The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's runtime image guides provide templates for downloading essential development tools from trusted and well-known sources.
Fetches the Miniforge installation script from the official Conda-forge GitHub repository.
Downloads the CUDA toolkit and repository keys from NVIDIA's official developer domain.
Configures package mirrors for Aliyun and Tsinghua University to accelerate downloads in the China region.
[SAFE]: The skill implements and enforces several security-positive practices.
[CREDENTIALS_UNSAFE]: Instructions in SKILL.md explicitly mandate the use of dotenv and os.getenv for credential management and prohibit the agent from reading .env files directly with file-read tools.
[DATA_EXFILTRATION]: No instances of unauthorized data transmission were detected. Data handling examples are confined to the intended MaxCompute/OSS ecosystem.
[COMMAND_EXECUTION]: The provided Python script scripts/lookup_operator.py is a safe utility for querying local API documentation and does not perform any high-risk system operations.
[PROMPT_INJECTION]: The skill instructions use clear, professional language to define agent behavior without attempting to bypass safety protocols or override fundamental system constraints.

alibabacloud-odps-maxframe-coding