testerarmy-cli
Use TesterArmy CLI to create, organize, schedule, and run dashboard-managed QA tests in TesterArmy cloud. Trigger when defining regression coverage, validating a flow, exploring an app for bugs, or inspecting remote test results and issues.
How do I install this agent skill?
npx skills add https://github.com/tester-army/cli --skill testerarmy-cliIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill provides a CLI interface for TesterArmy cloud QA testing and contains a minor vulnerability surface for indirect prompt injection.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · 1 issue
- Runlayerfail
3/3 files flagged
What does this agent skill do?
TesterArmy CLI
Create dashboard-managed QA coverage with ta / testerarmy.
The CLI is a control plane for the TesterArmy dashboard: manage projects, environments, credentials, and saved tests, and queue runs that execute in TesterArmy cloud. Use a cloud-reachable preview or tunnel environment to validate development changes.
When to Use
- Create persistent QA coverage for a feature or product area.
- Convert acceptance criteria into dashboard tests.
- Add or update smoke, regression, auth, billing, onboarding, or mobile flows.
- Explore an area for bugs without a saved test, or triage the issues runs found.
Setup
Check auth:
ta status --json
If needed:
ta auth
ta auth --api-key <key>
TESTERARMY_API_KEY=<key> ta status --json
ta signout
ta logout
Most commands that call the API accept --api-key <key> and --base-url <url>.
Use env vars or saved auth by default.
Discover scope:
ta projects list --json
ta projects get <projectId> --json
ta groups list --project <projectId> --json
ta tests list --project <projectId> --json
Use IDs exactly as returned.
ta docs <topic> prints the full reference for a topic: projects,
memories, tests, groups, runs, issues, discover, ci.
Project Context
Create a project:
echo '{"name":"Example","url":"https://example.com","projectType":"web"}' | ta projects create --json
echo '{"name":"Example staging"}' | ta projects update <projectId> --json
ta projects files <projectId> --json
ta projects files-upload <projectId> --file avatar.png --description "Profile photo" --json
Project commands: list, get, create, update, delete, environments,
environments-create, environments-delete, credentials,
credentials-create, files, files-upload, apps, apps-delete.
Project files (max 50 MB) are inputs runs can use, such as uploads for files
steps. On mobile projects, photos and videos are preloaded into the device
library.
Manage saved environments (remote run targets such as Staging or QA):
ta projects environments <projectId> --json
ta projects environments-create <projectId> --name Staging --url https://staging.example.com --json
ta projects environments-delete <projectId> <environmentId> --json
Production and PR Preview are built-in environments and cannot be deleted.
Target a saved environment on runs with --env <nameOrSlug> or
--project-environment-id <id>.
Delete projects only when explicitly requested:
ta projects delete <projectId> --json
Store app knowledge:
ta memories list --project <projectId> --json
echo '{"category":"site_structure","title":"Auth route","content":"Login is at /login","importance":"high"}' | ta memories create --project <projectId> --json
ta memories delete <memoryId> --project <projectId> --json
Memory categories: site_structure, test_insights, user_preferences.
Create credentials for login or inbox flows:
ta projects credentials <projectId> --json
echo '{"kind":"login","label":"Admin","username":"admin@example.com","password":"secret"}' | ta projects credentials-create <projectId> --json
echo '{"kind":"inbox","label":"Signup inbox"}' | ta projects credentials-create <projectId> --json
Never print real secrets in final messages.
Tests
Create:
echo '{"title":"Login flow","description":"User can sign in and reach the dashboard","steps":[{"title":"Navigate to /login","type":"act"},{"title":"Sign in with the saved admin credentials","type":"login","credentialId":"<credentialId>"},{"title":"Dashboard loads and shows the project list","type":"assert"}]}' | ta tests create --project <projectId> --json
Create in a group:
echo '{"title":"Pricing CTA","steps":[{"title":"Open /pricing","type":"act"},{"title":"Click the primary CTA","type":"act"},{"title":"Signup or dashboard flow starts","type":"assert"}]}' | ta tests create --project <projectId> --group <groupId> --json
Payload:
{
"title": "string, required",
"description": "string, optional",
"platform": "web or mobile, optional",
"steps": [
{ "title": "User action", "type": "act" },
{ "title": "Expected result", "type": "assert" },
{ "title": "Login instruction", "type": "login", "credentialId": "uuid" },
{ "title": "Use temporary email", "type": "login", "temporaryEmail": true },
{ "title": "Upload the sample invoice PDF", "type": "files" },
{ "title": "Screenshot label", "type": "screenshot" }
]
}
Write good steps:
- Write plain-language instructions you would give to a teammate.
- Describe the user's intent, not selectors, waits, DOM structure, or component names.
- Give each step one clear job with a clear stopping point.
- Split actions and checks into separate steps.
- Use labels visible in the UI, such as
Pricing,New Project,Save, orInvite teammate. - Include business context that disambiguates the goal, such as project names, emails, plan names, or expected statuses.
- Make assertions specific enough to pass or fail confidently.
- Quote exact copy only when exact copy matters; otherwise describe the expected meaning.
Examples:
| Avoid | Write |
|---|---|
Click the button with class .nav-item:nth-child(3), wait 500ms, then assert URL contains /pricing. | Open the Pricing page from the header. |
Log in, create a project, invite a teammate, run a test, and verify the result. | Split into focused steps: log in, create the project, invite the teammate, run the test, then verify the result. |
Add the Pro plan to the cart and make sure the cart is correct. | Add the Pro plan to the cart. then Verify the cart shows the Pro plan with the correct monthly price. |
Verify the dashboard looks good. | Verify the dashboard shows a Projects card, a Recent Runs card, and no visible error banner. |
Before creating or updating a test, check:
- Could a teammate follow these steps without seeing the code?
- Does each step have one clear intent?
- Are actions and assertions split?
- Are user-visible labels included where useful?
- Are credentials handled by a
loginstep or saved project credentials? - Is each expected result specific enough to evaluate?
Rules:
- Cover one user journey.
- Prefer 3-10 meaningful steps.
- Use
actfor navigation, clicks, typing, and other user actions. - Use
assertfor visible outcomes, persisted state, email delivery, or URL changes. - Use
loginwithcredentialIdortemporaryEmail; do not put passwords in step titles. - Use
filesfor uploading attached files. - Use
screenshotonly for important visual checkpoints. - Do not hide login, uploads, screenshots, or assertions inside broad action steps.
- Maximum 50 steps per test.
Inspect before changing:
ta tests list --project <projectId> --json
ta tests get <testId> --json
Update title, description, steps, or enabled:
echo '{"title":"Updated login smoke"}' | ta tests update <testId> --json
echo '{"steps":[{"title":"Open /login","type":"act"},{"title":"Sign in","type":"login","credentialId":"<credentialId>"},{"title":"Dashboard is visible","type":"assert"}]}' | ta tests update <testId> --json
echo '{"enabled":false}' | ta tests update <testId> --json
Replacing steps requires the complete array.
Toggle a test without deleting it:
ta tests disable <testId> --json
ta tests enable <testId> --json
Delete only when explicitly requested:
ta tests delete <testId> --json
Groups
Create suites:
ta groups list --project <projectId> --json
ta groups get <groupId> --json
echo '{"projectId":"<projectId>","name":"Smoke"}' | ta groups create --json
echo '{"name":"Core smoke"}' | ta groups update <groupId> --json
ta groups add-test <groupId> <testId> --json
ta groups remove-test <groupId> <testId> --json
Delete non-default groups only when explicitly requested:
ta groups delete <groupId> --json
Common groups: Smoke, Auth, Core journeys, Mobile smoke.
Run a group on a recurring schedule (web projects only). Use --frequency
(every_1h, every_3h, every_6h, every_12h, daily, weekly) with
--minute, --hour, and --day-of-week, or --cron for anything else:
ta groups schedule <groupId> --json
ta groups schedule-set <groupId> --frequency daily --hour 8 --timezone Europe/Warsaw --json
ta groups schedule-set <groupId> --cron "0 8 * * 1-5" --project-environment-id <environmentId> --json
ta groups schedule-delete <groupId> --json
A group has at most one schedule; schedule-set replaces it. Schedules run
every enabled test in the group against Production unless
--project-environment-id is set. Timezone defaults to UTC.
Runs
ta tests run queues the saved test in TesterArmy cloud. Add --wait for
validation; without it, exit code 0 only confirms that the run was queued.
ta tests run <testId> --wait --json
ta tests run <testId> --env staging --wait --json
ta tests run --group <groupId> --project <projectId> --wait --json
ta tests run <testId> --platform ios --app-id <appId> --wait --json
ta tests run <testId> --platform android --app-id <appId> --wait --json
Development targets must be reachable from the cloud browser. Save a preview or tunnel URL as an environment, then select it on the run:
ta projects environments-create <projectId> --name Development --url https://dev.example.com --json
ta tests run <testId> --env development --wait --json
Run options:
--env <nameOrSlug>or--project-environment-id <id>: select a saved project environment; do not combine them.--url <url>: target URL override for group runs; do not combine it with--env.--wait-timeout <ms>and--wait-interval <ms>: control how long to wait and the initial polling interval.--platform web|ios|android,--app-id, and--device-model: select the runtime and mobile build/device.- Group runs can use
--environment production|staging|previewto label the run origin. - Single-test runs can use
--mode fast|deep.
Save a local result artifact with shell redirection:
ta tests run <testId> --env development --wait --json > result.json
Runs:
ta runs list --project <projectId> --json
ta runs get <runId> --json
ta runs wait <runId> --timeout 600000 --json
ta runs messages <runId> --json
ta runs telemetry <runId> --json
ta runs cancel <runId> --json
ta runs recording <runId> --output ./recordings/
To debug a failed run: ta runs get first, then ta runs messages for what
the agent did, ta runs telemetry for console and network errors, and
ta runs recording for the screen recording. Without --output, recording
prints a short-lived MP4 link.
Group runs share a batch ID:
ta batches get <batchId> --json
ta batches cancel <batchId> --json
ta batches rerun <batchId> --json
Export JUnit XML for CI reporters or TestRail. It fails until every run
finishes unless --allow-incomplete is set:
ta runs junit <runId> --output report.xml
ta batches junit <batchId> --output report.xml
Discovery and Issues
ta discover queues a free-roam run: the agent explores toward a goal with no
saved test and reports bugs. A FAILED verdict means bugs were found. With
--wait, the command exits 1 for every outcome except a pass: bugs found
(FAILED), blocked, or cancelled. Read the run verdict in the --json output to
tell them apart.
ta discover --project <projectId> --goal "Explore checkout and find bugs" --wait --json
ta discover --project <projectId> --goal "Explore settings" --target-url https://staging.example.com --wait --json
ta discover --project <projectId> --goal "Explore onboarding" --platform android --wait --json
--target-url is web only. Mobile projects default to iOS and explore the
latest uploaded app; --app-id picks another build and needs
--platform ios|android. --device-model picks the device.
Found bugs land in issues, deduplicated across runs and tracked per environment:
ta issues list --project <projectId> --status open --json
ta issues list --project <projectId> --environment staging --limit 20 --json
ta issues get <issueId> --project <projectId> --json
Statuses: open, resolved, false_positive. ta issues get lists the runs
that surfaced the issue; inspect them with ta runs messages or
ta runs recording. Turn flows worth keeping into saved tests.
Mobile App Coverage
Upload an iOS Simulator app or Android APK before cloud runs:
ta upload-app --app-path ios/build/Build/Products/Release-iphonesimulator/MyApp.app --project <projectId> --json
ta upload-app --app-path MyApp.apk --project <projectId> --remove-after 3600 --json
ta tests run <testId> --platform ios --app-id <appId> --wait --json
ta tests run <testId> --platform android --app-id <appId> --wait --json
Supported uploads: .app, .app.zip, .zip for iOS Simulator apps and .apk
for Android. Use --remove-after <seconds> to auto-delete uploaded apps.
List or delete uploaded builds:
ta projects apps <projectId> --json
ta projects apps-delete <projectId> <appId> --json
Reporting
Report:
- Project ID/name
- Test IDs and titles created or updated
- Group IDs/names touched
- Remote validation command and result, if run
- Run ID or artifact/output path, if available
Do not claim durable coverage unless ta tests create or ta tests update ran.
References
| File | Description |
|---|---|
| reporting-template.md | Dashboard coverage report template |
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/tester-army/cli/testerarmy-cli">View testerarmy-cli on skillZs</a>