Run reports

A finished collection run can be saved in two formats:

Format	From the app	From zorvik run	Use it for
JSON	Runner tab → Export → JSON report…	--json (printed to standard output)	Scripts, dashboards, your own checks with jq
JUnit XML	Runner tab → Export → JUnit XML…	--junit <FILE>	CI test reports (GitHub, GitLab, Jenkins, Azure DevOps and others)

Both come from the same data, so a run exported from the app and the same run from the command line have the same shape. In the app, Export appears once the run has finished; the last 5 finished runs can be exported (older ones say That run is no longer available; run it again). The suggested file name is the run’s name and time, such as Users 2026-09-28 14-05.json.

Text output

Without --json, zorvik run prints one line per request as it finishes, then a summary:

Iteration 1 of 2
✓ POST    Log in  https://api.example.com/login  200 84 ms, 312 B
    ✓ status is 200
✓ GET     Profile  https://api.example.com/me  200 41 ms, 1204 B
    ✓ token works
    ✓ not bob
– WS      Live prices  (skipped: WebSocket connections are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests)

Iteration 2 of 2
✓ POST    Log in  https://api.example.com/login  200 77 ms, 312 B
    ✓ status is 200
✗ GET     Profile  https://api.example.com/me  200 39 ms, 1204 B
    ✓ token works
    ✗ not bob — expected 'bob' to not equal 'bob'
– WS      Live prices  (skipped: WebSocket connections are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests)

Requests  3 passed, 1 failed, 2 skipped (6 total)
Tests     5 passed, 1 failed
Time      0.6 s, 2 iterations

• ✓ passed, ✗ failed, – skipped. Each line has the method, name, the URL as sent, then the status, time and body size, or the error when nothing came back.
• Tests are listed under their request. Script errors start with !, and so does the list of undefined variables (! undefined variables: token).
• Iteration N of M headers appear when the run has more than one iteration.
• The summary ends with Stopped (Ctrl+C). or Stopped at the first failure (--bail). when that happened, and the run’s error if it had one.
• Console output from scripts is not printed; it’s in the JSON report.

Colors are used when standard output is a terminal and the NO_COLOR environment variable is not set. Control characters in names, URLs and messages are shown escaped (for example \u{1b}), so a response can’t take over your terminal.

JSON report

The JSON report is an object with the run’s summary and every request’s results:

{
  "summary": {
    "name": "Users",
    "environment": "Staging",
    "startedAt": 1790571915806.0,
    "durationMs": 1084.75,
    "iterations": 2,
    "requests": 4,
    "failed": 1,
    "skipped": 0,
    "testsPassed": 7,
    "testsFailed": 1,
    "testsSkipped": 0,
    "stopped": false,
    "bailed": false,
    "error": null,
    "passed": false,
    "perIteration": [
      { "iteration": 0, "requests": 2, "failed": 0, "skipped": 0, "testsPassed": 4, "testsFailed": 0, "durationMs": 540.2 },
      { "iteration": 1, "requests": 2, "failed": 1, "skipped": 0, "testsPassed": 3, "testsFailed": 1, "durationMs": 544.5 }
    ],
    "omitted": 0
  },
  "results": [
    {
      "iteration": 1,
      "path": "Users/Get profile.yaml",
      "name": "Get profile",
      "kind": "http",
      "method": "GET",
      "url": "https://api.example.com/me?key={{apiKey}}",
      "status": 200,
      "durationMs": 39.1,
      "size": 1204,
      "tests": [
        { "name": "token works", "passed": true, "skipped": false, "error": null },
        { "name": "not bob", "passed": false, "skipped": false, "error": "expected 'bob' to not equal 'bob'" }
      ],
      "console": [{ "level": "log", "message": "user bob" }],
      "error": null,
      "scriptErrors": [],
      "unresolved": [],
      "passed": false,
      "skipped": false
    }
  ]
}

summary

Field	Type	Description
name	string	The folder’s name, or the workspace’s for the whole collection
environment	string or null	The environment’s name
startedAt	number	Start time, Unix epoch milliseconds
durationMs	number	Duration of the whole run
iterations	number	Iterations planned
requests	number	Requests sent or tried (skipped ones not counted)
failed	number	Requests that failed
skipped	number	Requests skipped (kinds the runner doesn’t send)
testsPassed, testsFailed, testsSkipped	number	Test counts over the whole run
stopped	boolean	Stopped by the user (Stop, or Ctrl+C) before it finished
bailed	boolean	Stopped at the first failure (--bail / Stop on the first failure)
error	string or null	Why the run ended early otherwise, for example a setNextRequest loop
passed	boolean	No request failed and error is null
perIteration	array	For each iteration that ran: iteration (from 0), requests, failed, skipped, testsPassed, testsFailed, durationMs
omitted	number	Results left out of results (see Limits); they are still counted above

results

One entry per request per iteration, in the order they ran.

Field	Type	Description
iteration	number	Iteration, from 0
path	string	The request file under requests/
name	string	The request’s name
kind	string	http, sse, websocket, tcp, udp, dns, mqtt or grpc
method	string	The method sent
url	string	The URL as sent, with secret values shown as {{name}}; the saved URL when nothing was sent
status	number or null	HTTP status, null when there was no response
durationMs	number or null	Time of the request (for “repeat until”, from the first send to the end of the last)
size	number or null	Response body size in bytes
tests	array	{ name, passed, skipped, error } for each test, including “Matches the API spec”
console	array	{ level, message } lines from the scripts; level is log, info, warn, error or debug
error	string or null	Why there is no usable result: network or resolve error, pre-request script error, “repeat until” timeout, an unreadable request file
scriptErrors	string[]	Post-response scripts that failed, for example Post-response script of request 'Get profile' failed at line 3: …
unresolved	string[]	Variables the request used but that weren’t defined
passed	boolean	Whether the request passed
skipped	boolean	Not sent because of its kind
skipReason	string	Only for skipped requests: why
attempts	number	Only for requests with “repeat until”: how many times it was sent

A few jq examples:

zorvik run ./api --json > run.json

jq '.summary.passed' run.json                                   # true or false
jq -r '.results[] | select(.passed | not) | .name' run.json     # names of failed requests
jq '[.results[].durationMs | select(. != null)] | add / length' run.json   # average request time

JUnit XML report

The JUnit report follows the layout most CI tools read:

• One <testsuites> for the run, named after it, with the totals, the duration in seconds and the start time (timestamp, UTC).
• One <testsuite> per request, named after its path without .yaml (for example Users/Get profile), in the order the requests first ran.
• One <testcase> per test (per iteration), with classname set to the suite’s name.
• One extra <testcase> for the request itself, named METHOD Name (for example GET Get profile), when the request was skipped, had an error, or had no tests (then it passes, or fails with HTTP 404).
• With more than one iteration, every test case name ends with (iteration N) (N from 1).
• The request’s time is on its first test case; the others have 0.000, so sums stay right.

<?xml version="1.0" encoding="UTF-8"?>
<testsuites name="Users" tests="7" failures="2" errors="1" skipped="2" time="1.500" timestamp="2026-09-28T14:05:12">
  <testsuite name="Users/Get profile" tests="4" failures="1" errors="0" skipped="1" time="0.020">
    <testcase name="status is 200 (iteration 1)" classname="Users/Get profile" time="0.012"/>
    <testcase name="body (iteration 1)" classname="Users/Get profile" time="0.000">
      <failure message="expected &apos;a&apos; to equal &apos;b&apos;" type="AssertionError">expected &apos;a&apos; to equal &apos;b&apos;</failure>
    </testcase>
    <testcase name="later (iteration 1)" classname="Users/Get profile" time="0.000">
      <skipped/>
    </testcase>
    <testcase name="GET Get profile (iteration 2)" classname="Users/Get profile" time="0.008"/>
  </testsuite>
  <testsuite name="Health" tests="1" failures="0" errors="1" skipped="0" time="0.000">
    <testcase name="GET Health (iteration 1)" classname="Health" time="0.000">
      <error message="Could not connect to 127.0.0.1:8080: connection refused (is the server running?)" type="Error">Could not connect to 127.0.0.1:8080: connection refused (is the server running?)</error>
    </testcase>
  </testsuite>
  <testsuite name="Users/Missing" tests="1" failures="1" errors="0" skipped="0" time="0.000">
    <testcase name="GET Missing (iteration 2)" classname="Users/Missing" time="0.000">
      <failure message="HTTP 404" type="AssertionError">HTTP 404</failure>
    </testcase>
  </testsuite>
  <testsuite name="Live prices" tests="1" failures="0" errors="0" skipped="1" time="0.000">
    <testcase name="Live prices (iteration 2)" classname="Live prices" time="0.000">
      <skipped message="WebSocket connections are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests"/>
    </testcase>
  </testsuite>
</testsuites>

How outcomes map:

Outcome	JUnit element
A test passed	<testcase …/>
A test failed	<failure type="AssertionError"> with the test’s error (failed when it has none)
A test was skipped (pm.test.skip, or pm.test without a function)	<skipped/>
The request was skipped	A request test case with <skipped message="reason"/>
The request couldn’t be sent, a script failed, or “repeat until” timed out	A request test case with <error type="Error">; all messages, one per line
No tests, and the status failed the request	A request test case with <failure> HTTP <status>
No tests, and it passed	A passing request test case

Special characters are escaped for XML; control characters that XML can’t hold are written as \u{…}.

Skip reasons

A request is skipped when it’s of a kind the runner doesn’t send. It counts as neither passed nor failed. The reason is shown in the text output, in skipReason in the JSON report, and in the JUnit <skipped message>:

Request kind	Skip reason
WebSocket	WebSocket connections are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests
TCP	TCP connections are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests
UDP	UDP sockets are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests
MQTT	MQTT clients are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests
Socket.IO	Socket.IO clients are live sessions; the runner sends HTTP, GraphQL, SSE and MCP requests
gRPC	gRPC calls don't run in the collection runner; the runner sends HTTP, GraphQL, SSE and MCP requests
DNS	DNS queries don't run in the collection runner; the runner sends HTTP, GraphQL, SSE and MCP requests

A request file that can’t be read is not skipped: it fails with The request can't be read: ….

A test is skipped when it’s written with pm.test.skip(name, fn) or pm.test(name) without a function. Skipped tests are counted in testsSkipped and don’t affect whether the request passes.

Exit codes of zorvik run

Code	Meaning
0	Every request passed (skipped requests don’t count). Also when the run had no requests at all.
1	A request or test failed, --bail stopped the run, or the run ended with an error (such as a setNextRequest loop)
2	The run was stopped with Ctrl+C before it finished, or it couldn’t start: a wrong option or value, the workspace, folder or environment not found, a data file problem; also when the JUnit file couldn’t be written

Caution

A folder with no requests in it runs nothing and exits with 0. If that matters in your pipeline, check summary.requests in the JSON report.

See zorvik run for the options, and its CI examples.

Secrets in reports

Reported URLs never show the values of secret variables (variables marked secret in the environment or the workspace):

• Every secret value of three characters or more is replaced by {{name}} in the URL, both as written and URL-encoded.
• This covers values given with --var for a secret variable, and values scripts set for it during the run (for example a token saved by a login request).

Only URLs are masked. Test names, test errors and console output are reported as the scripts wrote them, so keep secret values out of them. Reports don’t contain request or response headers or bodies.

Limits

Limit	Value
Results in results	The first 50,000, then failed ones only, up to 100,000 in total. The rest are counted in summary.omitted and in all the other counts.
Console lines per result	200 (then N more console lines not kept.)
Characters per console line	4,096 (then … (truncated))