Zagros logoZAGROS v0.1.0
Documentation

What are you trying to do?

Zagros is designed as a shared, self-hosted security reference for teams and their AI agents. Start with the task you need to complete; implementation details and project internals are kept lower in the navigation.

TEAM

Deploy Zagros for a team

Run one shared instance, keep HelixDB private, and put MCP/UI behind your organization’s access boundary.

Open guide →AI

Connect an AI agent

Use MCP so agents and internal tools can query the same team-controlled security reference.

Open guide →TRUST

Understand what to trust

See how Zagros treats authoritative sources, provenance, freshness, and conflicting evidence.

Open guide →OPERATE

Operate and keep it fresh

Configure refresh behavior, runtime settings, and health checks for a shared deployment.

Open guide →
New installation

Start with a working instance

For a first deployment, use the installation guide. If you want an AI agent to perform the setup, use the AI install prompt.

Install ZagrosAI install
Reference

Browse by topic

Use Zagros

Connect AI agents with MCPCLI referenceData sourcesInvestigation example

Trust & security

Source trust policyThreat modelQuality & freshnessSecurity review

Agent integration

Agent harnessesSkill contractSkill compatibilitySkill trust policy

Project internals

How Zagros worksArchitectureDevelopmentTestingReleasingInterface statusProject vision
Looking for maintainer-only or historical material? Those documents remain available by direct URL and in the repository, but they are intentionally excluded from the primary documentation navigation so normal users do not have to sort through project-management material.