topos.evaluation.suggestions

Refactor-suggestion engine — turns a score into actionable next steps.

Maps the metrics that failed their policy gate (and any active security findings) into concrete, imperative, refactor-focused instructions an agent or developer can act on directly. Thresholds come from the central calibration singletons so messages quote the real numbers.

Pure and side-effect-free so both the CLI and (later) the MCP layer can render the same suggestions.

class topos.evaluation.suggestions.Suggestion(pillar, metric, severity, message)

Bases: object

One actionable, refactor-focused next step.

pillar

metric

severity

message

topos.evaluation.suggestions.suggest_refactors(result, *, active_findings=None)

Build actionable suggestions from a classification result.

active_findings are the security findings that are NOT allowlisted; only these produce SECURE suggestions.