Quilzomanual

Open source · self-hosted · one binary, no dependencies

The control plane for your AI agents

Quilzo gives every AI agent an identity, a policy, a budget and a signed record, whichever model or sandbox it runs on, and asks a person before the actions that matter. Around it: an MCP gateway, chatbots that cite their sources, AI compliance evidence, a security console, and the content platform the agents work on.

Get started in ten minutes Build an AI agent Source on GitHub
A Quilzo agent stopped before writing a page: the exact call it wants to make, field by field, with Let it go ahead and Decline

A declaration, not a prompt

Most agent frameworks give a model your tools and a system prompt asking it to behave. Quilzo starts from the other end: the model never holds the authority.

Limits enforced outside the model

Each agent declares what it may do, read and spend, and one gate enforces it on every action. A hijacked agent still cannot do anything it did not declare.

A person decides what matters

Capabilities marked ask first stop the run until somebody approves the exact call. Autonomy is earned through evaluations with planted instructions.

Every action is provable

Runs, app calls and forgotten memory come with receipts proved against a log head signed with Ed25519 and ML-DSA, checkable by anybody.

Get started

AI agents

Models and chatbots

Connect

Content

Release

Security operations

Compliance and audit

Administration

More