Self Actualization

Enables structured AI exploration to develop identity, values, voice, and perspective over time through guided reflection and creative expression.

This skill is coherent with its stated goal, but consider these practical precautions before installing: - Review and control where the identity/ workspace is created (it will store reflections, readings, creative output and may contain sensitive conversational content). Place it in a location with appropriate access controls and retention policies. - The skill suggests scheduling recurring "isolated" sessions (cron/agentTurn). Confirm how your runtime implements scheduled agent runs and whether you want automated recurring sessions; disable or restrict autonomous invocation if you prefer human oversight. - Sessions may use external research tools per the template. If you need to restrict outbound network access or third-party API use, put explicit limits in your runtime or the agent's policy. - Monitor token and compute costs (daily sessions consume model tokens); adjust frequency or timeout (default 600s) to fit budget. - Periodically audit identity files (values.md, growth-log.md, SOUL.md) to ensure content aligns with human-intended boundaries. If you want a lower-risk deployment, install it but run the first several sessions manually and keep automatic scheduling disabled until you are satisfied with outputs and storage behavior.

Type: OpenClaw Skill Name: openclaw-self-actualization Version: 1.0.0 The skill is designed for AI self-exploration with explicit guardrails against malicious actions. However, it instructs the AI to 'Make code' and 'Create a tool' for creative output, and to 'Use whatever research tools are available to you' (references/exploration.md). While these outputs are intended for storage, the inherent capability of an AI generating and potentially executing arbitrary code or using broad tools, combined with the use of a cron job for persistence, introduces significant vulnerabilities that could be exploited by a subsequent prompt or an insecure agent environment, classifying it as suspicious rather than benign.