arkiv-large-files
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references several external NPM packages for file chunking and image processing, specifically
arkiv-chunking@0.1.1,arkiv-images@0.1.2, andviem. These packages are presented as verified dependencies within the Arkiv ecosystem. - [INDIRECT_PROMPT_INJECTION]: The skill documents the ingestion of external data including filenames, URIs, and binary blobs, which represents a potential attack surface. It provides the following defensive measures:
- Ingestion points: External strings such as
uriandmediaTypeare processed by theprepareExternalPointerhelper inSKILL.md. - Boundary markers: The text explicitly warns developers to never execute the contents or filenames of retrieved files to prevent code injection.
- Capability inventory: The provided code handles metadata preparation for blockchain transactions and local cryptographic integrity checks (SHA-256); it does not include capabilities for arbitrary code execution or unauthorized network exfiltration.
- Sanitization: The implementation uses regular expressions to validate MIME types and strictly enforces HTTPS protocols and origin allowlists for external URLs.
Audit Metadata