AML.CS0051
OpenClaw Command & Control via Prompt Injection
- Type
- Exercise
- When
- 2026-02-03
- Actor, as ATLAS names it
- Not stated
- Target
- OpenClaw
- ATLAS attributes it to HiddenLayer (security evaluation).
Techniques, in order
- 1Search Open Websites/Domains: Code RepositoriesReconnaissance · AML.T0095.000
- 2Acquire Public AI Artifacts: AI Agent ConfigurationResource Development · AML.T0002.002
- 3Discover LLM System Information: Special Character SetsDiscovery · AML.T0069.000
- 4Discover LLM System Information: System Instruction KeywordsDiscovery · AML.T0069.001
- 5LLM Prompt CraftingAI Attack Adaptation · AML.T0065
- 6Acquire InfrastructureResource Development · AML.T0008
- 7Stage CapabilitiesResource Development · AML.T0079
- 8MasqueradingDefense Evasion · AML.T0074
- 9Drive-by CompromiseInitial Access · AML.T0078
- 10LLM Prompt Injection: IndirectExecution · AML.T0051.001
- 11LLM JailbreakDefense Evasion · AML.T0054
- 12AI Agent Tool InvocationExecution · AML.T0053
- 13Modify AI Agent ConfigurationPersistence · AML.T0081
- 14LLM Prompt Injection: DirectExecution · AML.T0051.000
- 15AI Agent Context Poisoning: ThreadPersistence · AML.T0080.001
- 16AI AgentCommand and Control · AML.T0108
- 17Machine Compromise: Local AI AgentImpact · AML.T0112.000
From MITRE ATLAS 2026.10 (Apache-2.0). ShieldMap lists who and what was involved and the techniques used, and links to ATLAS for the write-up; it does not copy the steps. Attribution is ATLAS's, not ShieldPi's.