alibabacloud-media-diagnostics
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes external media analysis tools (ffprobe and mediainfo) via the subprocess.run function in scripts/lib/cmd.py. It uses list-based arguments, which is a secure practice that prevents shell command injection.
- [EXTERNAL_DOWNLOADS]: The scripts/hls_check.py script performs network requests (HTTP GET and HEAD) to download HLS playlists and sample media segments from URLs explicitly provided by the user. This is necessary for the skill's primary function of diagnosing remote streams.
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external data (media files and playlists) and returns the extracted metadata to the agent, creating a vulnerability surface for indirect prompt injection. 1. Ingestion points: Technical metadata is extracted from media files via ffprobe and HLS playlist content is parsed in scripts/hls_check.py. 2. Boundary markers: The scripts return structured JSON data to the agent to distinguish data from instructions. 3. Capability inventory: The skill is capable of executing specific shell commands (ffprobe, mediainfo) and making network requests to user-specified URLs. 4. Sanitization: The tool extracts specific technical fields rather than raw text, though malformed metadata could still be processed.
Audit Metadata