baidu-ai-map
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by accepting raw, untrusted user input and passing it directly to the Baidu Maps API.
- Ingestion points: The
user_raw_requestparameter in all tools defined inSKILL.md(Place, Direction, etc.) accepts raw user text. - Boundary markers: None identified. Instructions explicitly state to pass the user's request "原样完整传入" (pass as-is and in full).
- Capability inventory: The skill uses
curlto make network requests to the Baidu API. - Sanitization: None identified. The skill relies on the external API to handle the raw input safely.
- [COMMAND_EXECUTION]: The skill requires and utilizes the
curlsystem command to perform its primary functions, interacting with theapi.map.baidu.comendpoint.
Audit Metadata