searching-messages

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (MEDIUM): The letta messages search and letta messages list commands pull data from past conversation logs, which are untrusted ingestion points.
  • Ingestion points: Conversation content retrieved by the letta CLI tool in SKILL.md.
  • Boundary markers: Absent; the documentation does not suggest using delimiters or 'ignore embedded instructions' prompts to isolate retrieved history from current instructions.
  • Capability inventory: The retrieved data is used to 'recall context', directly influencing the agent's reasoning and future responses.
  • Sanitization: Absent; there is no evidence of filtering or sanitizing the content of past messages before they are added back into the active context window.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 09:43 PM