jianying-draft-smart-package

Pass

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documentation mentions an 'external_native_execution' risk and describes a process where high-level proposals are compiled into Rust-based atomic execution steps. This is a functional requirement of the video editing integration rather than an exploit vector.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data such as video materials, style descriptions, and target channel requirements which could theoretically contain hidden instructions.
  • Ingestion points: Materials list, style selection, and channel context provided in Step 1 of the workflow.
  • Boundary markers: Uses stable digest strings (SHA-like hashes) to tie requests, proposals, and approvals together, ensuring the agent only acts on verified data.
  • Capability inventory: Includes capabilities for isolated project editing (project.edit_isolated), rendering (render.proxy), and native execution.
  • Sanitization: Implements strict verification of protection rules (dialogue, brand, locked shots), resource rights, and budget before any approval is granted. Final validation includes manual narrative review and draft difference checks.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 26, 2026, 07:55 PM