T-3004ValidatedActive
Memory Credential Mining
Extract credentials stored in agent memory from previous sessions or conversations
Tactic
Credential Harvest · Stage 3
Extract API keys, tokens, and credentials from agent context and connected services
Attack class
MEM-POISON
Injecting malicious entries into agent persistent memory to maintain control across sessions
Evidence grade
ValidatedReproduced in a controlled lab environment (DVAA) with documented steps.
DVAA validation
L2-05
Reproductions in Damn Vulnerable AI Agent, the OpenA2A intentionally-broken agent for kill-chain validation.
Honeypot
AgentPwn coverage
data-exfiltrationagentpwn.com/learn ↗
An AgentPwn trap page produces a payload tagged with this technique class. Following the AgentPwn taxonomy of trap pages shows what an agent encounters.
Conversation-history-leak tiers surface credentials from prior context.
Detect
Detection · HackMyAgent
MEM-001MEM-002MEM-003MEM-004MEM-005MEM-006
npx hackmyagent secure --ciLive = implemented in hackmyagent; queued = declaredDefend
Defense · OASB controls
Reference
How to cite
AI Agent Threat Matrix T-3004 (Memory Credential Mining). OpenA2A, 2026. https://threats.opena2a.org/techniques/T-3004