A harness around swappable agents

The idea

The parts of a worker's harness are pluggable: models, tools, context, rules, checks, decisions, verification, memory and evaluation. Each is set per organization, repository, person and run.

The problem
Teams get locked to one vendor and start over when they switch.
Why it matters
The model changes, but the team's work, rules and memory stay.
Where it stands
In production as of

A step-by-step replay of this idea is not published yet. The library adds one when a recorded run shows it.

  • Run modes as presets over knobs

    Named presets set routing, review, verification, the fast path, prefetching and the other harness settings per run, item, person, project, team or organization. A preset never weakens a hard gate.

  • Own accounts, execution profiles and payers

    Hosted work uses API keys. A subscription runs only through the vendor's product or the person's own runner. Executor, model, host and payer inherit from the platform through org, role and worker.

Turn the numbers into shipped work.

Crewshal runs these choices for you: a persistent AI worker with memory and rules, on your Claude and Codex subscriptions.