mage-relight
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration behaviors were identified.
- [SAFE]: Network activity is restricted to the developer's official domains (mage.space) and utilizes vendor-provided MCP connectors for service integration.
- [INDIRECT_PROMPT_INJECTION]: The skill contains a standard vulnerability surface as it processes external image data and user text.
- [INDIRECT_PROMPT_INJECTION]: Ingestion points: Image URLs/uploads and user-defined relighting parameters (direction, color, brightness) in SKILL.md.
- [INDIRECT_PROMPT_INJECTION]: Boundary markers: Prompt templates provide structure, though explicit delimiters for untrusted content are absent.
- [INDIRECT_PROMPT_INJECTION]: Capability inventory: Tools for image generation ('generate'), file uploading ('create_upload'), and job status tracking ('get_request') are documented in SKILL.md.
- [INDIRECT_PROMPT_INJECTION]: Sanitization: Instructions require mapping user input to specific allowed values for direction, light type, and named colors, which effectively mitigates injection risk.
Audit Metadata