Author: John Hartman

Background

Rapid7 Insight Agent and InsightVM Scan Assistant are executables that can be deployed to assist in understanding the vulnerabilities in your environment. Frequently there are questions around when and where you would deploy each, if you need both, what they actually monitor, etc. This article will answer those questions, but first let's look at each executable in more detail.

Rapid7 Insight Agent

Notice the name of this starts with Rapid7. This is important, because the Insight Agent can be used for multiple tools, primarily InsightVM and InsightIDR. However, the agent does different things for each. For InsightIDR, the agent monitors process start and stop events and has log collection abilities. For InsightVM, the Insight Agent is used for assessment of vulnerabilities. In this article, we’ll focus on using Insight Agent for InsightVM.

The Insight Agent performs an "assessment" roughly every six hours. Notice the word "assessment" and not "scan". The Insight Agent has the permissions necessary to gather information about the asset that it is installed on and then forward that information directly to the Insight Platform. The Insight Platform then forwards that data to the InsightVM Security Console. The Security Console then takes that data and runs it against a scan template to determine what vulnerabilities that asset has. Once done, the Security Console updates its own database with the results for that asset and then on the interval of communication with the Insight Platform it will forward the assessment results back to the Insight Platform.

With the Insight Agent, you do not determine a scan schedule or have the ability to kick off ad hoc or remediation scans on that asset. As noted above, assessments occur every six hours. However, not every agent is being assessed on the same six hour interval. The schedule is maintained entirely by the Insight Platform.

Another key takeaway about the communication path mentioned above: The Insight Agent does not communicate directly to the console. This makes Insight Agent particularly beneficial when it comes to protecting your remote workforce. Given that remote assets are not on your network, you typically cannot scan them directly. So, Insight Agent is the main option to view the vulnerabilities for those assets.

Recently, Rapid7 released the ability to perform Policy Scans using the Insight Agent as well. This ability is limited to assets that are available for the installation of the InsightAgent though (Windows, Linux, Mac), however that typically covers a large portion of the policy scanning needed. Policy scanning occurs every 12 hours.

InsightVM Documentation: Insight Agents with InsightVM

InsightVM Scan Assistant

The InsightVM Scan Assistant executable is solely dedicated to InsightVM and is configured to display a certificate on port 21047. The Scan Assistant can only be used when being accessed from a scan engine (distributed or local). Unlike the Insight Agent, which monitors and performs assessments on a scheduled basis, the Scan Assistant is dormant unless called upon by a Scan Engine either through a manual or scheduled scan configured from the Security Console.

For this to work, first you must generate a certificate from InsightVM in the credential setup. Then, you need to edit any scan templates being used to additionally look for port TCP 21047 on both Asset and Service discovery. From there, the Scan Engine will use those credentials and look for that port to be open on the endpoint servers. If the certificate being presented on that port matches the certificate created within InsightVM, the scan engine will use it to authenticate to the endpoint asset. The Scan Assistant has the permissions necessary to perform all local checks on the endpoint asset.

Using the Scan Assistant instead of regular domain credentials offers better security, as it eliminates the possibility of a domain account with elevated permissions to be used in your environment. Additionally, the Scan Assistant has proven to be more efficient and perform scans quicker than domain credentials.

InsightVM Documentation: Using the Scan Assistant

So why use both?

As stated above, the two executables are completely independent of each other. The Insight Agent communicates to the platform whereas the Scan Assistant talks directly to the Scan Engine performing the scan. The Insight Agent is not configurable in its scheduled assessment whereas the Scan Assistant is completely dormant until scanned and is completely reliant on an administrator configuring scanning.

So, WHERE should each executable be installed? I would suggest having the Insight Agent on all local and remote assets—everything capable of having the Insight Agent installed. For the Scan Assistant, only internal assets would be applicable. You could install the Scan Assistant on remote assets as well, if you have a policy that requires users to connect to the VPN on set schedules and you plan to scan through that VPN or office wi-fi. However, in most situations, the Insight Agent is the only way to assess your remote assets.

So that brings us to the internal assets that should have BOTH the Insight Agent and the Scan Assistant installed. You might be asking ‘why in the world would I want to deploy yet another executable if the Insight Agent is already performing the assessment on those assets?’ Well, let's circle back to the fact that the Insight Agent is only performing the local checks. So, you will need to perform at least monthly scanning of those assets to view network vulnerabilities. Additionally, as mentioned above, the Insight Agent is incapable of kicking off an ad-hoc scan. This is where the Scan Assistant comes into play for remediation scans specifically.

Scenario: I have an asset "abc.company.com." InsightAgent discovers a local vulnerability on the asset at 10AM and it's only 1030AM. I send the finding off to my system administrator to patch the vulnerability immediately. By 11AM the vulnerability is patched, and I want to verify that the vulnerability has been remediated. Without a credentialed scan, I have to wait another five hours before InsightAgent conducts another assessment. However, with the Scan Assistant I can immediately kick off an authenticated vulnerability scan against that asset to determine that the vulnerability is no longer present.

The other main use case for the Scan Assistant is to take advantage of the full breadth of the Policy Scanning. Currently, InsightAgent can only assess up to 100 different policies and can only assess for the default values of the policies through CIS or DISA.

Using the Scan Assistant with the scan engine you have access to ALL categories of Policy Scans, including CIS, DISA, FDCC, and USGCB. Additionally, you can use the custom policy builder to edit values within typical benchmarks. For example, you might change the minimum password length from 14 characters to 20 characters if that's what your internal policy dictates.

Trying to deal with a large network can be difficult. All too often, engineers and admins don't know the full scope of their environment and have trouble defining the actual subnets and the systems that exist on those subnets. They know of a couple /24 subnets here or there, but it's very possible they're missing a few. Once you get over a couple thousand assets, it can get fairly unruly pretty quick. Different teams own different servers and different network ranges. With regards to InsightVM, how do you know what sites create if you don't even know what you own?

Luckily, in InsightVM, we can use a little bit of SQL, an overarching site with a ping sweep, and a nifty little tag to help get a handle on things – all outside any third-party software or  other management tools you may acquire to help you wrangle in your IP space. This method in InsightVM lets you find all live assets and identify all network spaces being used in your environment. Then, we can correlate this list against our known subnets and begin building out defined sites for scanning. As we create our known sites, we can start whittling down the number of unknown or undefined subnets.

1. Ping Sweep template

The first step is to create a new scan template dedicated solely to a ping sweep. This template isn't scanning for any other services or ports, fingerprinting, or performing any other action –  it is simply sending pings to see what is alive. If we get a response back, we assume there is a live asset there, and this will help build out our known networks.

Create your template using these screenshots as guidance. Note that pretty much everything is off except ICMP and ARP pings, and we're not treating TCP resets as live assets (we don't want firewalls throwing us off). This scan should take just a few minutes to complete, as it's not doing all the other functions that a typical scan can do.

2. Overarching site

The second step in this process is to create an overarching site. Give it a simple name like "Full Network" or whatever floats your boat. What's important is that, within this site, you define as large of a network range as you know of. Think /16 here, or even a couple /16 networks. I don't know your network, so use your judgment as to what you think exists. The idea is to be as broad as possible.

Now, within this site, set the default scan template as your ".Ping Sweep" template, as in my example above. Set your default scan engine or pool, and then save and scan.

What you should get back now is a full list of every live IP that exists within the defined network. If your defined network includes all the possible IP space, and we are assuming that all assets are online and able to respond, then you should have a pretty robust list of found assets.

3. Known Networks report

The next step is to go to the Reports tab and create a SQL Query Export. Throw the following SQL query in the definition, and scope the query from the GUI to your "Full Network" site.

WITH a AS (
SELECT
asset_id,
CONCAT(split_part(ip_address,'.',1),'.',split_part(ip_address,'.',2),'.',split_part(ip_address,'.',3),'.0/24') AS Network
FROM dim_asset
)
 
SELECT DISTINCT Network
FROM a
ORDER BY Network ASC

Save and run this report, and you will get a CSV output of all the /24 networks that have at least one live IP in them. You can use this CSV to compare to your known list of networks and start defining the actual sites within your environment. For example, if this report lists out 10.0.0.0/24 and you know that network as your main corporate server’s VLAN, then you can include that network into a separate site for vulnerability scanning.

4. Dynamic tagging

Now that we've started defining our known networks into sites, we need to create a dynamic tag that gets applied to all assets within any site. Now, in my example, I exclude the Rapid7 Insight Agents site, because depending on your environment and whether people are working from home, the Insight Agent may report the IP of their computer when logged onto their home network. We obviously can't scan home networks, so we want to exclude this site to deter any of that bad data.

Create a dynamic tag with several lines to include each site. Note that if your site structure is large enough that you have hundreds of sites, you may want to use the API for this part, but we won't go into that here – that's a whole other conversation.

In my example below, I only have four sites – keep in mind I did not select the Rapid7 Insight Agents or my Full Network site. Make sure the operator is set to match ANY of the specified filters. Apply a tag called "Defined Network" to this criteria to tag all assets within a defined site.

You could also optionally create a secondary tag for "Undefined Networks," but it's not exactly necessary for this process. The below query would get you the Undefined Network assets. Basically, the query is just looking for any assets that don't have the Defined Network tag and are not in the Rapid7 Insight Agents sites.

5. Undefined Networks report

Now, we can set up our secondary SQL report to show us all networks that are not defined within the scope of a site. Once again, go to the Reports tab, create a SQL Query Export report, and throw this query into the definition.

WITH a AS (
SELECT
asset_id,
CONCAT(split_part(ip_address,'.',1),'.',split_part(ip_address,'.',2),'.',split_part(ip_address,'.',3),'.0/24') AS Network
FROM dim_asset
)
 
SELECT DISTINCT Network
FROM a
 
WHERE a.asset_id NOT IN (
SELECT DISTINCT asset_id
FROM dim_asset
LEFT JOIN dim_tag_asset USING (asset_id)
LEFT JOIN dim_tag USING (tag_id)
WHERE tag_name = 'Defined Network'
)
 
ORDER BY Network ASC

Save and run this report, and you will get a new CSV that lists out all /24 networks where there was at least one live asset found but the assets are within a /24 that has not been defined within the scope of a created site. You can use this CSV to work your way through those networks to determine what they are and who owns them and then ensure they are included in future or current sites.

Large environments with unknown network components can be difficult to manage and monitor for vulnerabilities. These five steps in InsightVM help make the process easier and more intuitive, so you can maintain better oversight and a stronger security posture within your environment.

Additional reading:

• What’s New in InsightVM and Nexpose: Q2 2022 in Review
• InsightVM Release Update: Let’s Focus on Remediation for Just a Minute
• Maximize Your VM Investment: Fix Vulnerabilities Faster With Automox + Rapid7