OpenClaw is an open-source personal assistant you configure on your own machine, and developers ask next about its API and harness. Kortix is the open-source AI Operating System: the agents, skills, memory, connectors and triggers are files in one git repo you own, and OpenCode is the harness.
The configuration is files in a git repo you own, not a file on one machine.
# kortix.yamlagents: support-triage: file: agents/support-triage.md # what it does, in plain markdown connectors: all # what it may reach$ curl -fsSL https://kortix.com/install | bash$ kortix init$ kortix shipA Kortix project is a git repo with kortix.yaml at its root. Agents, skills, memory, tools, triggers and connectors are files in that repo, so the whole setup is versioned, diffable and yours. kortix.yaml declares the machine image, connectors and triggers, and secret names and grants. Secret values never appear in the repo.
OpenClaw organizes the same idea around one Gateway. Its repository calls the Gateway the local control plane for sessions, tools, events and channel connections, keeps configuration on the machine that runs it, and says a personal assistant and a shared team deployment share one shape (OpenClaw's repository). For a developer the difference is where the program lives: a file on a host, or a repo the team clones, reviews and rolls back.
OpenCode is the harness, the open-source coding agent (OpenCode on GitHub). The runtime config sits in the same repo as the agents and skills, and covers agents, skills, commands, tools, plugins, models and providers. A session runs one agent on its own sandbox, clones the repo and works on a new branch named after the session.
The everyday loop is three commands and a review. kortix init scaffolds a project, kortix ship validates kortix.yaml, commits the change and pushes the branch, and kortix sessions new --prompt "…" starts an agent on its own sandbox. kortix cr ls lists the change requests those agents opened.
Any model provider with your own keys. Or the ChatGPT plan you already pay for. Or sign in with your OpenCode Console account for OpenCode Zen and Go. For a self-hosted instance, every model call routes through the LLM gateway on your own box (self-hosting Kortix).
Connectors reach 3,000+ apps in a click, plus MCP, OpenAPI, Postman, GraphQL and raw HTTP. Connector credentials are brokered server-side and never enter the machine. For your own code, the project exposes a REST API under /v1 with a reference at api.kortix.com/v1/docs, a typed TypeScript client in @kortix/sdk, and a Kortix MCP server at /v1/mcp that starts sessions and calls connectors as you.
OpenClaw's new capabilities usually belong in plugins built on its plugin SDK and shared through ClawHub, per its repository. In Kortix, a skill is markdown plus scripts that encode how the company does one job. An agent is agents/<name>.md (behavior) plus an agents.<name> block in kortix.yaml (governance). Both are files in the repo, so a change is a diff a teammate reads.
OpenClaw is a capable personal assistant for a machine you own. Kortix is the company system. Files live in one repo. One isolated sandbox per session. A person reads a change request before it merges. Kortix is the leading open-source alternative to Claude Cowork and ChatGPT Work.
If you already run OpenClaw, the OpenClaw alternatives guide and the Kortix vs OpenClaw comparison map the move, and you can read the blog for the write-up. A team that needs the whole control plane inside its own network starts with self-hosting Kortix.
Yes. The REST API is under /v1, with an interactive reference at api.kortix.com/v1/docs. @kortix/sdk is the typed TypeScript client, and the Kortix MCP server at /v1/mcp lets an MCP client start sessions and call connectors as you.
OpenCode is the harness. Its config, covering agents, skills, plugins, models and providers, lives in the project repo beside kortix.yaml.
Yes. A connector can be an MCP server, and connectors reach 3,000+ apps in a click, plus MCP, OpenAPI, Postman, GraphQL and raw HTTP.
In the project repo you own. Agents and skills are markdown files, memory accumulates as files, and kortix.yaml declares the machine image, connectors and triggers, and secret names and grants. Secret values never appear in the repo.
The agents, skills, memory and connector config are files, so they move into the repo. The OpenClaw alternatives guide maps the field, and the write-up compares the two projects directly.
next
Kortix is open source, self-host free, and every change lands through a review.
Open source · Any model, your keys · Self-host, VPC, or on-prem