Skip to content

Cloudflare HTTP requests

Overview

Cloudflare is a global network designed to make everything you connect to the Internet secure, private, fast, and reliable. In this documentation, you will learn how to collect and send Cloudflare HTTP requests to Sekoia.io.

  • Vendor: Cloudflare
  • Plan: Defend Core & Defend Prime
  • Supported environment: SaaS
  • Detection based on: Telemetry
  • Supported application or feature: Web logs, Web application firewall logs

Configure

Create the intake on Sekoia.io

Go to the intake page and create a new intake from the format Cloudflare.

Configure events forwarding on Cloudflare

Retrieve necessary information

First, you will have to retrieve configuration information. Connect to Cloudflare Console to collect the following :

  1. Cloudflare API Token

    • Go to My Profile, then on the left panel, click on API Tokens.
    • Click on the Create Token button and select the Create Custom Token entry.
    • Give a name to your token and set the following permissions:
    Scope Group Level
    Account Account Analytics Read
    Account Logs Read
    Account Logs Edit
    Zone Logs Read
    Zone Logs Edit
    • If you want zerotrust logs you should also add:
    Scope Group Level
    Account Zero Trust Read

    see the Cloudflare documentation

  2. Cloudflare Zone ID :

    • This information is specific to a Website.
    • On the left panel, click on Websites and select the Website you want.
    • On the right panel, there is an API section where you can retrieve the Zone ID.

Create a Logpush job

Configure a Logpush job with the following destination:

https://intake.sekoia.io/plain/batch?header_X-SEKOIAIO-INTAKE-KEY=<YOUR_INTAKE_KEY>

To do so, you can manage Logpush with cURL:

$ curl -X POST https://api.cloudflare.com/client/v4/zones/<CLOUDFLARE_ZONE_ID>/logpush/jobs \
-H "Authorization: Bearer <CLOUDFLARE_API_TOKEN>" \
-H "Content-Type: application/json" \
--data '{
    "dataset": "http_requests",
    "enabled": true,
    "max_upload_bytes": 5000000,
    "max_upload_records": 1000,
    "logpull_options":"fields=ClientIP,ClientRequestHost,ClientRequestMethod,ClientRequestURI,EdgeEndTimestamp,EdgeResponseBytes,EdgeResponseStatus,EdgeStartTimestamp,RayID&timestamps=unix",
    "destination_conf": "https://intake.sekoia.io/plain/batch?header_X-SEKOIAIO-INTAKE-KEY=<YOUR_INTAKE_KEY>"
}' # (1)
  1. will return 
    {
  "errors": [],
  "messages": [],
  "result": {
    "id": 146,
    "dataset": "http_requests",
    "enabled": false,
    "name": "<DOMAIN_NAME>",
    "logpull_options": "fields=<LIST_OF_FIELDS>&timestamps=rfc3339",
    "destination_conf": "https://intake.sekoia.io/plain/batch?header_X-SEKOIAIO-INTAKE-KEY=<YOUR_INTAKE_KEY>",
    "last_complete": null,
    "last_error": null,
    "error_message": null
  },
  "success": true
}

Important

Replace :

  • <YOUR_INTAKE_KEY> with the Intake key you generated in the Create the intake on Sekoia.io step.
  • <CLOUDFLARE_API_TOKEN> with the API Token you generated
  • <CLOUDFLARE_ZONE_ID> with the Zone ID you grabbed
Useful Cloudflare API endpoints

On their documentation, Cloudflare provides a list API endpoints you can use. Find below some useful endpoints:

  • https://api.cloudflare.com/client/v4/zones/<ZONE_ID>/logpush/jobs/<JOB_ID> to verify the job you previously created is correct (you need to specify the JOB_ID)
  • https://api.cloudflare.com/client/v4/zones/<ZONE_ID>/logpush/datasets/<DATASET>/jobs to get all the jobs for a specific dataset (dns_log, firewalls_events or http_requests in our case)
  • https://api.cloudflare.com/client/v4/zones/<ZONE_ID>/logpush/jobs/<JOB_ID> to update a job if you noticed a mistake after the creation of the job (wrong fields, wrong SEKOIA API Key...)

Raw Events Samples

In this section, you will find examples of raw logs as generated natively by the source. These examples are provided to help integrators understand the data format before ingestion into Sekoia.io. It is crucial for setting up the correct parsing stages and ensuring that all relevant information is captured.

{
    "ClientIP": "34.142.121.18",
    "ClientRequestHost": "foo-bar-baz.xyz",
    "ClientRequestMethod": "GET",
    "ClientRequestURI": "/wp1/wp-includes/wlwmanifest.xml",
    "EdgeEndTimestamp": 1658281702371000000,
    "EdgeResponseBytes": 279,
    "EdgeResponseStatus": 522,
    "EdgeStartTimestamp": 1658281671671000000,
    "RayID": "72d807ffeba5753d"
}

{
    "WAFMatchedVar": "",
    "WAFProfile": "unknown",
    "WAFRuleID": "",
    "WAFRuleMessage": "",
    "WorkerCPUTime": 0,
    "WorkerStatus": "unknown",
    "WorkerSubrequest": false,
    "WorkerSubrequestCount": 0,
    "ZoneID": 545468107,
    "ZoneName": "foo-bar-baz.xyz"
}

Detection section

The following section provides information for those who wish to learn more about the detection capabilities enabled by collecting this intake. It includes details about the built-in rule catalog, event categories, and ECS fields extracted from raw events. This is essential for users aiming to create custom detection rules, perform hunting activities, or pivot in the events page.

The following Sekoia.io built-in rules match the intake Cloudflare HTTP requests. This documentation is updated automatically and is based solely on the fields used by the intake which are checked against our rules. This means that some rules will be listed but might not be relevant with the intake.

SEKOIA.IO x Cloudflare HTTP requests on ATT&CK Navigator

Cloudflare HTTP Requests Rule Block Or Drop

Detects when one of Cloudflare Web Application Firewall (WAF) Managed rule blocked or dropped an HTTP request. It requires only Cloudflare HTTP requests logs.

  • Effort: master
Cloudflare WAF Correlation Alerts

Detection of multiple alerts (more than 5) triggered by the same source by Cloudflare detection rules

  • Effort: master
Covenant Default HTTP Beaconing

Detects potential Covenant communications through the user-agent and specific urls

  • Effort: intermediate
Cryptomining

Detection of domain names potentially related to cryptomining activities.

  • Effort: master
Discord Suspicious Download

Discord is a messaging application. It allows users to create their own communities to share messages and attachments. Those attachments have little to no overview and can be downloaded by almost anyone, which has been abused by attackers to host malicious payloads.

  • Effort: intermediate
Dynamic DNS Contacted

Detect communication with dynamic dns domain. This kind of domain is often used by attackers. This rule can trigger false positive in non-controlled environment because dynamic dns is not always malicious.

  • Effort: master
Exfiltration Domain

Detects traffic toward a domain flagged as a possible exfiltration vector.

  • Effort: master
Koadic MSHTML Command

Detects Koadic payload using MSHTML module

  • Effort: intermediate
Nimbo-C2 User Agent

Nimbo-C2 Uses an unusual User-Agent format in its implants.

  • Effort: intermediate
Potential Azure AD Phishing Page (Adversary-in-the-Middle)

Detects an HTTP request to an URL typical of the Azure AD authentication flow, but towards a domain that is not one the legitimate Microsoft domains used for Azure AD authentication.

  • Effort: intermediate
Potential Bazar Loader User-Agents

Detects potential Bazar loader communications through the user-agent

  • Effort: elementary
Potential Lemon Duck User-Agent

Detects LemonDuck user agent. The format used two sets of alphabetical characters separated by dashes, for example "User-Agent: Lemon-Duck-[A-Z]-[A-Z]".

  • Effort: elementary
Potential LokiBot User-Agent

Detects potential LokiBot communications through the user-agent

  • Effort: intermediate
Remote Access Tool Domain

Detects traffic toward a domain flagged as a Remote Administration Tool (RAT).

  • Effort: master
Remote Monitoring and Management Software - AnyDesk

Detect artifacts related to the installation or execution of the Remote Monitoring and Management tool AnyDesk.

  • Effort: master
SEKOIA.IO Intelligence Feed

Detect threats based on indicators of compromise (IOCs) collected by SEKOIA's Threat and Detection Research team.

  • Effort: elementary
Sekoia.io EICAR Detection

Detects observables in Sekoia.io CTI tagged as EICAR, which are fake samples meant to test detection.

  • Effort: master
TOR Usage Generic Rule

Detects TOR usage globally, whether the IP is a destination or source. TOR is short for The Onion Router, and it gets its name from how it works. TOR intercepts the network traffic from one or more apps on user’s computer, usually the user web browser, and shuffles it through a number of randomly-chosen computers before passing it on to its destination. This disguises user location, and makes it harder for servers to pick him/her out on repeat visits, or to tie together separate visits to different sites, this making tracking and surveillance more difficult. Before a network packet starts its journey, user’s computer chooses a random list of relays and repeatedly encrypts the data in multiple layers, like an onion. Each relay knows only enough to strip off the outermost layer of encryption, before passing what’s left on to the next relay in the list.

  • Effort: master

Event Categories

The following table lists the data source offered by this integration.

Data Source Description
Web logs Cloudflare act as a proxy and provide associated traffic logs
Web application firewall logs Cloudflare protect web application with its web application firewall and provide associated traffic logs

In details, the following table denotes the type of events produced by this integration.

Name Values
Kind ``
Category web
Type access

Transformed Events Samples after Ingestion

This section demonstrates how the raw logs will be transformed by our parsers. It shows the extracted fields that will be available for use in the built-in detection rules and hunting activities in the events page. Understanding these transformations is essential for analysts to create effective detection mechanisms with custom detection rules and to leverage the full potential of the collected data.

{
    "message": "{\"ClientIP\":\"34.142.121.18\",\"ClientRequestHost\":\"foo-bar-baz.xyz\",\"ClientRequestMethod\":\"GET\",\"ClientRequestURI\":\"/wp1/wp-includes/wlwmanifest.xml\",\"EdgeEndTimestamp\":1658281702371000000,\"EdgeResponseBytes\":279,\"EdgeResponseStatus\":522,\"EdgeStartTimestamp\":1658281671671000000,\"RayID\":\"72d807ffeba5753d\"}",
    "event": {
        "category": [
            "web"
        ],
        "dataset": "http_requests",
        "end": "2022-07-20T01:48:22.371000Z",
        "start": "2022-07-20T01:47:51.671000Z",
        "type": [
            "access"
        ]
    },
    "cloudflare": {
        "ClientIP": "34.142.121.18",
        "ClientRequestHost": "foo-bar-baz.xyz",
        "ClientRequestMethod": "GET",
        "ClientRequestURI": "/wp1/wp-includes/wlwmanifest.xml",
        "EdgeEndTimestamp": "1658281702371000000",
        "EdgeResponseBytes": 279,
        "EdgeResponseStatus": 522,
        "EdgeStartTimestamp": "1658281671671000000",
        "RayID": "72d807ffeba5753d"
    },
    "destination": {
        "address": "foo-bar-baz.xyz"
    },
    "http": {
        "request": {
            "method": "GET"
        },
        "response": {
            "bytes": 279,
            "status_code": 522
        }
    },
    "observer": {
        "type": "proxy",
        "vendor": "Cloudflare"
    },
    "related": {
        "ip": [
            "34.142.121.18"
        ]
    },
    "source": {
        "address": "34.142.121.18",
        "ip": "34.142.121.18"
    },
    "url": {
        "path": "/wp1/wp-includes/wlwmanifest.xml"
    }
}

{
    "message": "{\"WAFMatchedVar\":\"\",\"WAFProfile\":\"unknown\",\"WAFRuleID\":\"\",\"WAFRuleMessage\":\"\",\"WorkerCPUTime\":0,\"WorkerStatus\":\"unknown\",\"WorkerSubrequest\":false,\"WorkerSubrequestCount\":0,\"ZoneID\":545468107,\"ZoneName\":\"foo-bar-baz.xyz\"}\n\n",
    "event": {
        "category": [
            "web"
        ],
        "dataset": "http_requests",
        "type": [
            "access"
        ]
    },
    "cloudflare": {
        "WAFMatchedVar": "",
        "WAFProfile": "unknown",
        "WAFRuleID": "",
        "WAFRuleMessage": "",
        "WorkerCPUTime": 0,
        "WorkerStatus": "unknown",
        "WorkerSubrequest": false,
        "WorkerSubrequestCount": 0,
        "ZoneID": 545468107,
        "ZoneName": "foo-bar-baz.xyz"
    },
    "observer": {
        "type": "proxy",
        "vendor": "Cloudflare"
    }
}

Extracted Fields

The following table lists the fields that are extracted, normalized under the ECS format, analyzed and indexed by the parser. It should be noted that infered fields are not listed.

Name Type Description
@timestamp date Date/time when the event originated.
destination.address keyword Destination network address.
event.action keyword The action captured by the event.
event.category keyword Event category. The second categorization field in the hierarchy.
event.dataset keyword Name of the dataset.
event.end date event.end contains the date when the event ended or when the activity was last observed.
event.start date event.start contains the date when the event started or when the activity was first observed.
event.type keyword Event type. The third categorization field in the hierarchy.
http.request.bytes long Total size in bytes of the request (body and headers).
http.request.method keyword HTTP request method.
http.request.referrer keyword Referrer for this HTTP request.
http.response.bytes long Total size in bytes of the response (body and headers).
http.response.status_code long HTTP response status code.
network.protocol keyword Application protocol name.
observer.type keyword The type of the observer the data is coming from.
observer.vendor keyword Vendor name of the observer.
rule.id keyword Rule ID
rule.ruleset keyword Rule ruleset
source.as.number long Unique number allocated to the autonomous system.
source.geo.country_name keyword Country name.
source.ip ip IP address of the source.
source.port long Port of the source.
tls.cipher keyword String indicating the cipher used during the current connection.
tls.version_protocol keyword Normalized lowercase protocol name parsed from original string.
url.path wildcard Path of the request, such as "/search".
user_agent.original keyword Unparsed user_agent string.

For more information on the Intake Format, please find the code of the Parser, Smart Descriptions, and Supported Events here.