Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward iFlytek image-analysis helper that sends a user-selected image and question to a fixed iFlytek WebSocket API using the user's own API credentials.
Install only if you are comfortable sending selected image contents and prompts to iFlytek using your own iFlytek credentials. Avoid using it on sensitive documents, faces, receipts, screenshots, or proprietary images unless that remote processing is acceptable under your policies.
The entire skill description and usage guidance are presented only in Chinese, which can constitute a language/locale policy issue when no user opt-in or alternative language is offered. Nothing in the README indicates that the skill is region-specific or intentionally limited to Chinese-speaking users.
The skill advertises code-capable behavior that uses environment variables and outbound network access, but it does not declare an explicit tool or permission scope. This creates a transparency and governance gap: the agent may invoke a third-party API with secrets and user-supplied image content without clear permission boundaries or reviewable scope.
The skill description is broad enough to match many ordinary image-related requests, which increases the chance the agent routes user content to this skill by default. In this context, overbroad routing is risky because invocation can cause images and prompts to be sent to an external service and consume credentials without clear user intent.
The example invocation phrases are generic everyday requests like 'analyze this image' and 'what is in this picture,' with no constraints or negative examples. That makes accidental or overbroad activation more likely, which is especially sensitive here because user images may contain private data and are transmitted off-platform.
The documentation explains how to use the service but does not clearly warn users that images and prompts are transmitted to a third-party provider. This is a material privacy and data-handling omission because users may unknowingly submit sensitive images, OCR content, or personal information to an external API.
The script base64-encodes a local image and sends it to a third-party remote API for processing, but it does not provide an explicit privacy warning, consent prompt, or clear disclosure at the point of transmission. This can cause unintended exfiltration of sensitive image contents such as personal documents, faces, receipts, or internal screenshots, especially when invoked as an agent skill where users may assume local-only analysis.
The file switches into Chinese for the error-code and FAQ sections, which imposes a language expectation on users who may only read English. There is no opt-in, alternate language version, or note that the skill is intended only for Chinese-speaking users, so this may conflict with language-choice policy expectations.
The default question text is hard-coded in Chinese, and several usage examples are also Chinese-only. This imposes a specific language behavior without explicit user opt-in or a documented locale constraint, which can violate language/locale policy expectations.
Detected: suspicious.exposed_secret_literal
