memory-hygiene

Pass

Audited by Gen Agent Trust Hub on Apr 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes shell commands, specifically rm -rf, to remove local directory contents at ~/.clawdbot/memory/lancedb/. This is a destructive operation intended for clearing the vector database.\n- [COMMAND_EXECUTION]: Provides instructions to implement a persistent cron job that executes shell commands and restores data to the memory database periodically.\n- [PROMPT_INJECTION]: The skill uses the memory_recall tool to fetch data from the vector database, which introduces a surface for indirect prompt injection if historical memory content is attacker-controlled.\n
  • Ingestion points: Recall operations targeting all memory content (query='*') are utilized in SKILL.md.\n
  • Boundary markers: No delimiters or safety instructions are defined to separate recalled data from agent instructions.\n
  • Capability inventory: The agent has the capability to execute shell commands (rm, restart) and manipulate memory storage via tools.\n
  • Sanitization: No input validation or content filtering is specified for data retrieved from memory storage.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 3, 2026, 10:00 PM
Security Audit — agent-trust-hub — memory-hygiene