videodb
Pass
Audited by Gen Agent Trust Hub on Mar 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs the "videodb" and "python-dotenv" packages from standard registries to provide its core video processing functionality.- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute Python code, including running a background listener script ("scripts/ws_listener.py") and various SDK operations for media management.- [DATA_EXFILTRATION]: The skill accesses the local filesystem to manage configuration in ".env" files and session state in "~/.local/state/videodb". While it handles an API key, it includes explicit instructions for the agent to avoid manual handling of the secret.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its ingestion of external video and audio content.
- Ingestion points: External data is pulled from URLs, RTSP streams, and local recordings using "coll.upload()", "coll.connect_rtstream()", and "CaptureClient".
- Boundary markers: No specific delimiters are mandated when processing transcribed text or visual descriptions from these external sources.
- Capability inventory: The skill allows the agent to execute Python code which has the capability to perform network requests and file operations via the authorized "videodb" SDK.
- Sanitization: Transcripts and scene descriptions are used in prompts without evidence of sanitization to filter out potentially malicious instructions embedded in the media content.
Audit Metadata