vigolium-scanner
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The tool features an AI 'autopilot' mode that autonomously generates and executes CLI commands. Documentation specifies a terminal security model including a command allowlist, shell injection prevention via direct execution (no shell), and process isolation.
- [EXTERNAL_DOWNLOADS]: The scanner supports cloning application source code from Git repositories for whitebox analysis and integrates with external AI backends like Anthropic's Claude Code via 'npx'.
- [REMOTE_CODE_EXECUTION]: Users can author and execute custom JavaScript extensions using the embedded Sobek engine. These extensions have access to a utility API (vigolium.utils.exec) intended for performing system-level security testing tasks.
- [DATA_EXFILTRATION]: The skill provides commands for exporting vulnerability findings and captured HTTP traffic to local files or remote servers via ingestion endpoints.
- [PROMPT_INJECTION]: The skill documents an attack surface for indirect prompt injection as it processes untrusted inputs (e.g., target web content, OpenAPI specifications, and application source code) using AI models. This surface is identified as an inherent risk of the tool's primary function as an automated security analyzer.
Audit Metadata