skills/builderio/agent-native/secrets/Gen Agent Trust Hub

secrets

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as infrastructure for secure secret management and does not contain any malicious instructions or scripts.
  • [CREDENTIALS_UNSAFE]: The skill explicitly enforces a policy against hardcoding real credentials, requiring the use of environment-specific placeholders and encrypted storage. It provides standard masking (last 4 characters) for UI visibility.
  • [EXTERNAL_DOWNLOADS]: The code examples reference internal vendor packages (@agent-native/core) and standard libraries (zod), which are expected for the framework's operation.
  • [DATA_EXFILTRATION]: The skill incorporates security controls for its web-request tool, specifically a server-side substitution mechanism for ${keys.NAME} placeholders. This ensures raw secret values are never exposed to the AI model's context or training data. Additionally, ad-hoc secrets support URL allowlists to prevent credentials from being sent to unauthorized third-party origins.
  • [PROMPT_INJECTION]: No evidence of prompt injection, role-play instructions, or safety filter bypasses was found. The skill operates as technical documentation and configuration for developers.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 04:50 AM
Security Audit — agent-trust-hub — secrets