seedance-prompt-generator
Warn
Audited by Gen Agent Trust Hub on Jul 18, 2026
Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The 'filter-aware prompting' section provides explicit instructions on how to circumvent safety filters of external LLM-based systems. It suggests using 'cinematic language' to increase the 'pass rate' and provides techniques to disguise sensitive elements to avoid rejection by safety layers.- [PROMPT_INJECTION]: The skill is designed to ingest untrusted data from repository files (e.g., README.md, package.json, source code) to generate marketing scripts, creating a surface for indirect prompt injection. * Ingestion points: Files such as
README.md,package.json,pyproject.toml,Cargo.toml, and other project metadata or documentation. * Boundary markers: Absent. The skill does not instruct the agent to use delimiters or ignore instructions embedded within the repository files. * Capability inventory: Repository analysis and structured text generation. * Sanitization: No evidence of input validation or content sanitization is provided before the repository data is processed.
Audit Metadata