Skip to main content

​
Overview

OneTest accepts automated test results from any CI/CD pipeline. Results from your automated tests appear alongside manual and AI-driven test runs in a unified dashboard — giving you a single view of quality across all execution methods. There are three ways to report automated results:

ReportPortal Agents

Drop-in compatible with pytest-reportportal, Java agent, and other RP agents

JUnit XML Import

Upload JUnit XML files from any test framework

MCP API

Use the OneTest MCP tools directly from Claude Code or any MCP client

​
Prerequisites

Before reporting results, you need an API key:
  1. Go to tms.onetest.ai -> Settings -> Integrations and API Keys
  2. Click + Add API Key and copy the key immediately (it is shown only once)
  3. Note your Product UUID displayed on the same page

​
Method 1: ReportPortal Agents

OneTest is fully compatible with the ReportPortal v2 API. If your team already uses a ReportPortal agent, just point it at OneTest — no code changes required.

​
Configuration

Set these environment variables in your CI/CD pipeline: 
RP_ENDPOINT=https://tms.onetest.ai/api/receiver
RP_PROJECT=<your-product-uuid>
RP_API_KEY=<your-api-key>

​
Supported Agents

Any ReportPortal agent works out of the box:
FrameworkAgent
pytestpytest-reportportal
JUnit 5agent-java-junit5
TestNGagent-java-testng
Cypressagent-js-cypress
Playwrightagent-js-playwright
Robot Frameworkagent-Python-RobotFramework

​
Example: pytest

# pytest.ini
[pytest]
rp_endpoint = https://tms.onetest.ai/api/receiver
rp_project = a1b2c3d4-e5f6-7890-abcd-ef1234567890
rp_api_key = ak_your_api_key_here
rp_launch = Nightly Regression
rp_launch_attributes = env:staging build:2.1.0

pytest --reportportal

​
Example: GitHub Actions

- name: Run tests with reporting
  env:
    RP_ENDPOINT: https://tms.onetest.ai/api/receiver
    RP_PROJECT: ${{ secrets.ONETEST_PRODUCT_UUID }}
    RP_API_KEY: ${{ secrets.ONETEST_API_KEY }}
  run: pytest --reportportal

​
What Gets Reported

The ReportPortal agent sends a structured hierarchy: 
Launch (test run)
  └── Suite
        └── Test
              └── Step
                    └── Log (with screenshots, console output)
Each test item includes:
  • code_ref — the test’s fully qualified name (e.g., test_login.py:TestLogin.test_valid_login)
  • Status — passed, failed, skipped
  • Duration — execution time
  • Logs — console output, stack traces, screenshots
The code_ref value is what OneTest uses to link automated results to test cases. Set the Automation Test ID field on your test cases to match the code_ref format your agent sends. See Automation Coverage for details.

​
Method 2: JUnit XML Import

For frameworks that don’t support ReportPortal agents, upload JUnit XML results directly: 
curl -X POST "https://tms.onetest.ai/api/v1/products/{product_uuid}/import/junit" \
  -H "Authorization: Bearer {api_key}" \
  -F "file=@test-results.xml"

​
Example: GitHub Actions

- name: Run tests
  run: pytest --junitxml=results.xml

- name: Upload results to OneTest
  if: always()
  run: |
    curl -X POST "https://tms.onetest.ai/api/v1/products/${{ secrets.ONETEST_PRODUCT_UUID }}/import/junit" \
      -H "Authorization: Bearer ${{ secrets.ONETEST_API_KEY }}" \
      -F "file=@results.xml"
JUnit XML import creates a single launch with all test cases as flat items (no suite hierarchy). For hierarchical reporting, use a ReportPortal agent instead.

​
Method 3: MCP API

If you use the OneTest QA Agent or any MCP-compatible client, you can record results directly through MCP tools.

​
Setup

Add the MCP server to your .mcp.json: 
{
  "mcpServers": {
    "test-management": {
      "type": "http",
      "url": "https://tms.onetest.ai/api/test-management/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_API_KEY>"
      }
    }
  }
}

​
Recording Results

Use /qa-onetest run in Claude Code to execute and record a full test run, or call the MCP tools directly: 
# 1. Create a run with test cases
run = create_run(
    name="Sprint 5 Regression",
    identifiers=["TC-0001", "TC-0005", "TC-0012"]
)

# 2. Start the run
start_run(run_id=run["id"])

# 3. Get executions and record results
items = get_run_items(run_id=run["id"])
for item in items["items"]:
    record_test_result(
        execution_id=item["id"],
        status="passed",
        step_results=[
            {"step_number": 1, "status": "passed", "actual_result": "Page loaded"}
        ]
    )

# 4. Complete the run
complete_run(run_id=run["id"])
See Running Tests for the full /qa-onetest workflow.

​
API Endpoints Reference

EndpointMethodDescription
/api/v2/{product_uuid}/launchPOSTStart a new test launch
/api/v2/{product_uuid}/itemPOSTReport a test item
/api/v2/{product_uuid}/logPOSTAdd a log entry (text, screenshot, etc.)
/api/v1/products/{product_uuid}/import/junitPOSTImport JUnit XML results

​
Viewing Automated Results

Once results are reported, they appear in the Test Runs section alongside manual and AI-driven runs. You can filter by source:
  • All runs — see everything together
  • Automated — only CI/CD results
  • Manual — only manual/AI-driven runs
Each automated run shows:
  • Pass/fail/skip breakdown with percentages
  • Hierarchical test items — drill into suites, tests, and steps
  • Logs and screenshots — console output, stack traces, failure screenshots
  • Duration and timing — per-test and total execution time

​
Linking to Test Cases

When a code_ref from an automated result matches a test case’s automation_test_id, OneTest links them automatically. This enables:
  • Tracking automated vs. manual coverage on the Automation Coverage dashboard
  • Seeing automated results directly on the test case detail page
  • Identifying automation gaps (tests marked automated but not running in CI/CD)

​
Cost

Each API call costs 1 coin from your weekly budget. Browser UI usage is always free. See Usage & Billing for details.

​
What’s Next?

Integrations and API Keys

Manage API keys and view your Product UUID

Automation Coverage

Track which automated tests are actually running

Viewing Results

Analyze results and trends in the dashboard