AGENTFORCE-8.1: Context Traversal Exfiltration (ForcedLeak Mitigation)

🔴 High · Agentic Architecture

Detects Deep Object Graph Traversals in Agent Prompts. Agents utilizing multi-hop relational merge fields (e.g. {!Record.Rel1.Rel2.Field}) are susceptible to Context Traversal Data Exfiltration (like the ForcedLeak exploit). Deep traversals exponentially broaden the agent's PII attack surface via Indirect Prompt Injection.

Details

Field	Value
Rule ID	AGENTFORCE-8.1
Severity	High
Category	Agentic Architecture
Compliance	NIST_AI_RMF

Remediation

Refer to the SquireX documentation for remediation guidance specific to AGENTFORCE-8.1.

See Also

• Rule catalog index
• SquireX scan documentation