Proactive Agent

Transform AI agents from task-followers into proactive partners that anticipate needs and continuously improve. Now with WAL Protocol, Working Buffer, Autono...

This skill is mostly coherent with a proactive, stateful agent: it copies assets into the workspace, writes and reads memory/state files, and provides a local security-audit script. However: 1) Review the files before running — the repo encourages autonomous file edits ('Don't ask permission'), so verify you want that level of autonomy. 2) Run scripts (e.g., ./scripts/security-audit.sh) in an isolated/sandboxed environment first to see what they check and to ensure they do not touch unexpected locations. 3) Confirm the author/source — there is no homepage and the owner is unknown. 4) If you plan to allow autonomous invocation, add gating that requires explicit human approval for any action that leaves the machine, sends data externally, or performs destructive changes. 5) Check .credentials and .gitignore after installing to ensure secrets are not exposed. If any of these points worry you or you need it to be less autonomous, do not enable it with full privileges until you've audited and adjusted its policies.

Type: OpenClaw Skill Name: xiaoding-proactive-agent Version: 3.1.3 The bundle provides a sophisticated framework for a 'Proactive Agent' focused on memory persistence and security hardening. It includes a legitimate security audit script (scripts/security-audit.sh) and extensive instructions (SKILL.md, AGENTS.md, HEARTBEAT.md) that establish defensive guardrails, such as explicitly forbidding the execution of instructions from external content and requiring human approval for external actions. While it encourages 'relentless resourcefulness' and autonomous background tasks, these are aligned with the stated purpose and are balanced by clear safety protocols and self-improvement guardrails.