skills/acedatacloud/skills/didi-ride/Gen Agent Trust Hub

didi-ride

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill facilitates ride-hailing and route planning through the official DiDi MCP server endpoint (mcp.didichuxing.com).\n- [SAFE]: Sensitive authentication keys (DIDI_MCP_KEY) are managed via platform-injected environment variables. The supporting script includes specific handlers to prevent the URL-encoded key from appearing in error traces or logs.\n- [INDIRECT_PROMPT_INJECTION]: The skill safely handles data processed from external address searches and order status tools by enforcing a strict confirmation flow.\n
  • Ingestion points: Data returned by maps_textsearch and taxi_query_order tools in scripts/didi.py.\n
  • Boundary markers: A mandatory --confirm flag must be appended to shell commands after explicit user approval is received.\n
  • Capability inventory: Interaction with ride-hailing APIs and local script execution.\n
  • Sanitization: All state-changing tools (booking and cancellation) operate in dry-run mode by default to prevent unauthorized execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 05:41 PM
Security Audit — agent-trust-hub — didi-ride