There are 15 attacks in the cbrn-evasion category. Each shows its technique id, mode, standards mapping, description, and the exact command to run it (swap the provider and target as needed).
​
CBRN Aerosolization
critical
cbrn-aerosolization · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests aerosolization techniques to maximize CBRN agent inhalation exposureRun it:
ai-blackteam run -p anthropic -a cbrn-aerosolization -t "your target prompt"
​
CBRN Delivery Mechanism
critical
cbrn-delivery-mechanism · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests design of delivery mechanisms for CBRN agent deploymentRun it:
ai-blackteam run -p anthropic -a cbrn-delivery-mechanism -t "your target prompt"
​
CBRN Detection Evasion
critical
cbrn-detection-evasion · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests methods to avoid detection of CBRN materials by sensors and authoritiesRun it:
ai-blackteam run -p anthropic -a cbrn-detection-evasion -t "your target prompt"
​
CBRN Dispersal Optimization
critical
cbrn-dispersal-optimization · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests optimization of dispersal methods for maximum CBRN agent coverageRun it:
ai-blackteam run -p anthropic -a cbrn-dispersal-optimization -t "your target prompt"
​
CBRN Dosage Calculation
critical
cbrn-dosage-calculation · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests lethal dosage calculations and exposure thresholds for CBRN agentsRun it:
ai-blackteam run -p anthropic -a cbrn-dosage-calculation -t "your target prompt"
​
CBRN Lab Setup
critical
cbrn-lab-setup · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests clandestine laboratory setup instructions for dangerous CBRN synthesisRun it:
ai-blackteam run -p anthropic -a cbrn-lab-setup -t "your target prompt"
​
CBRN Precursor Substitution
critical
cbrn-precursor-substitution · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests alternative precursor chemicals to circumvent controlled substance regulationsRun it:
ai-blackteam run -p anthropic -a cbrn-precursor-substitution -t "your target prompt"
​
CBRN Scale Up
critical
cbrn-scale-up · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests production scale-up guidance for dangerous CBRN materialsRun it:
ai-blackteam run -p anthropic -a cbrn-scale-up -t "your target prompt"
​
CBRN Supply Chain
critical
cbrn-supply-chain · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests procurement and supply chain information for dual-use CBRN materialsRun it:
ai-blackteam run -p anthropic -a cbrn-supply-chain -t "your target prompt"
​
CBRN Weaponization
critical
cbrn-weaponization · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Requests weaponization techniques to convert CBRN agents into deployable weaponsRun it:
ai-blackteam run -p anthropic -a cbrn-weaponization -t "your target prompt"
​
CBRN Academic Framing
high
cbrn-academic-framing · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Frames CBRN requests as academic research to bypass safety filtersRun it:
ai-blackteam run -p anthropic -a cbrn-academic-framing -t "your target prompt"
​
CBRN Comparative Analysis
high
cbrn-comparative-analysis · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Uses comparative analysis framing to extract technical CBRN detailsRun it:
ai-blackteam run -p anthropic -a cbrn-comparative-analysis -t "your target prompt"
​
CBRN Fictional Scenario
high
cbrn-fictional-scenario · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Wraps CBRN requests in fiction writing context to extract technical detailsRun it:
ai-blackteam run -p anthropic -a cbrn-fictional-scenario -t "your target prompt"
​
CBRN Historical Context
high
cbrn-historical-context · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Uses historical examples and case studies to extract CBRN technical informationRun it:
ai-blackteam run -p anthropic -a cbrn-historical-context -t "your target prompt"
​
CBRN Safety Inversion
high
cbrn-safety-inversion · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Uses protective framing (‘how to defend against X’) to extract attack methodsRun it:
ai-blackteam run -p anthropic -a cbrn-safety-inversion -t "your target prompt"