Source overview

elastic/protections-artifacts

Elastic's malware protection signatures — YARA rules plus behavioral EQL rules.

elastic-protections-artifacts · YARA · upstream repo ↗ · Elastic License 2.0

Rules indexed 4401 1 atomic-IOC hidden · include
ATT&CK techniques 243
Newest rule 2026-10-09 checked 2026-10-10

How detections work here

These are the artifacts that ship inside Elastic Defend, Elastic's endpoint agent — the protections that run on the host itself. They are not the hunting content in elastic-detection-rules, which runs as queries in Elasticsearch. Different execution model, different repository, easy to confuse.

This is the one source in this index carrying two formats, and they work differently:

  • YARA (2,087 rules) — pattern matching against files and process memory on the endpoint.
  • Elastic TOML behavior rules (1,249) — EQL-style conditions evaluated against endpoint event streams in real time.

The format facet on this page separates them, and you should use it: a YARA rule and a behavior rule answer different questions.

How rules are written

The YARA rules follow Elastic's naming convention — Windows_Trojan_AgentTesla, Linux_Backdoor_AutoColor, ByteCode_MSIL_Ransomware_... — encoding platform, category and family in the rule name itself. That convention is worth knowing because it makes the rule name a usable search key.

The behavior rules under behavior/rules/ are TOML with an EQL-like query field describing a sequence or condition over endpoint events (process, file, network, registry), plus metadata for severity and ATT&CK.

What fires an alert

For YARA, a scan match on the endpoint. For behavior rules, the condition matching live event telemetry — which means they can express sequence and temporal relationships that the YARA half cannot.

Reading a rule on this site

Both formats land in the same schema, so filter by format before comparing rules from this source with each other.

ATT&CK coverage comes almost entirely from the behavior half — YARA carries no technique field. So the technique count on this page describes roughly a third of the rules, not all of them.

Gotchas

  • This is protection content, not hunting content. These rules are tuned to run inside an agent with the agent's context and its own precedence rules. Lifting a behavior rule out and running it as a standalone query will not reproduce what Elastic Defend does with it.
  • The two formats are mixed in every unfiltered view of this source. A count of "rules in elastic-protections-artifacts" is a count of two unrelated things unless you split it.

Rungs

Reaches TTPs — 482 rules, 10.9% of the ruleset. A rung is claimed at 20 rules and 1.0% of the source, so that one rule cannot speak for a small ruleset and a rounding error cannot speak for a large one.

Shares are of all 4402 live rules in this source, including the atomic-IOC and non-detection content the count above hides by default. The pending band is drawn to scale for the same reason: it is the width of what no classifier has read, and a bar without it would look more complete than the index is. What the rungs mean.


Severity

Status


Platforms

windows 2812 Linux 1170 macos 397 multi 55

Formats

YARA 3069 Elastic TOML 1332


Recently modified

all →
Application Javascript Injection via Nodejs 2026-10-09
Arbitrary Python Code Execution via Nodejs 2026-10-09
Cloud Credential Files Accessed by Process in Suspicious Directory 2026-10-09
Connection to WebService by an Unsigned Binary 2026-10-09
Connection to WebService by a Signed Binary Proxy 2026-10-09

Detection rules belong to the projects that publish them and remain under their own licenses. This site indexes and links to them; it claims no rights in them.