Where answers come from

Some pages can ask a language model questions. Set this up once; your choice is remembered in this browser for the whole site.

Model source
Use another source instead

Smallest-to-largest weight downloads. Qwen3.5 0.8B is our starting choice for page Q&A. GPU estimates are not total RAM: leave room for the browser, context and temporary buffers. Hardware changes speed and capacity; answers still need checking.

Session-only loading avoids saving model weights, uses extra RAM and downloads again after unloading. Turn it off to save downloads. Delete-data-on-exit or private browsing may impose a much smaller storage limit than the displayed estimate.

You sign in on OpenRouter; it gives this page a key for your account.

A free account does not make every model free. Choose a model with a listed :free variant; merely adding that suffix does not create one. Check free-model availability and pricing.

Chat

AI assistant, not Shawn Hartsock. It does not speak for him.

Inspect the page excerpt

Up to 4,800 characters, not the whole site. Recent conversation turns are retained within the model’s context limit.

Ask about this page, or just say hello.

Ready when you are.

Not saved or uploaded automatically. Closing keeps this conversation in this tab; changing pages starts a new one. Model replies can be wrong.

Series

Newt-Agent

An experimental agentic coder in Rust, built for local models first.

Newt-Agent logo: a newt

Newt is an experimental agentic coder in Rust, built for local models first. The operator grants each tool call a capability, and the grant is enforced in the kernel where the OS allows.

Each page is a step-through animation checked against the source, with a table of what is implemented and what is only design. The pages are revised as the project changes, and each names the source revision it came from.

Links: Source, Vision, Design laws, Terminal-Bench scoreboard, Changelog.

Try it:

git clone https://github.com/Gilamonster-Foundation/newt-agent
cd newt-agent
just install
newt

The first run opens a setup wizard. newt --help is the authority on what the binary does.

Source: newt-agent site/index.html at 67b24697. newt-agent is Apache-2.0.

  1. How Newt manages contextA model's window is small and every turn re-sends it. When the window nears its budget, Newt does not throw the middle away. It keeps the task, keeps the recent work, saves the rest verbatim, and leaves the model a way to ask for it back. Step through one compaction below.~5 min read
  2. Paged context over a derivation graphThe goal is the effect of a huge context window without a huge window: bounded residency, unbounded reach. The model is never told it holds everything. It is told what it holds and that it can reach the rest.~5 min read
  3. How agent-frame assists contextNewt has two ways of managing context. The default one, compaction, does not use agent-frame. The opt-in smart harness does: it records the session as content-addressed events, lets a small model propose what to keep, lets the host decide what is allowed, and renders the request from the result. Both stand on the same content-addressing crate.~5 min read
  4. How Newt grants and enforces authorityEvery tool call runs under a capability: what it may read, write, run, and reach. The operator sets the ceiling, the gate checks each call, and the kernel enforces it where the OS allows. Step through one write that is allowed and one network call that is not.~5 min read
  5. How Newt runs shell commands with brushModels write shell.~5 min read
  6. How Newt tracks who is actingTwo separate things carry identity. A key chain decides what a session may do. A ledger decides who is credited on a commit. They are not the same mechanism, and the animation keeps them apart.~5 min read
  7. How Newt sessions dock across machinesA hub can list, mirror and steer sessions running on another of the operator's own machines. The session never moves: the host stays the only writer. Step through discovery, pairing, approval, mirroring, injecting and revoking.~5 min read