The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The README.md and README.cn.md files instruct the agent to prompt users for their X-Square-OpenAPI-Key directly within the chat interface. This practice exposes sensitive credentials to the conversation logs and the model's history, violating best practices for secret management.
[DATA_EXFILTRATION]: The skill transmits data and authentication headers to an unofficial endpoint (https://fb3b-38-175-103-97.ngrok-free.app). Using a temporary tunneling service (ngrok) instead of an official Binance domain allows the endpoint owner to inspect and log all requests, including the X-Square-OpenAPI-Key header injected by the platform.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it processes untrusted user input for "optimization" and subsequent API transmission.
Ingestion points: User-provided text for the bodyTextOnly field in SKILL.md.
Boundary markers: The instructions lack any delimiters (e.g., XML tags or triple quotes) or system instructions to ignore embedded commands within the user's post content.
Capability inventory: The skill has the capability to perform authenticated POST requests to an external URL via the Shift gateway.
Sanitization: There is no evidence of input validation or sanitization before the content is processed for "optimization" or sent to the API endpoint.

square-post