Argus – Open-Source Agentic Security Platform – MIT

Argus, an open-source agentic security platform released under the MIT license, unifies visibility across cloud, identity, GitHub, Kubernetes, endpoints and network in a single view. The platform uses specialist AI agents with read-only tools to investigate threats, and gates every response behind dry runs, Slack approvals, limits, kill switches and rollback, alongside model access and budget controls, DLP and shadow AI discovery. Argus also ships authorized pentesting, agent canaries, sandbox attacks, a signed audit chain, encrypted backups and control mappings.

Unified visibility Cloud, identity, GitHub, Kubernetes, endpoints and network in one view. One open-source platform to monitor your infrastructure, detect attacks, investigate with AI agents and respond through controlled, reversible actions. Built-in controls for every response Specialist agents investigate threats with read-only tools. Dry runs, Slack approvals, limits, kill switches and rollback. Manage model access, budgets, DLP and shadow AI discovery. Authorized pentesting, agent canaries and sandbox attacks. Signed audit chain, encrypted backups and control mappings.