Skip to content
KortixKortix
Esc
↑↓navigate↵open⌘Jpreview
On this page

Kortix

The open-source AI Operating System: agents, skills, memory and connectors live in a git repo you own, and every session runs on its own cloud computer.

What Kortix is

Kortix is the open-source AI Operating System. Your agents, skills, company memory, and connectors live in one git repo you own. Each session runs on its own cloud computer. It is an open-source alternative to Claude Cowork and ChatGPT Work.

Kortix is for teams that want agents that do finished work, reviewed by people. Run it on Kortix Cloud, in your VPC, or on your own on-prem network. To start, follow the Quickstart.

How it works

A project is a git repo

A project is a git repo with kortix.yaml at its root. Kortix hosts the repo, or you connect your own. Agents, skills, memory, tools, triggers and connectors are files in it. See Git and the kortix.yaml reference.

A session is one agent run

A session runs one agent in its own sandbox. The sandbox clones the repo and works on a new branch. The branch name is the session_id, a UUID with no prefix. See Sessions.

The agent opens a change request

The agent commits, pushes the session branch, and opens a change request against the project’s default branch.

A person merges it

A person with the project.gitops.merge permission reviews the change request and merges it into the default branch. A session can merge its own change request only when its agent’s kortix_permissions grant in kortix.yaml includes project.gitops.merge. The starter kortix agent has kortix_permissions: all, so it can. To require a person, give the agent a kortix_permissions list without project.gitops.merge. See Change requests.

The parts

Each part of an operating system maps to a Kortix feature.

Operating system part Kortix feature Read
Processes Sessions: one sandbox and one branch per session Sessions
File system The project repo Projects
Programs Agents and skills Agents & permissions
Drivers Connectors Connectors
Scheduler Triggers Triggers & reminders
Permissions Roles, grants and connector policies Accounts & access
Shell Web, desktop, mobile, CLI, TUI, Slack, Teams, email and the MCP server Where you use Kortix
Updates Change requests Git

Where you use Kortix

Surface What it does Read
Web, desktop and mobile apps Start and follow sessions, review change requests, and manage projects and accounts. Web, desktop & mobile
CLI kortix creates projects, runs sessions, and manages agents, secrets, connectors and triggers from a terminal or a script. CLI
TUI kortix tui runs the whole product as a terminal app. The TUI is experimental. TUI
Git kortix projects clone clones a project through the authenticated Kortix git proxy. Any git client then commits and pushes. Git
Kortix MCP server Your MCP client starts sessions, reads files and calls connectors as you, through /v1/mcp on the API. Kortix MCP server
REST API The HTTP API under /v1. Its interactive reference is at api.kortix.com/v1/docs. REST API
SDK @kortix/sdk is the typed TypeScript client for the API and each session’s runtime. SDK
Slack, Teams and email Mention the bot in Slack or Microsoft Teams with a task, and it starts a session. An email to the agent’s inbox starts a session too. Email is experimental. Turn on the AgentMail Email flag in Customize → Settings → Feature flags. Slack, Teams & email
Self-hosting The kortix self-host commands install and run the full stack on your own infrastructure. Self-hosting

Security and data

  • Connector credentials are brokered server-side and never enter the machine. See Connectors.
  • A secret can be egress-enforced. The sandbox holds a handle. Kortix substitutes the real value outside the sandbox, only on approved HTTPS hosts. Environment is the default exposure. Egress-enforced is experimental. See Secrets.
  • Kortix-managed models run on zero-data-retention endpoints. See Models.
  • The experimental us_region flag places a project’s new Platinum sandboxes in a US region, where the instance offers one. It sets compute placement only. It does not move the API, the database or the archive. See Feature flags.
  • Every action is recorded in the audit log. The log keeps 90 days of events live, and the archive keeps them up to 365 days. Reading, exporting and streaming the audit log depend on the plan. See Accounts & access.

Where to go next

Was this page helpful?