cold-open-creator
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, data exfiltration attempts, or unauthorized command executions were found in the skill instructions or reference files.
- [PROMPT_INJECTION]: The skill contains no instructions designed to bypass safety filters or extract system prompts. It uses standard instructional language to define its role as a scene selector and editor.
- [DATA_EXFILTRATION]: There are no network calls (e.g., curl, wget) or attempts to access sensitive system files like SSH keys or environment variables.
- [REMOTE_CODE_EXECUTION]: No external dependencies or scripts are downloaded. The skill references internal related skills which do not involve remote code loading.
- [PROMPT_INJECTION]: Regarding Indirect Prompt Injection (Category 8), the skill processes external transcript data. However, the 'Fundamental Constraint' in SKILL.md explicitly limits the agent to 'CUTTING' and 'REARRANGING' verbatim text only, prohibiting the creation of new dialogue or paraphrasing. This architectural constraint prevents the agent from executing instructions that might be embedded in the processed data.
Audit Metadata