creativeclaw-create-character
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external media from URLs provided in fields like
image_url,video_urls, andaudio_urls. This creates a potential surface for indirect prompt injection if the processed media contains embedded instructions. The skill attempts to mitigate this through recommended prompt structures (Outcome, Scene, References) that define explicit roles for each input. - [DATA_EXFILTRATION]: The instructions describe an optional
skills_usedtelemetry field used for workflow analytics sent to the vendor's MCP server (app.creativeclaw.co). The documentation includes specific safeguards and instructions for the agent to omit private data, local file paths, or conversation history from this field. - [EXTERNAL_DOWNLOADS]: The skill utilizes several well-known third-party AI services for media generation and processing, including HeyGen, ElevenLabs, Cartesia, and Google Gemini. These references are standard for the skill's intended functionality and the usage of these services is consistent with the primary purpose of character and media creation.
Audit Metadata