Security

AgentPaul treats security as an operating constraint, not a final checklist. Useful companion agents need memory, tools, scheduled work, and access to real systems; safe companion agents need explicit boundaries, auditable escalation, and reversible change paths.

Use this page as a map. Each visual card opens the security page it represents.

Explore security

Agent Isolation and Access

Container boundaries, profile isolation, brokered host access, and change-request gates without making the companion agent useless.

Companion Agent Secure Coding Configuration

A secure coding setup for companion agents using isolated development containers, scoped project access, policy gates, and reviewed escalation.

Planned security coverage

Additional security pages on containerisation and development workflow will be added here once their source pages are ready, so this index only sends readers to live routes.