anti-amnesia

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze recent tool results and conversation history to answer user questions. This creates a surface for indirect prompt injection where instructions embedded in previous data inputs (e.g., file contents read by a tool) could potentially influence the agent when it recalls that record.
  • Ingestion points: Tool results, assistant responses, and conversation summaries as defined in SKILL.md.
  • Boundary markers: The instructions do not specify any delimiters or safety markers to differentiate between recorded data and instructions within the history.
  • Capability inventory: The skill explicitly advises against calling tools to rediscover information, which limits immediate execution risk, though the agent's general tool access remains a factor.
  • Sanitization: There are no instructions for filtering or sanitizing the content of the records being processed.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 11:56 AM
Security Audit — agent-trust-hub — anti-amnesia