Looptimal

Open-source loop discipline

From objective to verified outcome.

Looptimal starts with a simple idea: agent work should have a goal, durable state, an external verifier, and a stop condition before anyone calls it done.

Start with the repo Read the overview
Goal State Verifier Stop
run lo-0248 / checker lane armed / stop cap 3
Objective intake Reduce failed install attempts

One outcome, pinned before execution. No moving target after the maker starts.

Proof receipt Accepted only after live check

Evidence is stored as an artifact, not buried in a transcript.

ready to verify
Durable state State survives the session
attempt logged to state.md failure mode linked to run next action has owner
External verifier The gate can say no

Maker output is not accepted until a separate check passes.

Run evidence

A small standard for better agent loops.

Looptimal is for teams trying to move beyond one-off prompts and brittle automation. It gives loop-shaped work a structure that can be reviewed, run, rejected, and improved.

Define the loop

State the objective once, keep durable context, and make the success gate explicit before execution.

Separate maker and checker

The thing that produced the work should not be the only thing grading whether it is done.

Start open

The repo is the starting point: MIT licensed, inspectable, and designed to be adapted into your own workflow.

Start here

Use the repository install instructions.

The current source of truth is the GitHub README. Start there for plugin install, harness-specific setup, and the latest tagged command examples.

Open install instructions View repository
Quick start 
# Repository
https://github.com/Renn-Labs/Looptimal

# README section
https://github.com/Renn-Labs/Looptimal#install

# Quick self-test from the README
uvx --from git+https://github.com/Renn-Labs/Looptimal@v2.1.0 verify-outcome --selftest

Why we exist

Agents need operating discipline, not just better prompts.

The failure mode is familiar: the agent keeps going, forgets what happened, shifts the target, or declares success because its own output looks plausible. Looptimal exists to make the loop inspectable before it runs and accountable after it runs.

1
Fewer vague done claimsMake the acceptance gate visible instead of relying on self-report.
2
Reusable loop startersTreat useful workflows like repo starters: fork, adapt, run, and improve.
3
Human control where it mattersKeep judgment calls explicit while automating the repeatable parts.