cURL to Code

Paste a curl command copied from the DevTools Network panel, an API doc or a blog post and get runnable code for five targets: Python (requests), JavaScript (fetch), Node.js (axios), PHP (cURL) and Go (net/http).

The mapping follows what curl actually does, not what looks familiar. A few outputs look unusual because of that, deliberately:

• No Content-Type is invented. With -d, curl sends application/x-www-form-urlencoded by default. Many converters "helpfully" switch to JSON, which then does not match what the server expects. This follows curl and adds a note below the output: if the body is JSON, add -H 'Content-Type: application/json'.
• Multiple -d values are joined with & — curl's own rule, not a bug.
• -u user:pass becomes a ready-made Authorization: Basic ... header, with the base64 computed in the browser so each language does not need its own version.
• -k produces the "skip certificate verification" form (verify=False in Python, VERIFYPEER => false in PHP) with a warning underneath. It is not a recommendation — it translates the intent in your command.
• --data-urlencode is passed through as written, not encoded here: encoding belongs in your code, and pre-encoding hides the original value.

Multipart (-F) is generated in full only for Python and PHP. JavaScript gets a FormData skeleton; Node and Go are left for you, because a mime/multipart boilerplate runs twenty lines and would make the output harder to use than to write.

Unmapped flags are never silently dropped: --http2, -s, -v and similar are listed below the output so you know what was not translated.

What you get is a minimal skeleton that runs — no retries, no refined error handling, and timeout=30 in the Python version. Fill those in to taste.

Parsing and generation happen in the browser; tokens in the command are never uploaded.