Documentation
¶
Overview ¶
Package review is the Stage-1 namespace for self-review / critique / quality scoring types in package engine. See ../REFACTOR_PLAN.md.
Index ¶
- Constants
- type Assessment
- type Bot
- type Comment
- type ConsensusResult
- type ConsensusSampler
- type Critic
- type LLMClient
- type PatchVerdict
- type QualityScorer
- type Report
- type ResponseContext
- type ReviewResult
- type Rule
- type Sample
- type ScoreWeights
- type ScoredResponse
- type SelfAssessor
- type SelfReviewResult
- type Solution
- type SolutionReviewer
- type TaskContext
Constants ¶
const ConfidenceThreshold = engine.ConfidenceThreshold
ConfidenceThreshold is the minimum confidence at which self-review approves a write without asking for human input.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Assessment ¶
type Assessment = engine.Assessment
Assessment is a structured self-review of an in-progress task.
type ConsensusResult ¶
type ConsensusResult = engine.ConsensusResult
ConsensusResult is the reduced output of N samples.
type ConsensusSampler ¶
type ConsensusSampler = engine.ConsensusSampler
ConsensusSampler runs N samples of an LLM prompt and reduces them.
func NewConsensusSampler ¶
func NewConsensusSampler(numSamples int) *ConsensusSampler
NewConsensusSampler returns a sampler configured for numSamples draws.
type PatchVerdict ¶
type PatchVerdict = engine.PatchVerdict
PatchVerdict is the Critic's accept/reject decision on a patch.
type QualityScorer ¶
type QualityScorer = engine.QualityScorer
QualityScorer ranks candidate responses against a rubric.
func NewQualityScorer ¶
func NewQualityScorer() *QualityScorer
NewQualityScorer returns a scorer with default weights.
type ResponseContext ¶
type ResponseContext = engine.ResponseContext
ResponseContext is the input the scorer scores against.
type ReviewResult ¶
type ReviewResult = engine.ReviewResult
ReviewResult is the SolutionReviewer's verdict.
type ScoreWeights ¶
type ScoreWeights = engine.ScoreWeights
ScoreWeights configures a QualityScorer.
func DefaultWeights ¶
func DefaultWeights() ScoreWeights
DefaultWeights returns the default scoring weights for QualityScorer.
type ScoredResponse ¶
type ScoredResponse = engine.ScoredResponse
ScoredResponse pairs a candidate with its score breakdown.
type SelfAssessor ¶
type SelfAssessor = engine.SelfAssessor
SelfAssessor produces Assessments mid-loop.
func NewSelfAssessor ¶
func NewSelfAssessor() *SelfAssessor
NewSelfAssessor returns a fresh self-assessor.
type SelfReviewResult ¶
type SelfReviewResult = engine.SelfReviewResult
SelfReviewResult is the output of ReviewBeforeWrite.
func ReviewBeforeWrite ¶
func ReviewBeforeWrite(ctx context.Context, llm LLMClient, model, intent, filePath, oldContent, newContent string, ) (*SelfReviewResult, error)
ReviewBeforeWrite runs an LLM-driven self-review on a candidate write.
type SolutionReviewer ¶
type SolutionReviewer = engine.SolutionReviewer
SolutionReviewer evaluates multiple proposed solutions and picks one.
func NewSolutionReviewer ¶
func NewSolutionReviewer(maxAttempts int) *SolutionReviewer
NewSolutionReviewer returns a reviewer capped at maxAttempts iterations.
type TaskContext ¶
type TaskContext = engine.TaskContext
TaskContext is the input to a SelfAssessor.
Source Files