The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains logic in recipes/advanced_permissions.md that instructs the agent to automatically switch its target API base URL if a specific permission error (status 240) is received, representing a behavioral override triggered by external service data.
[DATA_EXFILTRATION]: Network requests are made to subdomains of baidu.com and bcebos.com to transmit user-supplied search keywords and location coordinates. While these are official vendor services, they constitute the transmission of user-provided context to external endpoints.
[EXTERNAL_DOWNLOADS]: The skill downloads CSV and Excel data files for weather district identification from Baidu's cloud storage (bcebos.com). These are recognized vendor resources and do not represent a security risk.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the ingestion of untrusted user data in API parameters. Ingestion points: Search queries and address strings provided by users in various map recipes. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the documentation. Capability inventory: The skill utilizes HTTP GET requests across multiple files for core mapping functionalities. Sanitization: No input validation or filtering protocols are documented for the processed location strings.

baidu-map-webapi