memory-reflect
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted interaction data from daily logs to update long-term memory, creating a surface for indirect prompt injection.\n
- Ingestion points: The skill reads content from daily notes using
read_noteand summarizes history fromrecent_activity, both of which contain raw user input from past conversations.\n - Boundary markers: The instructions lack explicit requirements for the agent to use delimiters or ignore embedded instructions within the ingested text during the evaluation phase.\n
- Capability inventory: The agent possesses the capability to modify the persistent
MEMORY.mdfile and log entries to daily notes based on the processing of this untrusted data.\n - Sanitization: No specific filtering, validation, or sanitization steps are defined to prevent the agent from extracting and consolidating malicious instructions into its long-term memory.
Audit Metadata