Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Sigma is an open YAML format for writing detection logic once and converting it into the query language of whichever SIEM you happen to run.

What Are Sigma Rules?
Detection logic has always been trapped in the product that runs it. A rule written in one SIEM’s query language is worthless in another, which means a platform migration throws away years of tuning, and a published piece of research describing a new attack cannot simply be deployed by the people who read it. Sigma exists to break that lock-in. It is an open YAML format for describing what to look for in log data, created by Florian Roth and Thomas Patzke in 2017, with converters that translate a rule into the query language of whichever platform you use.
The usual comparison is to YARA for files and Snort for network traffic. Sigma occupies the equivalent position for log events: a shared, readable, vendor-neutral way of writing down a detection so that other people can use it.
title: Suspicious Rundll32 Execution via comsvcs MiniDump
id: 9bd1a1b2-3c4d-4e5f-8a90-1b2c3d4e5f60
status: experimental
description: Detects LSASS credential dumping using the comsvcs.dll MiniDump export
references:
- https://attack.mitre.org/techniques/T1003/001/
author: Privia Security
date: 2026/03/11
logsource:
category: process_creation
product: windows
detection:
selection:
Image|endswith: '\rundll32.exe'
CommandLine|contains|all:
- 'comsvcs'
- 'MiniDump'
condition: selection
falsepositives:
- Rare administrative diagnostic collection
level: high
tags:
- attack.credential_access
- attack.t1003.001
Three blocks carry the meaning. The logsource block says what kind of data the rule applies to, using an abstract taxonomy rather than a product-specific log name. The detection block defines one or more named selections with field matches, and modifiers such as contains, startswith, endswith, re, and all refine them. The condition combines those selections with boolean logic, including the common pattern selection and not filter for excluding known-good activity.
The surrounding metadata is not decoration. The UUID gives the rule a stable identity across repositories, status warns whether it has been tested, falsepositives tells the analyst what else fires it, and the ATT&CK tags make coverage measurable.
Conversion was originally handled by the sigmac tool and is now done by sigma-cli with pySigma backends, which exist for Splunk, Microsoft Sentinel, Elasticsearch, QRadar, and many others. The part that newcomers miss is the processing pipeline. Field names differ everywhere — what Sysmon calls CommandLine may be process.command_line in one schema and ProcessCommandLine in another — and the pipeline maps the abstract taxonomy onto your actual field names and indices.
Build that pipeline once for your environment and every subsequent rule converts correctly. Skip it, and conversion produces syntactically valid queries that silently match nothing, which is the most dangerous failure mode available: a detection that appears deployed and is not.
Sigma handles single-event matching extremely well, and that covers a large share of practical detections. Later revisions of the specification added correlation rules for counting events, grouping within a time window, and chaining related rules, though backend support for these varies more than for basic rules. Logic that depends on joins across data sets, lookups against external tables, or statistical baselining is still better written natively.
Treat conversion output as a draft rather than a finished query. A generated search may ignore partitioning, scan a wider time range than necessary, or express a condition in a way that performs badly at your volume. Review it, optimise it, and test it before it reaches production, exactly as you would a rule written by hand.
The SigmaHQ repository holds thousands of community rules covering most published attack techniques, which makes it a reasonable way to find coverage gaps quickly. It is not a rule set to deploy wholesale. Community rules are written against generic environments, many carry an experimental status, and a rule tuned for someone else’s estate will generate noise in yours.
The workable approach is to convert selectively, run new rules in a monitoring mode first, record the ones you keep in your use case catalogue, and hold your modifications in your own repository with the upstream rule ID preserved so updates can be tracked. Rules you write from your own incidents are often the most valuable in the set, and contributing them back costs nothing.
Sigma complements rather than replaces the other open detection formats: YARA for file content and Sigma for log events. At Privia Security, we build and tune portable detection content of this kind through our SOC maturation services.
They operate on different data. YARA matches patterns inside files and process memory, so it identifies malware samples and families. Sigma matches conditions in log events, so it identifies behaviour recorded by endpoints, servers, and cloud services. Mature teams use both: YARA for sweeping systems for known tooling, Sigma for detecting the activity that tooling produces.
Not safely. Public rules target generic environments, many are marked experimental, and each organisation’s normal activity differs. Convert a selected subset, run them in a monitoring mode where alerts go to the engineering team, measure the volume and true positive rate, tune the exclusions, and only then promote them to the analyst queue.
Almost always field mapping. The abstract Sigma taxonomy has to be translated to your schema by a processing pipeline, and without one the generated query references fields that do not exist in your index. The query runs, returns nothing, and looks healthy. Build and test the pipeline for your environment before converting anything else.
Yes, for three reasons beyond portability: the public rule base gives you tested detection content for free, YAML in version control is far easier to review and track than queries edited in a console, and the day you change platform, migrate tenants, or add a second analysis environment, the logic comes with you rather than being rewritten.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.