HTTP Archive (HAR) command-line tool
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


HTTP Archive (HAR) command-line tool


The curl tool has many options, but it can be difficult to use. HAR is a great format for debugging and logging, but it's very static, and just sits there. This is where harcurl comes in. To make a request, simply fill out a HAR entry with the request fields you are sending, and harcurl will:

  • read the JSON object from stdin
  • convert it to libcurl options
  • call curl_easy_setopt so that libcurl can use them
  • call curl_easy_perform
  • call curl_easy_getinfo to fill in the timings
  • convert the response back to JSON
  • dumps the JSON object to stdout

It is a very simply idea, but hopefully one that helps people to understand, inspect, and use: CURL, HAR, and HTTP.


  • libcurl for obvious reasons (all HTTP logic is handled by it).
  • jansson for JSON parsing, reading and writing.
  • glib for byte-array utils, utf8 validation, etc.
  • zlib for GZIP compression utils.


$ cat > req.json <<EOF
    "request": {
        "method": "GET",
        "url": ""

$ harcurl < req.json > resp.json

HAR Extensions

HAR-1.2 is a great specification. It does miss a couple things, however, so harcurl uses a few extensions to it where appropriate.

  • entry.request._headersText
  • entry.request._requestLine entry.request._requestLine should be the same as {method} {_urlParts.path} {httpVersion}
  • entry.request._urlParts entry.request.url should be the same as {_urlParts.scheme}://{_urlParts.authority}{_urlParts.path}
  • entry.response._headersText
  • entry.response._contentType
  • entry.response._statusLine entry.response._statusLine should be the same as {httpVersion} {status} {statusText}