The ultrawork loop

ultrawork is the headline mode. Typing ultrawork (or the short alias ulw) anywhere in your prompt activates maximum-precision, outcome-first, evidence-driven orchestration.

"Plan, execute, verify, and keep the evidence attached."

The principle is simple: a task is not done because the agent says so. It is done when an Oracle verifies it against evidence.

What ultrawork enforces

When ultrawork is active, the harness does not settle for a hopeful "looks done":

• Strict TDD: RED → GREEN → SURFACE → CLEAN
• At least 3 realistic QA scenarios
• Real manual-QA channels (HTTP call, tmux, browser)
• A binding verification gate that loops until the work is genuinely done

$ulw-loop: verified completion

$ulw-loop is a self-referential development loop that runs until verified completion. The agent works continuously and emits <promise>DONE</promise> when it believes the task is complete — but that does not end the loop.

An Oracle must verify the result first. The loop ends only after the system confirms the Oracle verified it. If verification fails, it continues with the message:

Oracle verification failed. Continuing ULTRAWORK loop.

Syntax

$ulw-loop "task description" [--completion-promise=TEXT] [--strategy=reset|continue]

Iteration caps

The loop is bounded so it cannot run forever:

• 500 iterations in ultrawork mode
• 100 iterations in normal mode

Evidence over hopeful status

The loop does not stop at "wrote some code." It stops when the result is confirmed by evidence — the checks that ran and what they showed — rather than a status update the agent is hopeful about. That is the whole reason an Oracle sits between <promise>DONE</promise> and the loop ending.

Where it sits among the pillars

$ulw-loop is one of three command pillars, each for a different shape of work:

A common flow: $ulw-plan produces a decision-complete plan, $start-work executes it checkbox by checkbox, and $ulw-loop keeps an open-ended task running until the Oracle signs off.

Related