movie-maker-skill
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (HIGH): An automated scanner (URLite) flagged a blacklisted URL in
rules/start/requirements.md. Since the skill's instructions mandate reading this file during the 'Intake' phase, the agent is directed to process potentially malicious external content. - [PROMPT_INJECTION] (HIGH): The skill is highly vulnerable to Indirect Prompt Injection. It ingests untrusted user data—specifically character descriptions, styles, and reference images—and interpolates them into manifests and prompts for downstream AI models (fal, ElevenLabs).
- Ingestion points: Intake questions (1-9) in
SKILL.md. - Boundary markers: Absent; no delimiters or instructions to ignore embedded commands are present in the provided context.
- Capability inventory: Command execution via
ffmpegandremotion, and network API calls to external providers. - Sanitization: None; the skill lacks validation for user-supplied 'vibe' or 'character' strings.
- [COMMAND_EXECUTION] (MEDIUM): The skill documentation describes a local rendering pipeline using
ffmpegandremotion. This implies the agent will generate and execute shell commands or scripts based on a manifest JSON, which could be manipulated if the manifest is poisoned by malicious user input. - [REMOTE_CODE_EXECUTION] (MEDIUM): The skill references 'Remotion', which typically involves running a Node.js-based rendering engine. If the manifest generation is compromised, this could lead to arbitrary code execution within the rendering environment.
Recommendations
- AI detected serious security threats
- Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata