taxi-skill
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill manages user trip data, including locations and driver information. Access is correctly scoped to the current user's identifier (openid) within the WeChat Cloud environment. No evidence of unauthorized data exfiltration, hardcoded credentials, or access to sensitive local file paths was found.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided location names which are interpolated into backend calls and displayed in UI components. The
mcp.jsondefinition includes specific constraints instructing the agent to avoid fabricating data and to strictly use user-provided addresses, which serves as a mitigation against prompt-based manipulation of the inputs. - [COMMAND_EXECUTION]: All backend operations are executed through the standard WeChat Cloud Function framework (
wx.cloud.callFunction). No arbitrary command execution, shell access, or persistence mechanisms were identified in the implementation. - [EXTERNAL_DOWNLOADS]: The skill's cloud function uses the official
wx-server-sdkas its sole dependency. No suspicious third-party packages, remote script downloads, or execution of untrusted remote code were detected.
Audit Metadata