cleanup: remove stale v6.1 outputs, top-level suggest.py, and pycache

probes: K8s-honest framing + probe-5 rename

feat(galaxy): top 25 legend kinds + wider palette

fix: add tokenizers dependency

feat(v9): sub-tokenization — [UNK] collisions fixed + namespace probe passes + apiVersion probe added

fix: add scikit-learn dependency

feat: structural probes tab + galaxy ns fix

feat: add manifest galaxy tab + overview landing

deploy: v8 mainline (post-cleanup) — renamed v8_model to model, v7 archived

deploy: v8 (MLM+recon, 276K) — atomic-vocab + doc_vec architecture

deploy: v7 — broader output vocab, status keys now predictable

deploy: restructure — Space bundle now sourced from hf-space/ in main repo

suggest: re-push with empty-parent + None-value handling

suggest: also probe under 'key:' (None value) as empty parent — PyYAML parses 'resources:' as None, not {}

suggest: probe empty-mapping parents (e.g. securityContext: {}, resources: {}) and refactor probe loop into shared _run_probe helper

ui: drop top_k slider — it was a no-op given threshold + kind-conditioning

ui: auto-derive example labels from YAML (Kind: name; MultiDoc for bundles)

logs: timestamped progress through import / model load / UI build

infer: kind-condition kind_head predictions (mask non-matching-kind logits before softmax)

ui: default threshold 0.1 (show more by default; slide up to filter)

ui: compact example labels + no pagination + drop redundant hint

examples: drop redundant threshold/top_k columns

examples: add 8 more (StatefulSet, CronJob, HPA, NetworkPolicy, Ingress, Pod+probes, Secret, incomplete Deployment)

feat: multi-document YAML support (split on ---, process each doc separately)

ui: trim intro + limitations — drop internal version chatter

ui: 300-line gate with friendly error for oversized YAMLs

ui: default confidence threshold 0.75 (STRONG only)

ui: textbox starts at 22 lines, grows to 100 then scrolls

fix: cap textbox height; button-only trigger (no auto-fire on change)

fix: unpack (suggestions, skipped) tuple from suggest_missing_fields

Initial app: Gradio missing-field suggester (v6.1 model)

initial commit