What is an HTML report?

An HTML report is a web page generated by a tool to present the results of a run — a test suite, a coverage measurement, a security scan, a profile, a data analysis — as browsable, usually interactive output rather than console text. It is produced as a build output, not authored by hand.

Is an HTML report a single file?

Usually not. Most generators emit a directory: an entry `index.html` plus CSS, JavaScript, fonts, images and often a JSON data tree the page fetches at runtime. Copying the entry file out of that directory produces a page that renders blank or unstyled.

How is an HTML report different from a PDF?

A PDF is a fixed rendering; an HTML report keeps its behaviour. Sorting, filtering, expanding a failed assertion, drilling into a covered line, hovering a chart series — all of that is the report, and printing it to PDF discards it.

Why do HTML reports open fine locally and break when sent?

Three conditions hold locally and vanish in transit: the whole output directory is present, the page has a permissive origin, and the network reaches every CDN it references. Emailing the entry file preserves none of them.

HTML report

An HTML report is a web page generated by a tool to present the results of a run — tests, coverage, a security scan, a profile, an analysis — as browsable output rather than console text. It is a build output, not a hand-authored document, and it is usually interactive.

The category is broad on purpose. A Playwright run, a JaCoCo coverage tree, a Trivy scan, a Lighthouse audit, an exported Jupyter notebook and a dbt docs site are all HTML reports: each one is machine-generated, each one is meant to be read by a person, and each one loses something essential the moment it is flattened into text or an image.

What is actually inside one

Almost never a single file. A typical generator emits a directory:

Part	Example	What happens without it
Entry document	`index.html`	Nothing to open
Stylesheets	`assets/app.css`	Unstyled wall of text
Scripts	`assets/app.js`	Dead tables, no interactivity
Data tree	`data/test-cases/*.json`	Empty shell, blank page
Fonts, icons, PNGs	`static/`	Broken glyphs, missing charts
CDN references	`<script src="https://cdn…">`	Charts silently absent

The data tree is the one that surprises people. Allure, Playwright and several coverage tools render a shell and then fetch() the results — so the entry file alone is not a degraded report, it is an empty one (why Allure opens blank).

The three conditions that only hold locally

The whole directory is present. Relative references resolve because the siblings exist.
The origin is permissive. A local server gave the page a real http:// origin; opening it from disk gives an opaque file:// origin where fetch() is blocked (same-origin policy).
The network reaches the CDN. Yours does. A reviewer behind a corporate proxy, or a viewer with a strict Content Security Policy, may not.

Every common "it works on my machine" report failure is one of those three, and the browser console names which one before any guessing starts.

Human report, machine report

Most tools emit both, and both are worth keeping:

Machine-readable — JUnit XML for tests, SARIF for findings, lcov or Cobertura for coverage. These are for CI gates, code-host annotations and dashboards.
Human-readable — the HTML report. This is the one someone opens when the gate fails and they need to know why.

Treating the HTML one as a build artifact is the common mistake: artifacts are addressed by links that expire under a retention policy, which is correct for a cache and wrong for a document someone will cite in six months.

Making one readable by other people

Publish the output directory rather than passing the entry file around. Assets upload alongside the entry HTML with their relative references rewritten, the page is served over HTTPS from a real origin, and scripts keep running inside a sandboxed iframe (allow-scripts, no allow-same-origin) so sorting, filtering and charts behave as they do locally (interactive HTML reports).

Limits: entry HTML 5 MB; assets 25 MB per file, 250 MB and 500 files per report.

Try it

Comma is free — unlimited reports, unlimited commenters, unlimited revision history.

Publish an HTML report →

Share an HTML report · Share an HTML folder
Self-contained HTML · Build artifact
Report types — 200+ per-tool guides · Glossary