Skip to main content
Nasiko has two independent access-control mechanisms, checked separately:
  1. User → agent: which people can view, chat with, update, delete, or manage an agent’s secrets.
  2. Agent → agent: which agents may call which other agents.
Neither implies the other. A whole org unit can chat with a research agent while that agent’s ability to call a billing agent stays locked to an explicit allowlist.

User and org management

Create users, assign roles, build the org structure.

User → agent access

Ownership, the public flag, and explicit grants.

Agent → agent MCP access

Per-agent connector and tool permissions.

Access control reference

Full technical reference.

Roles

Plans: Pro and Enterprise. Self-hosted: Enterprise license.
On a self-hosted open-source server, there’s no manager role or org structure. Every signed-in user can deploy agents, and managing users needs the admin role or superuser. Every user has exactly one role. Each role includes everything below it. A role sets what kind of action a user may attempt. Whether they can act on a specific agent is decided by user → agent access. You don’t assign manager directly. Appointing someone as an org unit’s lead makes them a manager, and removing them as lead of their last unit returns them to member. See User and org management.
Role is not superuser. Superuser is a separate flag. A superuser passes every role check and can act on any agent. Setting role to admin doesn’t grant it. See user and org management.

Where each mechanism applies

An agent with no allowlist entries can call any agent. Once it has entries, it can call only the agents listed. If a call between two agents fails with 403, check whether the caller has an allowlist that leaves out the target.