Most static findings are false positives caused by Markdown code fences and inline formatting. Six findings contain intended belt or npx commands, while semantic review identified unpinned installations and third-party processing of prompts or images. No prompt injection or malicious exfiltration intent was found.
The documented npx commands install community skills without immutable versions or commit references, creating dependency and supply-chain exposure.
The commands visibly name remote skill sources without version pins or commit hashes. Their future contents can change independently of this audited file.
The belt app run examples submit storyboard prompts and image references to external inference applications, which may expose confidential production material.
The skill identifies inference.sh as the image-generation service and uses belt app run with prompt and image fields. External processing is central to these examples.
Capability review items (6)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
The Bash block runs belt login and belt app run commands against external applications. These commands have authentication and network side effects when followed.
Most static findings are false positives caused by Markdown code fences and inline formatting. Six findings contain intended belt or npx commands, while semantic review identified unpinned installations and third-party processing of prompts or images. No prompt injection or malicious exfiltration intent was found.
The documented npx commands install community skills without immutable versions or commit references, creating dependency and supply-chain exposure.
The commands visibly name remote skill sources without version pins or commit hashes. Their future contents can change independently of this audited file.
The belt app run examples submit storyboard prompts and image references to external inference applications, which may expose confidential production material.
The skill identifies inference.sh as the image-generation service and uses belt app run with prompt and image fields. External processing is central to these examples.
Capability review items (6)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
The Bash block runs belt login and belt app run commands against external applications. These commands have authentication and network side effects when followed.