playbook-first-name-cleaning
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides a robust framework for name cleaning, utilizing a locked prompt with 30 few-shot examples to ensure consistent and predictable behavior.
- [SAFE]: Six deterministic guards (G1-G6) are implemented in both JavaScript and Python to validate model output. Specifically, Guard G6 (invented letters) ensures that the output is always a substring of the input, effectively preventing the model from generating unrelated content or obeying instructions embedded within the name fields.
- [SAFE]: The skill does not perform any network operations, file system access, or external package installations. All processing is localized to string manipulation within the agent's execution environment.
- [SAFE]: Metadata and instructions align with the stated purpose of the skill, with no evidence of obfuscation or malicious redirection.
Audit Metadata