content-repurposing

Pass

Audited by Gen Agent Trust Hub on Feb 25, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill documents the official installation method for the infsh CLI via a shell script fetched from https://cli.inference.sh. This is a vendor-controlled resource used for environment setup and includes guidance for manual verification.
  • [EXTERNAL_DOWNLOADS]: Uses npx to download and install additional skill dependencies from the vendor's official repository.
  • [COMMAND_EXECUTION]: Executes multiple infsh command-line operations to run AI applications for image generation, text-to-speech, and video creation, including models from trusted providers like Google.
  • [PROMPT_INJECTION]: The skill facilitates the processing of external long-form content (blogs, transcripts) into various formats, which creates a surface for indirect prompt injection.
  • Ingestion points: Data provided to the --input flag in infsh app run commands.
  • Boundary markers: Uses structured JSON formatting for input parameters to help delineate data from instructions.
  • Capability inventory: Accesses remote AI services and social media APIs via the infsh CLI.
  • Sanitization: No explicit sanitization of the source content is mentioned in the markdown recipes.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 25, 2026, 05:36 PM