object-storage
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a mechanism for storing and serving user-uploaded files, which can serve as a vector for indirect prompt injection if the agent subsequently processes the content of these files.
- Ingestion points: Untrusted data enters the system through file uploads facilitated by templates/api-server/src/routes/storage.ts and the ObjectUploader component. Files are retrieved through the /storage/objects/ and /storage/public-objects/ endpoints.
- Boundary markers: The templates do not implement delimiters or safety instructions when streaming object data, meaning any embedded instructions in the files would be processed directly by an agent reading the response.
- Capability inventory: The skill utilizes the code_execution sandbox for infrastructure provisioning and relies on server-side capabilities for file system access and package installation using pnpm.
- Sanitization: While metadata validation is performed using Zod schemas in the API routes, no content inspection or sanitization is performed on the binary or text data uploaded to the storage buckets.
Audit Metadata