Open protocol from Krystal Unity

Krystal Loop Protocol

Hard-working agents. Big-brain control. No chaos.

Let fast, affordable AI agents do the heavy lifting while a stronger lead keeps the job focused, checks what comes back, and knows when to stop.

Why it exists

More work done without losing the plot.

Agent loops can move incredibly fast. They can also burn through your best model's attention, wander away from the original job, or confidently tell you that broken work is finished.

KLP gives the team a simple working rhythm: define the job, let a worker build it, check the result, get a fresh second opinion, and repair only what the evidence supports. Normal work keeps moving automatically. You step in only when the job genuinely changes or reaches a real limit.

The loop

Five clear stages. One controlled outcome.

Each stage has a purpose, and no agent gets to mark its own homework.

01

Scope

Define the outcome, boundaries, budget, checks, and stop conditions.

02

Worker

Give a capable, affordable agent one bounded job with a clear handover.

03

Checks

Use real tests and direct evidence before asking another AI for an opinion.

04

Critic

A fresh agent reviews the exact result and its findings are verified.

05

Integrate

Repair within the agreed limits, then check the combined result again.

What changes

Vibe coding without the hangover.

Spend the big brain wisely

Use stronger models for planning, judgment, and hard decisions instead of making them carry every line of routine work.

Keep every agent on the same job

Clear boundaries and exact handovers stop workers from overlapping, reopening solved work, or quietly changing the brief.

Stop loops becoming endless

Repair limits and stop conditions prevent the team from polishing the same corner forever or spending beyond the value of the task.

Current release

A working reference, not a magic promise.

KLP Core v0.2 is public, provider-neutral, and includes a reproducible offline example of a failed build being reviewed, repaired, and checked again.

Release KLP Core v0.2
Models Provider-neutral
Proof Offline fail-review-repair fixture
License Open source and open documentation

Give it to your AI coding agent.

Ask what it would change about the way your agents plan, build, check, and repair work today.