wa:heuristic-status

View heuristic observation stats by running a dashboard that reads observations.jsonl and heuristics.md to present counts, tool distribution, confidence buckets, and top heuristics. This dashboard surfaces live metrics so you can monitor pipeline health and curate active guidance.

Manage heuristic lifecycle actions by suppressing or promoting heuristics by title through dashboard controls. Promoting a heuristic updates its status and applies confidence flooring rules deterministically, while suppressing removes it from active guidance.

Export active heuristics as a shareable markdown block to move heuristic definitions across environments. You can also import heuristics from an external file, with built-in duplicate detection ensuring clean migration and validation.

Heuristic confidence scoring categorizes observations into confidence buckets to indicate rule reliability. Manual promotion applies a confidence floor, ensuring that any promoted heuristic meets a minimum confidence threshold before joining active guidance.

Heuristic lifecycle management requires reading observations.jsonl and heuristics.md files. These specific formats provide the raw observation data and markdown guidance definitions needed to generate dashboard metrics and execute status updates.

The best way to curate active heuristic guidance is using a dashboard that visualizes observation stats and enables deterministic lifecycle controls. This lets you promote high-confidence rules and suppress unreliable ones to maintain pipeline quality.