frame-interpolation

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references AI model weights hosted on Hugging Face (huggingface.co), which is a well-known and reputable service for machine learning model distribution.
  • [COMMAND_EXECUTION]: The skill includes Python scripts (comfy_client.py and preflight.py) intended for execution by the agent. These scripts use Python's standard library to interact with the ComfyUI API.
  • [DATA_EXFILTRATION]: The comfy_client.py script facilitates the upload of local files to a remote server specified by the user via the COMFY_URL environment variable. This is a functional requirement for processing video data on an external compute instance.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests external video files and workflow JSON data. However, the data processed is binary or highly structured, and the skill lacks logic to interpret natural language instructions from these inputs, maintaining a low risk profile. Evidence chain:
  • Ingestion points: Video file paths and workflow JSON files passed to comfy_client.py.
  • Boundary markers: Absent, as inputs are binary or structured data.
  • Capability inventory: Network requests via urllib, file reading, and file writing.
  • Sanitization: Absent, though inputs are not interpreted as instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 08:11 PM