Documentation
¶
Overview ¶
Package grading provides parsing and scoring for agent output reports.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func FormatResult ¶
func FormatResult(r *GradeResult) string
FormatResult returns a human-readable multi-line summary of the grade result.
func RequiredSections ¶
RequiredSections returns the list of required sections for a given mode.
Types ¶
type AgentStats ¶
type AgentStats struct {
Agent string `json:"agent"`
TotalRuns int `json:"total_runs"`
LatestGrade string `json:"latest_grade"`
AvgScore float64 `json:"avg_score"`
AvgReportQuality float64 `json:"avg_report_quality"`
AvgIterationEfficiency float64 `json:"avg_iteration_efficiency"`
GradeCounts map[string]int `json:"grade_counts"`
}
AgentStats provides aggregate statistics for an agent.
type GradeHistory ¶
type GradeHistory struct {
Grades []*GradeResult `json:"grades"`
}
GradeHistory is the top-level container written to and read from the store file. It holds all recorded GradeResult values in append order.
type GradeOptions ¶
type GradeOptions struct {
// Agent is the name of the agent being graded.
Agent string
// Iterations is the number of model iterations consumed by the run.
Iterations int
// FileCount is the number of files in the target codebase.
FileCount int
// RunID is an optional identifier linking the grade to a specific run.
RunID string
}
GradeOptions configures the grading calculation.
type GradeResult ¶
type GradeResult struct {
// Identification
Agent string `json:"agent"`
Timestamp time.Time `json:"timestamp"`
RunID string `json:"run_id,omitempty"`
// Overall grade
Grade string `json:"grade"`
TotalScore float64 `json:"total_score"`
// Component scores (0-100)
ReportQuality float64 `json:"report_quality"`
IterationEfficiency float64 `json:"iteration_efficiency"`
// Manual review indicators
FindingQualityReview bool `json:"finding_quality_review"`
SkipDisciplineReview bool `json:"skip_discipline_review"`
// Metadata
Iterations int `json:"iterations"`
FileCount int `json:"file_count"`
FilesTouched int `json:"files_touched"`
IssuesFixed int `json:"issues_fixed"`
IssuesSkipped int `json:"issues_skipped"`
Mode string `json:"mode"`
// Breakdown details
MissingSections []string `json:"missing_sections,omitempty"`
Notes []string `json:"notes,omitempty"`
}
GradeResult represents the computed grade for an agent run.
func ComputeGrade ¶
func ComputeGrade(parsed *ParsedOutput, opts GradeOptions) *GradeResult
ComputeGrade calculates the automated grade for a parsed agent output. It scores report quality (10% of total) and iteration efficiency (15%), returning a GradeResult with a letter grade and component breakdowns. The remaining 75% requires manual review.
type ParsedIssue ¶
ParsedIssue represents a single issue found or fixed.
type ParsedOutput ¶
type ParsedOutput struct {
// Report sections detected
HasChangesSummary bool
HasAnalysisSummary bool
HasFilesTouched bool
HasValidation bool
HasIssuesFixed bool
HasIssuesSkipped bool
HasFindings bool
// Extracted data
FilesTouched []string
IssuesFixed []ParsedIssue
IssuesSkipped []ParsedSkip
Findings []ParsedIssue
ValidationPassed bool
// Metadata
Mode string // "edit" or "readonly"
}
ParsedOutput represents the structured data extracted from agent output.
type ParsedSkip ¶
ParsedSkip represents an issue that was skipped.
type Parser ¶
type Parser struct {
// contains filtered or unexported fields
}
Parser extracts structured data from agent markdown output.
func (*Parser) Parse ¶
func (p *Parser) Parse(output string) *ParsedOutput
Parse extracts structured data from agent output markdown.
type Store ¶
type Store struct {
// contains filtered or unexported fields
}
Store persists grade results to a JSON file on disk. Use NewStore to create one at the default cache location or NewStoreAt to specify an explicit path.
func NewStoreAt ¶
NewStoreAt creates a store at a specific path.
func (*Store) List ¶
func (s *Store) List(agentName string, limit int) ([]*GradeResult, error)
List returns grades for a specific agent, sorted by timestamp (newest first).
func (*Store) ListAll ¶
func (s *Store) ListAll(limit int) ([]*GradeResult, error)
ListAll returns all grades, sorted by timestamp (newest first).
func (*Store) Save ¶
func (s *Store) Save(result *GradeResult) error
Save appends a grade result to the store.