astra:change-me@localhost:3306/astra |
MySQL connection |
REDIS_URL |
redis://localhost:6379/2 |
Cache and inter-process notifications |
CELERY_BROKER_URL |
redis://localhost:6379/3 |
Celery broker |
CELERY_RESULT_BACKEND |
redis://localhost:6379/4 |
Celery result backend |
CELERY_QUEUE_NAMESPACE |
empty | Queue namespace for shared Redis deployments |
OPENAI_CONFIGS |
empty | Multiple OpenAI-compatible providers as single-line JSON |
OPENAI_API_KEY |
empty | API key for the single-provider configuration |
OPENAI_API_BASE |
empty | Model endpoint base URL |
OPENAI_MODEL |
gpt-4o |
Vision model name |
OSS_ACCESS_KEY_ID |
empty | Alibaba Cloud OSS access key ID |
OSS_ACCESS_KEY_SECRET |
empty | Alibaba Cloud OSS access key secret |
OSS_REGION |
cn-shenzhen |
OSS region |
OSS_BUCKET_NAME |
empty | OSS bucket |
LDAP_SERVER |
ldap://localhost:389 |
LDAP endpoint |
JWT_SECRET_KEY |
change-me-in-production |
JWT signing key; replace in production |
TEST_USERS |
empty | Local users that bypass LDAP; disable in production |
BASE_URL |
http://localhost:8000 |
External URL for reports and package downloads |
DEFAULT_APPIUM_URL |
http://localhost:4723 |
Default Appium node |
IOS_XCODE_ORG_ID |
empty | Global fallback Apple Team ID |
IOS_XCODE_SIGNING_ID |
Apple Development |
Global fallback signing identity |
IOS_KEYCHAIN_PASSWORD |
empty | Legacy global Keychain password fallback |
Never commit backend/.env or write OSS, LDAP, AI, or device credentials into images, logs, or issues.
Astra/
├── runtime/ # Independent Appium, device-agent, and Celery entries
│ ├── appium.sh
│ ├── device_agent.py
│ ├── celery.sh
│ └── kill_celery.sh
├── backend/
│ ├── app/api/ # HTTP API and WebSocket
│ ├── app/core/ # Execution engine, assertions, sessions, AI, logging
│ ├── app/services/ # Devices, AI, OSS, reports, and domain services
│ ├── app/tasks/ # Celery Worker and Beat tas
Astra is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by Echoxiawan. Natural-language mobile UI testing with AI execution on real Android/iOS devices, generating standalone pytest + Appium scripts. 用自然语言写移动端测试用例,AI 在 Android / iOS 真机上执行,跑通后生成可独立运行的 pytest + Appium 脚本。. It has 60 GitHub stars.
Astra's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/Echoxiawan/Astra" and add it to your Claude Code skills directory (see the Installation section above).
Astra is primarily written in Python. It is open-source under Echoxiawan on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh Astra against similar tools.
No comments yet. Be the first to share your thoughts!