Guide

Explorer

Every harness's paths expanded for a platform and home directory you choose, and the command each invoke mode would spawn.

The explorer is the registry in a page. Pick a harness, a platform and a home directory, and it shows every path the way resolve() returns it, with the entries hidden on that platform counted. Pick a mode, a model and a prompt, and it builds the exact command line invoke() would run, or the rejection you'd get.

Nothing is spawned and nothing is read from your machine. One liberty: on Windows the page fills %PROGRAMDATA% and friends with the usual defaults, where the library would read your environment and leave an unset variable as written. The page ships a snapshot of the published registry, and the same snapshot feeds the landing and the harness pages. For the real answer on your disk:

harnesses paths claude
harnesses run claude --read-only "…"   # rejected, and the explorer shows you why before you try

Every state of the page is in the URL, so a link to /explorer?id=codex&platform=win32&mode=readOnly opens exactly that. "Where does Grok keep sessions on a Mac" is a question a page answers faster than a file in src/harnesses/, and that's the whole reason it exists.

Edit this pageorReport an issue

Custom harnesses

Extend the abstract Harness class, register it, and it shows up in getHarness, detection, the CLI and the tools.

Overview

Twelve harnesses, one page each, with every path, template and marker straight from the registry.