mediaio-generate

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (text prompts and media files) which is then used as parameters for generation tasks.
  • Ingestion points: User-provided text prompts and local media files uploaded via the mediaio CLI as described in SKILL.md and references/media-inputs.md.
  • Boundary markers: The skill instructs the agent to treat CLI output as the source of truth and requires explicit user approvals for network and filesystem access.
  • Capability inventory: The agent possesses network access and filesystem write capabilities via the mediaio CLI and standard shell utilities.
  • Sanitization: The skill recommends utilizing structured output validation and schema inspection before submission.
  • [COMMAND_EXECUTION]: The skill executes various mediaio CLI subcommands and standard shell utilities (such as mktemp, grep, file, and curl) to manage generation workflows and local artifacts.
  • [EXTERNAL_DOWNLOADS]: The skill retrieves generated media files from Media.io storage. These downloads are performed using the vendor's CLI or curl fallbacks, targeting authenticated result URLs provided by the service.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 07:54 PM
Security Audit — agent-trust-hub — mediaio-generate