Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Caldera is MITRE’s open adversary emulation platform. Agents on target hosts run chained ATT&CK techniques under a planner that decides the next step from what the previous one returned.

What Is MITRE Caldera?
Caldera is an open-source adversary emulation platform developed by MITRE and released in 2017. Where a script runs a fixed sequence, Caldera runs an operation: agents report what they find, and a planner selects the next technique based on the facts collected so far. That autonomy is what distinguishes it from simpler test harnesses, and it is why the output resembles an intrusion rather than a checklist.
The fact system is the interesting part. An ability that enumerates domain users writes those usernames as facts, and a later credential access ability consumes them. The operation adapts to the environment instead of executing a predetermined list, which is exactly the behaviour that defeats detections written around a specific command.
Caldera ships as a core with plugins, and several matter for defensive work. Stockpile provides the main library of abilities and adversary profiles. Atomic imports the Atomic Red Team test library, so individual techniques can be run inside an operation. Emu provides adversary profiles derived from MITRE’s own Center for Threat-Informed Defense emulation plans, including profiles modelled on named groups. Pathfinder adds attack path discovery, and Manx supplies a reverse shell contact for scenarios where a persistent agent is not wanted.
There is also a Caldera for OT project supplying abilities for industrial protocols, which is a rare capability in open tooling and should only be used in a laboratory.
The common deployment is a controlled exercise against a representative segment, run with the detection team either informed or not depending on the question being asked. The outputs that matter are the operation timeline, showing precisely what was executed and when, and the comparison against what the detection pipeline produced. Mapping those results onto a coverage map gives you evidence-based scoring for chained behaviour, not just for isolated techniques.
Caldera also serves as the execution engine inside a detection testing pipeline. Its REST API lets an operation be started, polled, and torn down from a scheduler, which means regression testing of chained scenarios can run nightly without anyone logging in.
Caldera is attacker tooling with a web interface. The server holds agent credentials and can execute commands across every host running an agent, so it belongs on a hardened, access-controlled system, never exposed to the internet, with its default credentials changed before first use. Agents should be removed after an exercise rather than left in place, and every operation should be logged with scope, approval, and time window so that resulting alerts can be attributed.
Run it in a lab or a controlled segment first. Abilities vary in impact, some write files or modify configuration, and cleanup is defined per ability rather than guaranteed. Agree the scope with system owners and keep a named person able to stop an operation immediately.
Compared with Atomic Red Team, Caldera chains techniques and adapts, which makes it better for testing whether a sequence is detected as one intrusion rather than as unrelated events. Compared with commercial breach and attack simulation, it is free and transparent but requires engineering effort, has no managed safety controls, and produces no executive reporting. Teams with the skills to run it get most of the capability for the cost of the time. Our red team services use emulation tooling of this kind alongside manual operation, because the automation reaches a point where a human operator is still required.
Only with careful scoping and owner agreement. It deploys agents that execute commands with the privileges they hold, abilities vary in impact, and cleanup is defined per ability rather than guaranteed. Start in a lab, move to a limited production segment with named approvers and a defined time window, keep someone able to halt an operation immediately, and remove the agents afterwards.
Atomic Red Team runs individual techniques on demand, one at a time, with a visible command for each. Caldera runs operations in which an agent collects facts and a planner chooses subsequent techniques based on them, producing a chained sequence that adapts to the environment. Use atomics to validate a specific rule and Caldera to test whether a multi-stage intrusion is recognised as one event.
No. It automates technique execution but has no judgement: it will not notice an unexpected opportunity, social engineer a helpdesk, improvise around an unfamiliar control, or decide that a different objective is more valuable. It is best used to maintain continuous assurance between human-led engagements, and to validate that the fixes from the last engagement are still in place.
Stockpile for the core ability library and adversary profiles, and Atomic to import the Atomic Red Team tests so individual techniques are available inside operations. Emu adds adversary profiles derived from MITRE emulation plans, which is useful once you know which groups matter to your industry. Add others only as a specific need arises, since each increases the capability of a sensitive system.
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.