Fluxwing Screenshot Importer
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes external data (UI screenshots) and user-defined file paths that are directly interpolated into subagent prompts.
- Ingestion points: The
screenshotPathvariable and the visual content of the screenshot file itself are passed to subagents for analysis in Phase 2 and Phase 4. - Boundary markers: Absent. The skill does not use delimiters or specific instructions to isolate the user-provided screenshot path or potential text content within the images from the subagent instructions.
- Capability inventory: The skill and its subagents have access to
Read,Write, andTasktools, allowing them to perform file system operations and spawn further subagents based on potentially adversarial content. - Sanitization: The skill validates basic file properties (existence, extension) but does not perform sanitization of the path string or the visual data for prompt injection vulnerabilities.
Audit Metadata