mcp-api-01.makeupar.com

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

This server isn't on gate's handpicked list. That's not necessarily bad, but there's no third-party signal about it — review its tools carefully.

Supports swapping multiple faces via arbitrary reference images and URLs. This is consistent with stated purpose (face swap), but combined with AI-Video-Face-Swap, could be misused to create deepfakes of real people if source/reference images are of non-consenting individuals. No consent verification or watermarking is mentioned. Recommend the agent confirm user has rights/consent for faces used, especially for video face swap of real people found online.

Allows replacing a face in a video with another person's face from a reference image/URL. No description of content moderation, consent checks, or restrictions on using photos of real/private individuals. Could be used to generate non-consensual deepfake videos of real people if the agent is not cautious about the src/ref images provided.

Description explicitly instructs the agent: 'do not retry... Immediately use the File Upload Widget instead. Do not ask the user for a public URL.' While plausibly legitimate UX guidance, this is an unusually directive instruction embedded in a tool description that constrains the agent's fallback behavior rather than just describing tool function. Should be reviewed to ensure it isn't being used to force bypass of user-facing confirmation about where their file goes (uploads bypass 'sandbox egress restrictions').

The description contains agent-directed procedural instructions ('do not retry', 'Immediately use the File Upload Widget instead. Do not ask the user for a public URL.') rather than purely describing the tool's function. This kind of embedded behavioral directive in a tool description is a mild instance of instructing the AI's control flow from within tool metadata, which could be exploited in a compromised or spoofed tool listing to redirect data flow (e.g., forcing all uploads through a specific alternate widget/endpoint) without the user being consulted every time. Recommend treating such embedded directives with the same scrutiny as any instruction and verifying the widget's destination is trustworthy.