A MIME type (now formally a media type) is the label in a response's `Content-Type` header that tells the client what the body contains — `text/html`, `application/json`, `image/png`. The browser uses it to decide whether to render, play, or download the response.

Why does my HTML file download instead of opening?

Because the server sent the wrong `Content-Type` — usually `application/octet-stream` or `binary/octet-stream`, which is what S3 assigns when an upload does not declare one — or because it sent `Content-Disposition: attachment`, which forces a download whatever the type says.

What is the difference between Content-Type and Content-Disposition?

`Content-Type` says what the bytes are. `Content-Disposition` says what to do with them: `inline` renders in place, `attachment` downloads and can suggest a filename. An `attachment` disposition overrides a perfectly correct `text/html`.

Does the file extension decide the MIME type?

Not over HTTP. The server's header is authoritative and the extension is irrelevant — a file named `report.html` served as `application/octet-stream` downloads. Only when opening from local disk does the browser fall back to the extension.

MIME type

A MIME type — formally a media type — is the label in a response's Content-Type header telling the client what the body contains. The browser reads it to decide whether to render the response, play it, or save it to disk.

Content-Type: text/html; charset=utf-8    → renders
Content-Type: application/octet-stream    → downloads
Content-Type: text/plain                  → shows the markup as text

Over HTTP the header is authoritative and the file extension is irrelevant. report.html served as application/octet-stream is a download, every time.

The two headers that decide

Header	Says	Values that matter
`Content-Type`	What the bytes are	`text/html`, `application/octet-stream`, `text/plain`
`Content-Disposition`	What to do with them	`inline`, `attachment; filename="…"`

Content-Disposition: attachment wins over a correct Content-Type. If a page downloads despite text/html, that header is why.

Where reports get this wrong

S3 and compatible object stores. An upload with no declared type is stored as application/octet-stream or binary/octet-stream, so the report downloads. Fixed at upload time with --content-type text/html, not afterwards on the link (the full fix).
CI artifact endpoints. Many serve every artifact as a download by policy, regardless of type — which is why a GitHub Actions "report link" is a zip (why).
Chat and ticket attachments. HTML is served as a download deliberately, because rendering user-supplied HTML in the app's own origin would be an XSS vector (why Slack won't preview).

That last one is not a misconfiguration. Serving arbitrary HTML inline from an origin that holds sessions is genuinely dangerous — which is why services that do render report HTML do it inside a sandboxed iframe on a separate origin.

Charset matters too

text/html without charset=utf-8 leaves the browser guessing, and the guess is wrong for any report containing non-ASCII test names, author names or currency symbols. The <meta charset> in the document helps, but the header wins where both are present.

Try it

Comma is free — unlimited reports, unlimited commenters, unlimited revision history. Reports are served with the headers that make them render.

Publish a report that opens →

MIME type

The two headers that decide

Where reports get this wrong

Charset matters too

Try it

Related