Security audit
Security checks for vulnerabilities and agentic risk
This skill coherently helps generate travel-photo collages from user-provided photos and destinations, with no evidence of hidden code, persistence, credential use, or destructive behavior.
Before installing, be aware that the skill may analyze uploaded personal photos to preserve appearance across generated images and may search the web for destination landmarks. The publisher should declare web_search in the metadata and make follow-up questions match the user's language.
The trigger condition is broad enough to activate on common combinations of photo uploads and destination mentions, which can cause the skill to run in situations where the user did not explicitly request this specific functionality. Over-broad triggering increases the chance of inappropriate processing of user photos and unintended tool usage.
The skill hardcodes a Chinese-language prompt when asking for a missing destination, regardless of the user's language. While not a direct security exploit, this can override user preference, create confusion, and contribute to unsafe automation behavior by reducing informed consent and clarity around what the skill is doing.
The manifest declares only the image-generation tool, but the workflow explicitly requires web_search to research landmarks. This mismatch can cause the agent to act outside the declared capability boundary, undermining least-privilege assumptions and making review, policy enforcement, and user expectations inaccurate.
The skill directly instructs the agent to perform web searches without that capability being justified in the manifest context. This expands data access and external interaction beyond what reviewers and operators may expect, increasing the risk of unintended browsing, policy bypass, or exfiltration through search queries.
The instruction hard-codes a Japanese search phrase ("必去景点") as part of the recommended query without offering a language choice or explaining why that locale-specific phrasing is required. This can violate language/locale policy because it imposes a specific language in natural-language guidance rather than letting the user choose.
No suspicious patterns detected.
