Wants the team coding agentically, but must stay in control of spend, data sensitivity and which code leaves the building.
Platform · Agents
Coding agents you dare to put into production
Your developers run their own coding agent (Claude Code, opencode, whatever they like) locally, but against the Numaga gateway. Each agent is bound to a user and inherits their limits and guardrails. Your source code and prompts stay inside your governance.
Who it’s for
Want to use Claude Code or opencode, but it is not allowed: the tool ships source code to a US service. Through Numaga it is.
Wants agent traffic under exactly the same policy and audit as any other AI use. No shadow IT.
The problem
Agentic coding is the productivity leap your policy forbids.
Tools like Claude Code send your source code, prompts and context straight to hyperscalers under US jurisdiction. So many companies ban them. Then developers use them in the shadows, or miss out on the gains. Either way it costs you.
Numaga puts the gateway in between: developers connect their own agent to Numaga instead of straight to a provider. Traffic runs through the same control plane, code stays within jurisdiction, and spend and access are capped per user.
How it works
From zero to a connected coding agent.
An agent is not an exception to your policy; it is a user of it.
Create an agent
In the platform: Agents → new agent → give it a name. You immediately get a personal Numaga API key plus connection instructions.
Connect your local tool
Drop the key into your coding agent: Claude Code, opencode, or another. The agent now talks to the Numaga gateway instead of straight to a model provider.
Bound to the user
The agent inherits the user’s tier, consumption limits and guardrails. Spend and data sensitivity stay capped: no separate, uncontrolled access.
Through the control plane, fully audited
Every agent call passes the same six checkpoints (classify, redact, risk-gate, route, meter, log) and sits in the same immutable audit trail.
In practice
Proven where data is most sensitive.
Numaga is running a pilot with Ontdek Zorg, a home-care organisation in Gelderland. Healthcare, with special-category personal data under the GDPR, is where responsible AI is hardest. That is exactly where we put the control plane to the test.
- Home care in Gelderland
- Special-category data (GDPR Art. 9)
- Policy and redaction on every prompt
Give your developers agentic coding, without losing control.
In a demo we show how a connected coding agent stays inside your governance.
Request demo access