The Agent Skills Directory

[CREDENTIALS_UNSAFE] (SAFE): The skill correctly retrieves the AMap API key from the AMAP_MAPS_API_KEY environment variable rather than hardcoding secrets.
[DATA_EXFILTRATION] (SAFE): Network activity is restricted to the trusted domain restapi.amap.com for geographic, weather, and routing data.
[COMMAND_EXECUTION] (SAFE): The application logic uses a strictly defined set of commands and arguments. It does not utilize unsafe functions like eval() or exec() for processing user input.
[REMOTE_CODE_EXECUTION] (SAFE): No mechanisms for downloading or executing remote code or scripts were found in the skill's source files.
[PROMPT_INJECTION] (SAFE): The instruction files provide clear, safe guidelines for the agent and do not contain attempts to bypass safety filters or override system prompts.
[DATA_EXPOSURE] (LOW): The skill exposes an indirect prompt injection surface by returning raw JSON data from an external API (AMap).
Ingestion points: AMap API responses (e.g., POI descriptions, address names) in scripts/lib/commands.ts.
Boundary markers: Absent; the skill prints raw JSON to stdout.
Capability inventory: Network access to restapi.amap.com.
Sanitization: None; the agent receives raw data from the external service.

amap