Solution · CI/CD mobile testing
Real iPhones in your release pipeline
HelmSwarm runs your test scenarios on real iPhones, and runs can be started and read over the same HTTP API the panel uses, signed in with a workspace account's session token (dedicated API keys and CI plugins aren't available yet). A CI job can start a run on the devices you choose, wait for it, and publish the JUnit XML report — while scheduled runs cover the nightly suite. Because pricing is per device, not per minute, the pipeline can run as often as it needs to.
What works in a pipeline today
Sign in and start a run
The job signs in with a dedicated HelmSwarm account and calls
POST /api/scenarios/{id}/runwith the device IDs to use.Wait for the result
It polls
GET /api/runs/{run_id}until the run finishes. Per-device results and screenshots are kept in HelmSwarm.Publish the report
It downloads
GET /api/runs/{run_id}/junit.xmland hands it to your CI tool's test report step.
A pipeline step, in outline
Pseudocode — adapt it to your CI system and keep the credentials in its secret store.
# 1. Sign in (dedicated CI account) and keep the token
TOKEN=$(curl -s -X POST https://app.helmswarm.com/api/auth/login \
-H 'Content-Type: application/json' \
-d "{\"username\":\"$HS_USER\",\"password\":\"$HS_PASSWORD\"}" | jq -r .token)
# 2. Start scenario 42 on devices 7 and 9
RUN=$(curl -s -X POST https://app.helmswarm.com/api/scenarios/42/run \
-H "Authorization: Bearer $TOKEN" -H 'Content-Type: application/json' \
-d '{"device_ids":[7,9]}' | jq -r .id)
# 3. Poll GET /api/runs/$RUN until the status is final, then:
curl -s -H "Authorization: Bearer $TOKEN" \
https://app.helmswarm.com/api/runs/$RUN/junit.xml -o helmswarm-junit.xmlWhy real iPhones belong in the pipeline
Simulators are fast and great for unit and UI tests, but they don't reproduce real performance, memory pressure, camera and sensor behaviour, push and SMS flows, or the exact rendering of a physical screen. A short real-device suite before release catches the problems simulators can't.
When real-device testing is metered by the minute, every pipeline run becomes a cost decision and the runs that catch regressions are the first to be cut. With a flat price per iPhone, the gate before merge can run on every build. Read more in cost-conscious QA.
What to plan for
- One device, one run at a time. Parallelism comes from having more iPhones; size your device count to your pipeline's peak.
- Authentication. The API uses session tokens from a normal sign-in. Use a dedicated account with only the permissions the pipeline needs.
- Builds. Installing a new build happens from the panel today. We'll help you design the flow during onboarding.
CI/CD questions
Can I trigger HelmSwarm runs from GitHub Actions, GitLab CI or Jenkins?
Yes, from any CI system that can make HTTPS requests. A pipeline step signs in to the HelmSwarm API, starts a scenario run on chosen devices, polls the run until it finishes and downloads the JUnit XML report.
Is there a HelmSwarm CI plugin or API key?
Not yet. Today the API uses the same token-based sign-in as the panel, so use a dedicated account for your pipeline. Customer API keys and ready-made CI integrations are not available yet.
How do I get test results into my CI dashboard?
Every run can be exported as JUnit XML, which GitHub Actions, GitLab, Jenkins and most other CI tools can display. The run page in HelmSwarm keeps per-device results and screenshots.
Can I install the new build from the pipeline?
Builds are installed from the panel today, on one device or up to 20 at once. Tell us about your pipeline and we'll help you set up the flow that fits.
Take the meter out of CI
Tell us how many devices you need and how you test. We'll confirm compatibility, pricing and the next steps by email.