muapi-ugc-video-factory (Generative-Media-Skills) — agentic threat model

Not certain from the listing — likely relies on external multimodal/generative video and text models via MuAPI. Primary threats include prompt injection leading to generation of inappropriate/violating UGC content or model reprogramming to bypass safety filters.

Not certain from the listing — the agent processes input assets (scripts, prompts, brand assets) to generate videos. Risks include data exfiltration of proprietary brand assets or injection of malicious metadata into the video generation pipeline.

The agent orchestrates batch workflows to generate videos. Vulnerabilities include insecure tool integration with the MuAPI backend, where manipulated inputs could trigger excessive API calls, leading to denial of wallet (DoW) or financial exhaustion.

The agent writes video files locally or to cloud storage and makes authenticated external API calls. Threats include insecure storage of MuAPI credentials/API keys and potential path traversal if output file-writing paths are not strictly sandboxed.

Not certain from the listing — there is no mention of automated guardrails, content moderation, or output verification before videos are written. Gaps here could allow the generation and distribution of toxic or copyrighted synthetic media.

The agent handles paid API credentials and generates public-facing marketing assets. Lack of explicit access controls or rate-limiting on the batch generation endpoint poses compliance risks regarding unauthorized resource consumption and intellectual property.

The agent acts as a specialized media generation skill within a broader ecosystem. If integrated into an automated pipeline, a compromised upstream agent could abuse this skill to mass-produce spam or disinformation at scale.