uploads-cli

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill references the installation of the '@buildinternet/uploads' package from the npm registry, which is a tool provided by the vendor.
  • [DYNAMIC_EXECUTION]: The screenshot functionality includes support for '--eval' and '--init-script' flags, which allow the execution of JavaScript in the browser context to handle dynamic web content.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill facilitates uploading local files to the 'uploads.sh' service. The documentation explicitly warns that these uploads are public and permanent, advising against the inclusion of sensitive data.
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes content from external URLs during screenshot and fetch operations, and mirrors GitHub PR media.
  • Ingestion points: External URLs fetched via 'uploads put --url' and captured via 'uploads screenshot', and GitHub PR/issue data processed by 'uploads ingest'.
  • Boundary markers: No explicit boundary markers or instruction-ignoring warnings are present for processed content.
  • Capability inventory: Execution of the 'uploads' CLI tool, browser-based screenshotting with JS execution capabilities, and network uploads to the vendor's API.
  • Sanitization: The skill mentions sanitization of branch names but does not specify sanitization for fetched URL content or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 06:54 PM
Security Audit — agent-trust-hub — uploads-cli