Go to file
Thomas Strömberg b0549b56e2
Merge pull request #393 from tstromberg/oct16
net traffic & security framework: Broaden exceptions to reduce false positives
2024-10-16 10:14:59 -04:00
.github Add Harden Runner audit configs 2024-03-12 11:51:40 -06:00
detection refactor to reduce false positives 2024-10-16 09:44:19 -04:00
fragments fpr: macOS, yubikey, Premiere, dnf, vagrant, etc 2023-05-23 11:31:37 -04:00
images v0.0.1 2022-10-13 09:11:17 -04:00
incident_response Add events and extra tags to relevant event-based queries 2024-09-24 15:36:03 -04:00
policy Run reformat 2024-02-16 17:21:00 -05:00
vulnerabilities split detection pack into subpacks 2023-09-20 17:43:39 -04:00
.gitignore v0.0.1 2022-10-13 09:11:17 -04:00
LICENSE.txt Add license file 2022-10-13 09:21:11 -04:00
Makefile policy.conf: Fix makefile typo (extra o) 2024-09-24 15:57:29 -04:00
README.md Update query count from 220 to 250 2023-07-03 07:37:01 -04:00
osquery.conf Simplify makefile, reduce config targets to 4 2024-01-09 16:56:40 -05:00

README.md

osquery-defense-kit

Actions Status Latest Release stable

osquery queries for Detection & Incident Response, containing 250+ production-ready queries.

osquery-defense-kit

ODK (osquery-defense-kit) is unique in that the queries are designed to be used as part of a production detection & response pipeline. The detection queries are formulated to return zero rows during normal expected behavior, so that they may be configured to generate alerts when rows are returned.

At the moment, these queries are predominantly designed for execution on POSIX platforms (Linux & macOS). Pull requests to improve support on other platforms are fully welcome.

Requirements

  • osquery v5.7.0 or above
  • macOS or Linux
  • If you plan to do local development you will also need Go v1.20+ for osqtool

Usage

Local Detection

Run make detect for point-in-time detection. This will not detect as much as a production installation as it will not have access to historical events.

Production Detection

Download a released query pack into a convenient location, and point to these files from the packs stanza of your osquery.conf file

Local Data Collection for IR

Run make collect. This is particularly useful for before/after analysis.

Local pack generation

Run make packs. For more control, you can invoke osqtool directly, to override default intervals or exclude checks.

Local verification testing

Run make verify

File Organization

  • detection/ - Threat detection queries tuned for alert generation.
  • policy/ - Security policy queries tuned for alert generation.
  • incident_response/ - Data collection to assist in responding to possible threats. Tuned for periodic evidence collection.

The detection queries are further divided up by MITRE ATT&CK tactics categories.

At release time, the queries are packed up in osquery query pack format. See Local Pack Generation for information on how to generate your own packs at any time.

Case Studies

Linux: Shikitega (September 2022)

https://cybersecurity.att.com/blogs/labs-research/shikitega-new-stealthy-malware-targeting-linux

Here is a partial list of what queries would have fired an alert based on these queries:

  • Initial Dropper Execution, detected by:
    • execution/tiny-executable-events.sql
    • execution/tiny-executable.sql
  • Next Stage Dropper Execution, detected by:
    • execution/tiny-executable-events.sql
    • execution/tiny-executable.sql
    • execution/unexpected-shell-parents.sql
  • Escalation Prep, detected by:
    • execution/sketchy-fetchers.sql
    • execution/sketchy-fetcher-events.sql
    • c2/unexpected-talkers-linux.sql
    • c2/exotic-command-events.sql
    • c2/exotic-cmdline.sql
  • Escalation Tool Execution detected by:
    • execution/unexpected-executable-permissions.sql
    • execution/unexpected-executable-directory-linux.sql
    • execution/unexpected-tmp-executables.sql
    • c2/exotic-command-events.sql
    • c2/exotic-cmdline.sql
    • initial_access/unexpected-shell-parents.sql
    • evasion/missing-from-disk-linux.sql
  • Privilege Escalation detected by:
    • privesc/unexpected-setxid-process.sql
    • privesc/unexpected-privilege-escalation.sql
    • privesc/events/unexpected-privilege-escalation-events.sql
    • evasion/name_path_mismatch.sql
  • Persistence detected by:
    • persistence/unexpected-cron-entries.sql
    • execution/unexpected-executable-directory-linux.sql

macOS: CloudMensis (April 2022)

https://www.welivesecurity.com/2022/07/19/i-see-what-you-did-there-look-cloudmensis-macos-spyware/

Here is a partial list of what stages would have been detected by particular queries:

  • Initial Dropper Execution, detected by:

    • c2/unexpected-talkers-macos.sql
  • Second Stage Execution, detected by:

    • execution/unexpected-executable-directory-macos.sql
    • persistence/unexpected-launch-daemon-macos.sql
    • execution/unexpected-mounts.sql
  • TCC Bypass, detected by:

    • evasion/unexpected-env-values.sql
  • Spy Agent Execution, detected by:

    • c2/unexpected-talkers-macos.sql
    • execution/exotic-command-events.sql
    • execution/unexpected-executable-directory-macos.sql

Policies

Contributions

Help Wanted! We support any new queries so long as they can be easily updated to address false positives.

Users may submit false positive exceptions for popular well-known software packages, but may be asked to provide evidence for the behavior.

Platform Support

While originally focused on Linux and macOS, we support the addition of queries on any platform supported by osquery.

In particular, we've been asked about Windows support: Chainguard doesn't have any Windows machines, but if you have Windows queries that you think would be useful and match our philosophy, we're more than willing to accept them!

False Positives

We endeavor to exclude real-world false positives from our detection queries.

Managing false positives is easier said than done - pull requests are welcome!

CPU Overhead

In aggregate, queries should not consume more than 2% of the wall clock time across a day on a deployed system.

Intervals

Deployed intervals are automatically determined based on the tags supported by the osqtool, which we use for pack assembly.