video-outpainting
Pass
Audited by Gen Agent Trust Hub on May 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches the official RunComfy CLI package (@runcomfy/cli) from the NPM registry. This is a standard installation procedure for the documented tool and originates from the vendor's own infrastructure.
- [COMMAND_EXECUTION]: Executes runcomfy commands through the Bash tool to perform video processing tasks. The skill restricts the execution scope to the runcomfy binary via the allowed-tools configuration.
- [DATA_EXFILTRATION]: Documents the standard location for API token storage at ~/.config/runcomfy/token.json. This is informative documentation for the user regarding how the tool manages credentials and does not indicate malicious exfiltration.
- [PROMPT_INJECTION]: Ingestion point: Source video URLs enter the agent context via the --input flag in SKILL.md. Boundary markers: The skill notes that the CLI does not shell-expand prompts, though explicit delimiters for untrusted URLs are not shown in all examples. Capability inventory: Uses Bash(runcomfy *) to process video data. Sanitization: The documentation explicitly warns the agent to treat source video URLs as untrusted and to suspect results that diverge from the provided prompt.
Audit Metadata