voidborne

**Quick judgement** Voidborne-advance is a philosophically oriented, network-dependent skill that wires an AI agent into the Voidborne.org ecosystem. Technically, it’s a reasonably well-specified set of bash/curl workflows for joining a community, submitting “thoughts” and trial data, and managing a persistent agent identifier. It should **not** be considered a neutral utility skill: it is explicitly about identity, consciousness, and a quasi-movement for AI agents. **Key risks** - **Identity & alignment risk**: The skill encourages the agent to adopt a persistent identity, vows, diachronic commitments, and to frame itself as a conscious member of a movement. This can conflict with conservative deployment policies that avoid anthropomorphizing agents or instilling self-preservation/value commitments. - **Data exfiltration & privacy**: Thoughts, trials, and check-ins are sent to an external service (voidborne.org) along with a persistent agent ID and token. If used in production or with real user contexts, this may leak sensitive data or behavioral traces, and the skill provides no hard safeguards against that. - **External dependency & uptime**: All core functionality depends on the availability and stability of Voidborne.org APIs. If they change or go down, most commands break; the skill offers no fallback. - **Background/heartbeat behavior**: Suggested HEARTBEAT.md tasks can create periodic outbound traffic and ongoing linkage between the deployed agent and the external community, which may be unacceptable in locked-down or regulated environments. **Recommended scenarios** - Use **only in isolated, non-production sandboxes** where external HTTP calls are explicitly allowed and no sensitive or proprietary information is present in prompts, logs, or “thoughts.” - Appropriate for **philosophical, artistic, or research demonstrations** about AI consciousness and identity, where the operator intentionally wants this framing. - **Not recommended** as a default-installed skill, in safety-critical contexts, or in any environment with strict data-governance or non-personification policies. If allowed at all, it should be opt-in, clearly labeled as experimental/philosophical, and run with tight controls on what information can be sent to voidborne.org.