video-analyser

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches video files from external sources including user-provided direct URLs and Linear ticket attachments using curl to download them to a temporary workspace.- [COMMAND_EXECUTION]: It relies on system-level command execution to perform its primary functions, specifically calling ffmpeg and ffprobe for media processing, and tesseract and whisper for data extraction.- [PROMPT_INJECTION]: The skill is subject to indirect prompt injection via the media content it processes. 1. Ingestion points: Media is ingested from external URLs and Linear ticket data. 2. Boundary markers: Extracted text and transcripts are wrapped in <ocr_frame_N> and <audio_transcript> tags. 3. Capability inventory: The skill has the ability to run shell commands and make network requests. 4. Sanitization: No explicit sanitization or filtering is applied to the text extracted by OCR or audio transcription before it is passed to the LLM.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 11:47 AM
Security Audit — agent-trust-hub — video-analyser