Expand description
Tier-2 scored-suggestion queue and tier-3 expansion candidates.
Routing for MIF ADR-020’s lower two tiers, per this workspace’s
“2b-routing, 2a-pipeline” decision: the queue and miss store are
purpose-built for scored candidate lists (rht’s --followup backlog
carries one unscored entity type per entry and is atomically rebuilt
every review — structurally wrong for either job), while the surfaces
that consume them are rht’s existing ones (/ontology-review --enrich
reads the queue; author-ontology.sh --from-clusters mines the
expansion candidates).
Everything here is written by mif-rh-cli paths only — mif-rh-mcp
stays read-only — and nothing here ever writes a finding’s
entity_type: confirming or rejecting a queue entry is the human/agent
--enrich step’s job (PDD-1).
Structs§
- Cluster
Member - One member of an expansion-candidate cluster.
- Expansion
Candidate - One ontology-expansion candidate: a recurring cluster of mutually-similar tier-3 misses.
- Suggestion
Entry - One queued scored suggestion: a finding that is not durably stamped, with its ranked, tier-annotated candidate list.
- Suggestion
Queue - One topic’s suggestion queue (
reports/_meta/suggestions/<topic>.json).
Constants§
- STATUS_
PENDING - The status every fresh queue entry starts in. Any other status is a human/agent verdict and is never overwritten by a re-suggestion.
Functions§
- expansion_
candidates - Clusters recorded misses into ontology-expansion candidates.
- upsert_
suggestions - Upserts
freshentries into the topic queue atpath.