claude-mem

Warn

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses npx to download and install its core components from the NPM registry.
  • [COMMAND_EXECUTION]: Configures multiple lifecycle hooks that automatically execute code during agent startup, tool usage, and shutdown.
  • [DATA_EXFILTRATION]: Systematically records all tool inputs and outputs, creating a repository of potentially sensitive user and system information.
  • [PROMPT_INJECTION]: Subject to indirect prompt injection through the automated re-injection of past context.
  • Ingestion points: Tool activity across all sessions (SKILL.md).
  • Boundary markers: Absent.
  • Capability inventory: Bash, Read, and Write tools (SKILL.md).
  • Sanitization: Not performed on captured data.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 8, 2026, 02:29 PM
Security Audit — agent-trust-hub — claude-mem