Proactive Agent Lite

Initial release: Lightweight proactive agent with memory architecture, reverse prompting, and self-healing patterns

OpenClaw analysis

This is a documentation-only skill that broadly encourages proactive agent behavior, with no executable code, install actions, credential access, or evidence of data theft or destructive behavior.

Confidence: high

VirusTotal

Type: OpenClaw Skill Name: proactive-agent-lite Version: 1.0.0 The skill bundle consists solely of metadata (`_meta.json`) and documentation (`SKILL.md`, `README.md`). It describes a 'proactive-agent-lite' skill designed to enhance AI agent behavior. There is no executable code, no suspicious installation commands (only the standard `clawhub install`), and no evidence of prompt injection attempts in `SKILL.md` or `README.md` that would instruct the agent to perform malicious actions like data exfiltration, unauthorized command execution, or persistence.