afa-email
Pass
Audited by Gen Agent Trust Hub on May 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data handling were found within the skill files. The skill operates within the defined brand marketing scope.
- [PROMPT_INJECTION]: The skill instructions do not contain attempts to override safety guidelines or bypass constraints. It uses a structured context matrix and preamble protocols to maintain operational scope and state management.
- [DATA_EXFILTRATION]: There are no network calls (curl, wget, etc.) or access to sensitive local files such as SSH keys, environment variables, or AWS credentials. Data processing is restricted to local brand context files like products.md and learnings.jsonl.
- [REMOTE_CODE_EXECUTION]: No external script downloads, package installations, or dynamic code execution patterns (eval, exec) were identified.
- [COMMAND_EXECUTION]: The skill does not invoke shell commands or interact with the operating system. All logic is represented as natural language instructions for the AI agent.
Audit Metadata