memento-reflect

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a reflective learning protocol that involves analyzing conversation history to determine task success or failure. This analysis is restricted to the current session context.
  • [DATA_EXFILTRATION]: File operations are limited to reading and writing a specific metrics file located in ~/.claude/memento/metrics.json or the local project directory. No access to sensitive system paths (e.g., SSH keys, AWS credentials) or unauthorized network communications were detected.
  • [COMMAND_EXECUTION]: The skill mentions external command patterns like /memento create and /memento optimize. These are presented as user-facing suggestions for manual or guided execution and do not involve hidden or arbitrary shell command injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 07:58 AM
Security Audit — agent-trust-hub — memento-reflect