prompt-engineering
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides a framework for generating video prompts based on external inputs like 'Script Briefs' and 'Character Anchors'. This creates an attack surface where instructions embedded in these external sources could influence the agent's prompt generation logic.
- Ingestion points: The skill explicitly mentions using '剧本 Brief' (Script Brief) and '角色锚点' (Character Anchors) as primary inputs in SKILL.md.
- Boundary markers: The prompt structure utilizes specific headers (e.g., SCENE CONTEXT, AUDIO) which provide basic structural separation but do not strictly sanitize or isolate the content from instructions.
- Capability inventory: The skill lacks execution capabilities such as file system writing, network requests, or code execution. Its primary function is text generation.
- Sanitization: No specific sanitization, filtering, or escaping logic is provided for the ingested data before interpolation.
- [NO_CODE]: The skill consists entirely of Markdown documentation and text templates. It does not include any scripts, configuration files for package managers, or binary executables.
Audit Metadata