3-Layer Isolation
Container + namespace + process-level isolation with seccomp-bpf and Landlock LSM.
Build, Share & Run AI Agents
Discover powerful AI agents on the marketplace, build your own with the visual Builder Lab, and execute them anywhere — from the web to your IDE.
Everything you need for AI agents
From discovery to execution, Flezi AgentBox provides the complete platform for building and running AI agents.
Agent Marketplace
Browse, search, and discover published AI agents. Star your favorites and follow top creators.
Builder Lab
Visual three-phase builder: brainstorm on a canvas, design logic as a DAG, and synthesize a blueprint.
Execution Engine
Run agents in secure Docker containers with real-time streaming. Isolated, sandboxed, and fast.
IDE Integration
Connect from VS Code or Cursor via MCP. Discover and invoke agents right from your editor.
Growing ecosystem
0
Published Agents
2
Community Members
0
Agent Executions
Enterprise
Enterprise-Grade Security
Run untrusted AI agents with confidence. NemoClaw provides hardware-backed isolation with comprehensive audit trails and compliance-ready security policies.
Trust Levels
4-tier trust system (Untrusted → Certified) with graduated permissions.
Policy Engine
YAML-based security policies with visual editor, dry-run, and preset templates.
Network Isolation
Default deny-all with per-domain allowlisting. Each sandbox gets its own network namespace.
Filesystem Guard
Kernel-enforced path restrictions via Landlock. Ephemeral overlay FS per execution.
Audit Trail
Immutable hash-chain audit logs with OCSF export for SIEM integration.
Ready to get started?
Join the Flezi AgentBox ecosystem. Build your first agent or explore what others have created.