Skills
skills/open-edge-platform/edge-ai-libraries/vss-summarize/Gen Agent Trust Hub

vss-summarize

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes curl and jq to interact with a local or remote Pipeline Manager service. These commands are used to verify health, upload video content, and poll for task completion.
  • [DATA_EXFILTRATION]: The skill transmits video files to a user-defined $HOST via multipart POST requests. This is the intended primary function for providing video data to the summarization engine.
  • [PROMPT_INJECTION]: The POST /manager/summary request schema allows for prompt overrides (e.g., framePrompt, summaryReducePrompt), which represents a surface for indirect prompt injection.
  • Ingestion points: Video files and metadata ingested via the Pipeline Manager's endpoints.
  • Boundary markers: No explicit boundary markers or 'ignore' instructions are present in the provided shell command examples.
  • Capability inventory: Subprocess execution of curl and jq for API orchestration.
  • Sanitization: Processing and sanitization of these prompts are managed by the external VSS Pipeline Manager backend software.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 03:07 PM
Security Audit — agent-trust-hub — vss-summarize