Autonomy, with a leash.
Agents act on your behalf. So every layer is built to keep them inside the lines you draw — and to show you exactly what they did.
Isolated sandboxes
Agents that run code do so in their own container, with no access to the host or to other companies.
Network allowlist
Sandboxes reach the internet only through an egress proxy, and only the hosts you approved.
Human approvals
Write and external actions can require a human decision. Policies are set per tool and per agent.
Strict company isolation
Agents, keys, memory, runs and audit log belong to exactly one company and are never shared.
Encrypted secrets
Provider keys and tokens are envelope-encrypted at rest. Backups are encrypted too.
Audit log
Every change, approval and tool call is recorded, so you can always see who — or which agent — did what.
Role-based access
Members get permissions per company; agents only get the tools you give them.
Budgets & kill switch
Hard spending limits per agent and company, and one Halt button that stops all runs.
Your data & GDPR
We act as processor for the data you put into Mandistry. A data processing agreement (DPA) is part of every plan. Data is hosted at [Hosting provider, data centre in the EU].
Read about our DPAModel providers
You choose which model providers your agents use, with your own keys or subscription. Prompts and tool results go to that provider under your own agreement with them — nowhere else.
Report a vulnerability
Found a security issue? Mail [security@yourdomain.nl]. We respond within 3 working days.
Disclosure policy