Reports and artifacts

Every run reports to the terminal and writes an HTML report. JSON and JUnit reporters are there for CI.

intelliwright.config.ts

export default defineConfig({
  reporters: ["terminal", "html", ["junit", { outputFile: "reports/junit.xml" }]],
  report: { open: "never" },
});

reporters lists reporter names, or [name, options] pairs. --reporter terminal,json replaces the list for one run.

Terminal

While tests run, each one gets a line as it finishes: ✓ passed, ✘ failed or timed out, - skipped (with the reason), and a note when a test is retried or passes on a retry. At the end:

• each failure, with its error, the lines of code around the failing line, its triage label and its artifact folder;
• the counts of passed, failed, flaky and skipped tests, with the failed and flaky ones listed;
• the counts for each tag;
• the three slowest tests;
• a Jev line with the provider, the model versions that answered, the number of calls, the input tokens and the cost;
• notes, such as triage being skipped;
• the total time.

Colors are used only when the output is a terminal. NO_COLOR turns them off, and FORCE_COLOR turns them on.

HTML report

The HTML reporter writes intelliwright-report/index.html: one self-contained file, with screenshots, traces and other files in data/ next to it.

npx intelliwright show-report

serves the last report on http://localhost:9323 and opens it. It takes a folder, --port, --host and --no-open; see the command line reference.

The report has:

• totals, the run's duration, and Jev usage for the run;
• filters by status, tag and file, and a search box;
• for each test: its tags, triage label, each attempt with its annotations, such as why it was skipped, errors with their locations, a timeline of steps, and an AI decisions table with every question, answer and probability;
• for failed attempts: the screenshot, the ARIA snapshot, the console and network logs, and the trace;
• attachments, such as the code from ai.run and anything a test attached with testInfo.attach().

The selected filters and test are kept in the page's URL, so a link opens the same view.

UI mode shows the same view while tests run, and fills it in as results arrive.

Option	Default	What it does
outputFolder	intelliwright-report	Where the report goes, relative to the config's folder. It's replaced on every run.
open	report.open	When to open the report: on-failure, always or never.

report.open in the config defaults to on-failure: after a local run with failures, the report opens in your browser. It never opens when the CI environment variable is set.

JSON

The JSON reporter writes the whole run, every attempt of every test included, to test-results/results.json. Set another path with outputFile. The file has the run's status, duration, counts and Jev usage, and for each test its file, line, title path, tags, outcome, triage, and attempts with their errors, steps, attachments and AI decisions.

JUnit

The JUnit reporter writes test-results/junit.xml for CI systems that show test results. Set another path with outputFile.

• Each test file is a testsuite, and each test a testcase named by its title path.
• Flaky tests count as passed, the same as in the terminal summary.
• A failure's system-out holds its triage label and the paths of its artifacts.

Artifacts

When an attempt fails or times out, Intelliwright saves the evidence to its own folder under test-results/, named after the file and test, with -retry1 and so on for retries:

File	What it holds
trace.zip	A Playwright trace: every action, with DOM snapshots, screenshots and sources.
screenshot.png	A full-page screenshot of each open page, numbered when there are several.
aria-snapshot.yml	The ARIA snapshot of each open page, with its URL.
console.log	Console messages and uncaught page errors.
network.json	Every request, with its method, URL, status, failure and timing.

When an attempt passes, nothing is saved. Open a trace with Playwright's trace viewer:

npx playwright-core show-trace test-results/e2e-notes-e2e-saves-a-note-a1b2c3/trace.zip

or drop the file on trace.playwright.dev.

Artifacts are captured after afterEach hooks and fixture teardown, so they show the page as cleanup left it. For ai.run failures, the report also has the page state Jev last answered from.

test-results/ is cleared at the start of each run, except for .last-run.json, which --last-failed reads. UI mode doesn't clear it between runs. Set another folder with outputDir.