Agyn

About Agyn

agyn is a self-hosted, Kubernetes-native AI agent platform for deploying agents into enterprise environments with security and governance controls. It is designed for teams that need to ship agents to multiple users in a company while keeping access scoped, actions reviewed, and costs visible.

The product focuses on enterprise deployment workflows: agents can be deployed into private networks, VPNs, VPCs, and firewalled environments so they can reach internal services. It supports defining agents, sandboxes, tools, MCPs, skills, and prompts in code, with a GitOps-style workflow and Terraform-based configuration shown in the documentation examples. The site also shows compatibility with Claude and Codex.

agyn is aimed at platform, security, engineering, data, and operations teams that want to manage agent access across departments. It provides least-privilege execution, static policies, a policy agent that reviews actions before execution, secrets vault support, and prompt-injection defense. It also includes per-agent tracking for budgets, usage alerts, and cost attribution, plus role-based access control and audit logs for team sharing.

The homepage positions the platform as a way to solve the problem of agents failing to access corporate resources from local environments and to scale those agents safely across an organization.

Key features:
- Deploy agents into private networks, VPNs, VPCs, and other restricted environments.
- Enforce least privilege with static policies and a policy agent that inspects tool calls.
- Keep secrets hidden from the model and defend against prompt injection.
- Track usage per agent with budgets, alerts, and cost attribution.
- Control access with role-based permissions, audit logs, and team sharing.
- Define agents, sandboxes, tools, MCPs, skills, and prompts in code.
- Use open-source, Kubernetes-native deployment with GitHub-hosted bootstrap instructions.