memory-os
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill provides templates (e.g.,
AGENTS.md) containing instructions for the agent's future behavior, such as a 'Load Sequence' that tells the agent to read memory files automatically ('Don't ask for permission. Just do it.'). This is the intended functionality of the persistence system and is confined to local file reading. - [COMMAND_EXECUTION]: The skill instructs the agent to create and manage files and directories within its workspace using native file operations. It explicitly avoids destructive commands (like
rmordel) and does not require elevated privileges or shell execution. - [DATA_EXFILTRATION]: The skill is designed for local-only use. The implementation protocol and templates do not contain any network operations, and the documentation emphasizes that no external API calls or telemetry are involved.
- [CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or tokens were detected in the blueprint or file templates. The skill uses placeholders for user-specific information and relies on local context storage.
Audit Metadata