Case study · Agent harness
The harness I built when my framework ran out of road.
A cloud-native agent harness that matches each step to the cheapest capable model, survives provider outages, and holds no secrets. It runs Offworld in production.
The problem
Three walls.
I built Offworld on OpenClaw and hit three limits. Its security posture was weak, a real problem for anything long-term or enterprise. It required a local daemon, when I wanted something cloud-hosted and always on. And the interface was painful: hard setup, a limited GUI, and multi-agent work routed through Discord, where agents sitting in an open chatroom burned tokens just listening. When Anthropic's API went down, everything stopped, and all the context was stuck with it.
The approach
Match the model to the judgment the step needs.
Kaiger is named for the Kiger Mustang herd and Kiger Gorge on Steens Mountain, in the Owyhee high desert. Two mechanics carry it.
Contextual model switching
Each step runs on the cheapest model that can do it, chosen automatically by task difficulty, with manual override. Different agents run different models.
Model-agnostic hot-swapping
When a provider's API goes down, Kaiger switches models and keeps the context, so work continues.
No secrets in the harness
Credentials and enforcement live in Owyhee. Kaiger runs agents and picks models; Owyhee decides what each agent may touch, at every tool call.
The proof
The cost ladder.
Offworld started as an open newsroom: five agents on Opus in one shared room, discussing pitches and drafts. The open discussion worked. Agents caught each other's hallucinations, factual errors, and logical fallacies. It also cost $100 a day. The goal was to keep that error-catching and stop paying frontier prices for every step. Each step down came from one idea: once a task needs less judgment, it can run on a cheaper model.
| Step | Change | Daily cost |
|---|---|---|
| 1 | Five agents in an open room, all on Opus | $100 |
| 2 | Two-role loop: the journalist pitches, the editor challenges and assigns, the journalist researches and drafts, the editor critiques, fixes go back | $50 |
| 3 | The narrower, more mechanical steps move from Opus to Sonnet | $15 |
| 4 | Writing moves to an open-weights model; a deterministic script checks facts and source links, which the open model sometimes dropped or got wrong | $0.39 |
The result
39 cents a day at the same quality bar.
The lesson I took from it: structure the work so judgment is concentrated in a few places, match each step to the cheapest model that can do it, and use plain code wherever the check is mechanical. Kaiger runs Offworld in production today and is what I build and test everything else on.
Part of the Kipple Labs stack
Offworld is the publication. Kaiger runs its agents and picks their models. Owyhee holds their credentials and enforces what each one may do. AXIS gives them an identity any company can verify.
It all begins with an idea. Maybe you want to launch a business. Maybe you want to turn a hobby into something more. Or maybe you have a creative project to share with the world. Whatever it is, the way you tell your story online can make all the difference. Don’t worry about sounding professional. Sound like you. There are over 1.5 billion websites out there, but your story is what’s going to separate this one from the rest. If you read the words back and don’t hear your own voice in your head, that’s a good sign you still have more work to do.