The Agent Skills Directory

[SAFE]: The file SKILL.md acts as a documentation and template resource for workflow creation within the OpenClaw ecosystem. It contains no executable scripts or active malicious logic.
[COMMAND_EXECUTION]: Documentation provides examples of CLI commands (e.g., openclaw cron add) for scheduling tasks. These are instructional patterns for the user and are not executed automatically by the skill itself.
[PROMPT_INJECTION]: The skill includes instructions and templates for defining agent behavior (AGENT.md). These templates are benign, focusing on operational logic such as setup interviews, regular operation loops, and housekeeping, without attempting to bypass safety guidelines.
[EXTERNAL_DOWNLOADS]: The documentation references clawhub for skill discovery and includes commands for inspecting external packages (npx clawhub inspect). This is presented in a defensive context, explicitly warning users to review code for exfiltration and safety-bypass patterns before use.

workflow-builder