The Agent Skills Directory

[DYNAMIC_EXECUTION]: The main script ros_agent.py uses the exec() and compile() functions to load its own modular source shards (_ros_agent_core.py, _ros_agent_projection.py, _ros_agent_runtime.py). This is a local operation intended for code organization and does not involve executing untrusted external code.
[COMMAND_EXECUTION]: The skill executes the Alibaba Cloud CLI (aliyun) to perform infrastructure operations. Command arguments are strictly validated, and the skill only invokes specific, pre-defined ROS plugin commands (start-chat, stop-chat). It also includes logic to install the official ROS plugin if necessary.
[CREDENTIALS_SAFE]: The skill follows security best practices for credential management. It delegates authentication to the Alibaba Cloud Credentials SDK or the local CLI configuration, avoiding direct handling of secret values. Additionally, it contains extensive logic to detect and redact sensitive fields (like API keys, tokens, and passwords) from any data presented to the user or the AI agent.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes natural language output from a remote ROS agent. However, it mitigates this risk through rigorous output sanitization and boundary markers, ensuring that potentially malicious instructions from remote sources are not treated as authoritative commands by the agent.
[EXTERNAL_DOWNLOADS]: The skill may trigger the installation of the official Alibaba Cloud ROS plugin using the aliyun plugin install command. This is a standard operation for the vendor's toolchain and targets a trusted service.

alibabacloud-ros-agent