fix-my-look
Pass
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via the
<change_prompt>input which is used to influence the behavior of generative AI tools. - Ingestion points: The skill accepts untrusted user input via the
<change_prompt>and<source>(URL/path) parameters. - Boundary markers: The
<change_prompt>is directly concatenated into prompt templates in Step 2 (gpt-image-2) and Step 4 (kling) without robust delimiters or instructions to ignore potential commands embedded in the user text. - Capability inventory: The skill utilizes several capabilities that could be misused if a prompt injection is successful, including
upload_asset,generate_image_edit,generate_reference_video, andedit_audio_replace. - Sanitization: No sanitization, escaping, or validation of the
<change_prompt>is performed before it is interpolated into the instructions sent to the image and video generation providers.
Audit Metadata