Probely Developers

Probely is a Web Vulnerability Scanning suite for Agile Teams. It provides continuous scanning of your Web Applications and lets you efficiently manage the lifecycle of the vulnerabilities found, in a sleek and intuitive ~~web interface~~ API.

Quick-Start

Authentication

To use the API, you first need to create a token (API Key). To create a token, select a target from the drop-down list, go to the "Settings" page, and click on the "Integrations" tab.

Write a name for the API Key. For example, if you want to use the API Key for travis, you could name it "travis". In this example, we chose "example.com_key"

The API key was created successfully:

On every request, you need to pass this token in the authorization header, like this:

Authorization: JWT eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJqdGkiOiJBRlNJQlp
3elFsMDEiLCJ1c2VybmFtZSI6IkNIZ2tkSUROdzV0NSJ9.90UwiPGS2hlvgOLktFU0LfKuatNKm
mEP79u17VnqT9M

WARNING: Treat this token as a password. With this token, you have the power to fully manage the target.

In the following examples, the token will be named as PROBELY_AUTH_TOKEN.

Scan target

First let's view our target list:

curl https://api.probely.com/targets/ \
  -X GET \
  -H "Content-Type: application/json" \
  -H "Authorization: JWT PROBELY_AUTH_TOKEN"

From the results, we need the target id:

{
   "count":1,
   "page_total":1,
   "page":1,
   "length":10,
   "results":[
      {
         "id":"AxtkqTE0v3E-",
         "name":"test-site",
         "desc":"",
         "url":"https://test-site.example.com",
         "settings":
            "(...)"
         ,
         "stack":
            "(...)"
         ,
         "verified":true,
         "(...)": "(...)"
      }
   ]
}

Now we can send a request to start a scan on target id AxtkqTE0v3E-

curl https://api.probely.com/targets/AxtkqTE0v3E-/scan_now/ \
  -X POST \
  -H "Content-Type: application/json" \
  -H "Authorization: JWT PROBELY_AUTH_TOKEN"

And we get a response saying that the scan is scheduled: the status is queued, and we've got a scan id:

{
   "changed":"2017-08-01T13:37:00.843339Z",
   "started":null,
   "completed":null,
   "mediums":0,
   "changed_by":
    "(...)"
   ,
   "highs":0,
   "status":"queued",
   "id":"S6dOMPn0SnoH",
   "created_by":
    "(...)"
   ,
   "target":
    "(...)"
   ,
   "created":"2017-08-01T13:37:00.843339Z",
   "lows":0
}

Using the scan id S6dOMPn0SnoH, we can pool the scan status:

curl https://api.probely.com/targets/AxtkqTE0v3E-/scans/S6dOMPn0SnoH/ \
  -X GET \
  -H "Content-Type: application/json" \
  -H "Authorization: JWT PROBELY_AUTH_TOKEN"

And we get a response saying that the scan status is now started:

{
   "id":"S6dOMPn0SnoH",
   "changed":"2017-08-01T13:38:12.623650Z",
   "started":null,
   "completed":null,
   "mediums":0,
   "changed_by":
    "(...)"
   ,
   "highs":0,
   "status":"started",
   "created_by":
    "(...)"
   ,
   "target":
    "(...)"
   ,
   "created":"2017-08-01T13:37:00.843339Z",
   "lows":0
}

The possible statuses are:

| Status | Name | Description | | ------ | ---- | ----------- | | queued | Queued | The scan is queued to start | | started | Started | The scan is currently running | | under_review | Under Review | The scan is complete but has some findings under review | | completed | Completed | The scan is complete | | completed_with_errors | Completed with errors | The scan is complete even after getting some error(s) | | failed | Failed | The scan failed | | canceled | Canceled | The scan was canceled | | canceling | Canceling | The scan is being canceled |

During the scan, the keys "lows", "mediums", and "highs" will be updated with the findings, as they are being found.

After we get either the status completed or completed_with_errors, we can view the findings.

Get vulnerabilities

Using the previous scan id S6dOMPn0SnoH, we can get the scan results:

curl https://api.probely.com/targets/AxtkqTE0v3E-/scans/S6dOMPn0SnoH/ \
  -X GET \
  -H "Content-Type: application/json" \
  -H "Authorization: JWT PROBELY_AUTH_TOKEN"

We get a response saying that the scan status is now completed, and that 45 vulnerabilities were found. 14 low, 11 medium and 20 high:

{
   "id":"S6dOMPn0SnoH",
   "target":
    "(...)"
   ,
   "status":"completed",
   "started":"2017-08-01T13:37:12.623650Z",
   "completed":"2017-08-01T14:17:48.559514Z",
   "lows":14,
   "mediums":11,
   "highs":20,
   "created":"2017-08-01T13:37:00.843339Z",
   "created_by":
    "(...)"
   ,
   "changed":"2017-08-01T14:17:48.559514Z",
   "changed_by":
    "(...)"
}

You can now view the results of this scan, or the target findings.

Let's start with the scan results:

curl https://api.probely.com/targets/AxtkqTE0v3E-/findings/?scan=S6dOMPn0SnoH&page=1 \
  -X GET \
  -H "Content-Type: application/json" \
  -H "Authorization: JWT PROBELY_AUTH_TOKEN"

{
   "count":45,
   "page_total":5,
   "page":1,
   "length":10,
   "results":[
      {
         "id":79,
         "target":
          "(...)"
         ,
         "scans":
          "(...)"
         ,
         "labels":
          "(...)"
         ,
         "fix":"To fix an SQL Injection in PHP, you should use Prepared Statements. Prepared Statements can be thought of as a kind of compiled template for the SQL that an application wants to run, that can be customized using variable parameters.\n\nPHP's PDO extension supports Prepared Statements, so that's probably your best option.\n\nIn the example below you can see the use of prepared statements. Variables ```$username``` and ```$hashedPassword``` come from user input.\n\n```\n$stmt = $dbg->prepare(\"SELECT id, name FROM users\n                       WHERE username=? AND password=?\");\n$stmt->bindParam(1, $username);\n$stmt->bindParam(2, $hashedPassword);\nif ($stmt->execute()) {\n\t$user = $stmt->fetch();\n\tif ($user) {\n\t\t$_SESSION['authID'] = $user['id'];\n\t\techo \"Hello \" . $user['name'];\n\t} else {\n\t\techo \"Invalid Login\";\n\t}\n}\n```  \n\nAs an added bonus, if you're executing the same query several times, then it'll be even faster than when you're not using prepared statements. This is because when using prepared statements, the query needs to be parsed (prepared) only once, but can be executed multiple times with the same or different parameters. \n",
         "requests":[
            {
               "request":"(...)",
               "response":"(...)"
            },
            {
               "request":"(...)",
               "response":"(...)"
            }
         ],
         "evidence":null,
         "extra":"",
         "definition":{
            "id":"xnV8PJVmSoLS",
            "name":"SQL Injection",
            "desc":"SQL Injections are the most common form of injections because SQL databases are very popular in dynamic web applications. This vulnerability allows an attacker to tamper existing SQL queries performed by the web application. Depending on the queries, the attacker might be able to access, modify or even destroy data from the database.\n\nSince databases are commonly used to store private data, such as authentication information, personal user data and site content, if an attacker gains access to it, the consequences are typically very severe, ranging from defacement of the web application to users data leakage or loss, or even full control of the web application or database server.",
         },
         "url":"http://test-site.example.com/login.php",
         "path":"login.php",
         "method":"post",
         "parameter":"username",
         "value":"",
         "params":{
            "username":[
               "probely'"
            ],
            "password":[
               "probely"
            ]
         },
         "reporter":
          "(...)"
         ,
         "assignee":null,
         "state":"notfixed",
         "severity":30,
         "last_found":"2017-08-01T14:03:56.207794Z",
         "changed":"2017-08-01T14:03:56.207794Z",
         "changed_by":
          "(...)"
         ,
         "comment":""
      },
      "(...)"
   ]
}

You can also view all the target findings, which will show all the findings that are not yet fixed. \ The structure is similar to the previous result.

curl https://api.probely.com/targets/AxtkqTE0v3E-/findings/ \
  -X GET \
  -H "Content-Type: application/json" \
  -H "Authorization: JWT PROBELY_AUTH_TOKEN"

Get vulnerability details

You can also get details for a particular finding in a target. \ In this example we will get the details for the same finding as in the previous section:

curl https://api.probely.com/targets/AxtkqTE0v3E-/findings/79/ \
  -X GET \
  -H "Content-Type: application/json" \
  -H "Authorization: JWT PROBELY_AUTH_TOKEN"

This will result on the same information, but just for this particular finding:

{
   "id":79,
   "target":
    "(...)"
   ,
   "scans":
    "(...)"
   ,
   "labels":
    "(...)"
   ,
   "fix":"To fix an SQL Injection in PHP, you should use Prepared Statements. Prepared Statements can be thought of as a kind of compiled template for the SQL that an application wants to run, that can be customized using variable parameters.\n\nPHP's PDO extension supports Prepared Statements, so that's probably your best option.\n\nIn the example below you can see the use of prepared statements. Variables ```$username``` and ```$hashedPassword``` come from user input.\n\n```\n$stmt = $dbg->prepare(\"SELECT id, name FROM users\n                       WHERE username=? AND password=?\");\n$stmt->bindParam(1, $username);\n$stmt->bindParam(2, $hashedPassword);\nif ($stmt->execute()) {\n\t$user = $stmt->fetch();\n\tif ($user) {\n\t\t$_SESSION['authID'] = $user['id'];\n\t\techo \"Hello \" . $user['name'];\n\t} else {\n\t\techo \"Invalid Login\";\n\t}\n}\n```  \n\nAs an added bonus, if you're executing the same query several times, then it'll be even faster than when you're not using prepared statements. This is because when using prepared statements, the query needs to be parsed (prepared) only once, but can be executed multiple times with the same or different parameters. \n",
   "requests":[
      {
         "request":"(...)",
         "response":"(...)"
      },
      {
         "request":"(...)",
         "response":"(...)"
      }
   ],
   "evidence":null,
   "extra":"",
   "definition":{
      "id":"xnV8PJVmSoLS",
      "name":"SQL Injection",
      "desc":"SQL Injections are the most common form of injections because SQL databases are very popular in dynamic web applications. This vulnerability allows an attacker to tamper existing SQL queries performed by the web application. Depending on the queries, the attacker might be able to access, modify or even destroy data from the database.\n\nSince databases are commonly used to store private data, such as authentication information, personal user data and site content, if an attacker gains access to it, the consequences are typically very severe, ranging from defacement of the web application to users data leakage or loss, or even full control of the web application or database server.",
   },
   "url":"http://test-site.example.com/login.php",
   "path":"login.php",
   "method":"post",
   "parameter":"username",
   "value":"",
   "params":{
      "username":[
         "probely'"
      ],
      "password":[
         "probely"
      ]
   },
   "reporter":
    "(...)"
   ,
   "assignee":null,
   "state":"notfixed",
   "severity":30,
   "last_found":"2017-08-01T14:03:56.207794Z",
   "changed":"2017-08-01T14:03:56.207794Z",
   "changed_by":
    "(...)"
   ,
   "comment":""
}

Concepts

The short version is that you run scans on targets, and findings are created for any issue that is found. However, there are a few more concepts that must be explained in order to get a complete picture of how Probely works. We will spend the next few sections detailing the most important concepts.

Target

A target defines the scope of a scan, what will and won't be included in the scan plan. This is done by filling a target's site and assets.

The entry point for the web application (and authentication) is setup in the target's site.

In modern web applications, you are probably loading resources from multiple domains. A single page app, for example, will usualy load the page from one domain and make AJAX requests to another. This is what assets are for: they specify what domains our scanner should follow and create requests for.

Site

A URL is probably not the only thing you will need to setup when scannning your application. Does the application have an authenticated area? Does it use basic auth? Does it expect a certain cookie or header? These parameters are all configured in the target's site.

We need to ensure that only allowed web applications are scanned. Therefore, we must verify that you have control of any site you wish to include. This can be done by:

• Placing a file on a well-known location, on the site's server;
• Creating specific DNS records.

Asset

An asset is very similar to a site. The difference is that it is a domain instead of a URL. Additionally, an asset has no login or basic auth support. You can still have custom cookies and headers per asset.

As with the site, you will need to prove an asset's ownership. We have added some rules to make your life easier, if you already have verified a site and the domains match, the validation is fast-tracked.

Scans

This is what you're here for. After configuring your target, you will want to run scans against it. You can either start a one off scan, or schedule one for later - recurring or not.

During the scan, we will spider and run several modules to check for security issues, which we call findings. You can check the findings even before a scan ends. If everything goes well, the scan will complete and that is it.

With some findings, our automated processes may have difficulties determining if it is a false positive or a legitimate issue. In these instances, a scan will be marked as under review, and we will further analyze the finding before making a decision. We will only show findings that, for some degree of confidence, are true positives. A finding that we are not sure of will never be displayed.

As much as we try to prevent it, a scan (or a sub-module) can malfunction. If this happens, a scan is marked as:

• "failed": the problem was irrecoverable;
• "completed with errors": some module failed but the scan itself completed.

During a scan, we try to determine what frameworks you are using and add this information to the site and asset objects discussed previously.

Findings

The last core concept is the finding, this is a security issue that we have found during our scans. If the same issue is found in a new scan it will not open a new finding but update the previous.

A finding will have a lot of information about the issue. Namely, where it was found, URL, insertion point (e.g. cookie), parameter, and method. Evidence we gathered, and the full request and response that we used. Sugestions of how to go about fixing it. A full description of the vulnerability is also present in the definition property. We also assign a severity and calculate the CVSS score for each.

Besides all this, there are also actions that you can perform on a finding. You can assign it to one user, leave comments for your team or add labels, and reduce or increase the severity.

If you don't plan on fixing the finding and accept the risk, or you think we reported a false positive, you can mark the finding to reflect that.