CALIBER
Quickstart
Use CALIBER

Calibration

The practical calibration path: choose the asset family, generate candidates, inspect evidence, and move to explicit review or apply.

System userEvaluatorDeveloperHow ToGA
PrerequisitesAn evidence set or runtime question
Reviewed 2026-08-10 · current main branch docs contract

Use this page for the practical calibration path: choose the asset family, launch candidate generation, inspect the evidence, and move to explicit human review or apply only where the product supports it.

At a glance

TaskStart hereDeep reference
understand whether calibration exists for an assetcheck the per-asset modelCalibration architecture
generate better candidateslaunch the asset-specific loopCalibration architecture
inspect evidence and targetsconnect back to evaluation and releaseEvaluation and test sets, Review and release flows
recover a lost or stuck jobuse the operator runbookOperations runbook

1. What calibration is for

Calibration is proposal generation backed by evidence. It is not hidden autopilot. Different asset families expose different candidate-generation and apply semantics, so always reason about the target asset first.

2. Common tasks

You want to...Read this next
calibrate prompts or skillsCalibration architecture
understand how evaluation feeds the loopEvaluation and test sets
inspect the live-control boundary after calibrationReview and release flows

3. Common failure modes

SymptomFirst thing to check
Calibration produced a candidate but nothing went liveapply/release is a separate control path
The candidate improved one slice but hurt anotherthe evidence set or judge mix is incomplete
The job vanished or stalleduse the runbook recovery path instead of retrying blindly

CALIBER : Contextual Adaptive Lifecycle for Intelligent Build, Evaluation, and Refinement — this page is generated from the authoritative Markdown sources in docs/ and the repository-level ARCHITECTURE.md.