output

code_context_agent.models.output ¶

Structured output models for agent analysis results.

These models define the shape of the agent's final structured response. File generation (CONTEXT.md, bundles, etc.) still happens via tools. The structured output captures the analysis summary and metadata.

BusinessLogicItem ¶

Bases: FrozenModel

A ranked business logic item discovered during analysis.

ArchitecturalRisk ¶

Bases: FrozenModel

An identified architectural risk.

GeneratedFile ¶

Bases: FrozenModel

Record of a file generated during analysis.

RefactoringCandidate ¶

Bases: FrozenModel

A suggested refactoring opportunity.

CodeHealthMetrics ¶

Bases: FrozenModel

Aggregate code health metrics.

PhaseTiming ¶

Bases: FrozenModel

Timing data for an analysis phase.

AreaRiskProfile ¶

Bases: FrozenModel

Risk profile for a specific codebase area, used for review routing.

RiskProfile ¶

Bases: FrozenModel

Overall codebase risk profile for review routing decisions.

AnalysisConfidence ¶

Bases: FrozenModel

Confidence metadata attached to any analysis judgment.

CalibratedJudgment ¶

Bases: FrozenModel

A judgment with attached confidence and auditable reasoning.

ArchitecturalPattern ¶

Bases: FrozenModel

A codified architectural pattern detected in the codebase.

PatternViolation ¶

Bases: FrozenModel

A specific deviation from an established pattern.

ConsistencyReport ¶

Bases: FrozenModel

How consistent a set of changes is with established patterns.

RiskSnapshot ¶

Bases: FrozenModel

Point-in-time risk measurement for a codebase area.

TemporalRiskProfile ¶

Bases: FrozenModel

Risk trajectory for a codebase area across analysis snapshots.

ContractConsumer ¶

Bases: FrozenModel

A repo that consumes a service contract.

ServiceContract ¶

Bases: FrozenModel

A detected interface contract between repos.

CrossRepoImpact ¶

Bases: FrozenModel

Cross-repository impact assessment for a set of changes.

VerdictSignal ¶

Bases: FrozenModel

A single signal contributing to a change verdict.

ReviewerRecommendation ¶

Bases: FrozenModel

Recommended reviewer with rationale.

DecisionBoundary ¶

Bases: FrozenModel

Transparency about how close a verdict is to changing tier.

ChangeVerdict ¶

Bases: FrozenModel

Machine-readable verdict for a set of code changes.

Produced by the deterministic verdict engine from a diff against pre-computed codebase context (risk profiles, GitNexus structural data, git history). Designed for CI/CD consumption.

IndexFreshness ¶

Bases: FrozenModel

Tracks how current the pre-computed analysis is.

VerdictResponse ¶

Bases: FrozenModel

Top-level CI/CD output consumed by pipelines and bots.

Bundle ¶

Bases: FrozenModel

A narrative bundle about a specific codebase area.

AnalysisResult ¶

Bases: FrozenModel

Structured output for the complete analysis.

This model captures the analysis summary returned by the coordinator agent. The actual files (CONTEXT.md, bundles, etc.) are created by tools.