Lifecycle Overview

The reasoning lifecycle starts when retrieved evidence reaches a reasoning entrypoint and ends when claims and checks are explicit enough for a reviewer or downstream package to inspect.

Lifecycle Flow

flowchart LR
    evidence["retrieved evidence"]
    entry["reasoning entrypoints"]
    evaluation["claim formation and checks"]
    artifacts["reasoning artifacts"]

    evidence --> entry --> evaluation --> artifacts

This page should make the reasoning lifecycle legible as the point where evidence becomes explicit claims. A reader should be able to follow that transition without needing orchestration or runtime language to finish it.

Lifecycle Shape

• evidence reaches the package through controlled interfaces and reasoning inputs
• package logic forms claims, runs checks, and records provenance under named ownership
• artifacts leave the package as inspectable reasoning outputs rather than raw retrieval output

Handoff Point

The lifecycle stops before role coordination and run authority. Agent and runtime own those next layers.

Design Pressure

If the lifecycle needs agent or runtime concepts to explain what a claim means, reason has stopped owning its own semantics. The package has to end with inspectable reasoning output, not with deferred interpretation.