The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill guides the user to download its core components from a public GitHub repository using git clone https://github.com/O-ozzz/OctoGuard--Free-OpenClaw-Security-Supervision-System.git. This is a standard setup procedure for the management system described.
[PROMPT_INJECTION]: The skill architecture includes a gateway that processes external data, creating an indirect prompt injection attack surface.
Ingestion points: The interceptRequest middleware in SKILL.md ingests user_input and tool_call data from the request body for security evaluation.
Boundary markers: The implementation lacks explicit delimiters or instructions to the AI to ignore embedded commands within the evaluated strings.
Capability inventory: The system can execute network requests (axios), send alerts to external webhooks (DingTalk, WeChat), and interact with system processes via child_process.spawn.
Sanitization: While the system uses regex-based matching to block known patterns, it does not perform comprehensive sanitization of the input before it is logged or passed to alert handlers.
[COMMAND_EXECUTION]: The GatewayController module in SKILL.md references the use of child_process.spawn for controlling the OpenClaw service status, which is a functional requirement for the system's management capabilities.

octoguard-openclaw-security-supervision