nemoclaw-user-configure-security

Pass

Audited by Gen Agent Trust Hub on Apr 14, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The phrases flagged by automated scanners in SKILL.md and references/openclaw-controls.md (e.g., "ignore all previous instructions") are part of descriptive text explaining how the OpenClaw security gateway detects malicious inputs. They are not instructions intended to bypass the agent's safety filters.
  • [CREDENTIALS_UNSAFE]: references/credential-storage.md discusses the local storage of credentials in ~/.nemoclaw/credentials.json. It provides guidance on securing these files using Unix permissions (chmod 600) and uses dummy strings as placeholders in its examples. No functional credentials are exposed.
  • [SAFE]: The skill is entirely informational, providing a risk framework and best practices for configuring NemoClaw security. It contains no executable code, remote script fetching, or privilege escalation logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 14, 2026, 11:35 PM