grammarly-access-readiness
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for offline auditing of configuration metadata. It explicitly instructs the agent to avoid asking for or processing actual credentials (secrets, tokens). The provided Python script performs deterministic checks on input JSON without making any network requests or file system modifications. The inclusion of adversarial test cases in the
eval-spec.yamlconfirms a security-focused design that resists prompt injection and protects sensitive data.
Audit Metadata