talking-head-guide
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: Instructional Integrity: The content consists entirely of legitimate operational guidelines for video editing workflows, including A-roll cleanup, B-roll placement, and caption management.
- [SAFE]: Platform Tooling: The skill references internal platform tools like
read_script,apply_script, andmulticam_syncfor their intended purposes without any attempt to bypass restrictions or escalate privileges. - [SAFE]: No Malicious Patterns: There is no evidence of prompt injection, data exfiltration, obfuscation, or unauthorized network communication. All workflows are focused on media processing within the designated environment.
- [SAFE]: Indirect Prompt Injection Surface: The skill interacts with user-provided transcripts, but it establishes clear structural boundaries using segment markers (e.g.,
[sN]) and specific editing syntax, which mitigates the risk of the agent misinterpreting data as instructions.
Audit Metadata