File-backed tracker for active work, completed tasks, and good ideas.
Next: Fix remaining cron/script drift around event reporting and batter-props path issues.
Next: Design a more visual MLB dashboard so daily-picks is not the primary way to consume same-day MLB updates.
Next: Document upgrade procedure, validation checklist, rollback path, and staging/test strategy.
Next: Decide whether to add council integration, grading, daily automation, or leave as research-only.
Next: Observe for a week and tune if major artifacts still slip past memory/shared and dashboard.
Next: Reproduce safely or inspect the exact flow that rewrites openclaw.json to minimal state.
Next: Leave blocked unless there is strong reason to reopen; current workaround is dedicated lanes or mention-gated separation.
Patched mlb_score_nrfi_today.py to stream raw_json-derived inputs instead of materializing the large blob column in pandas.
backup.py hardened so Git stages curated code/config/docs while B2 owns bulky artifacts.
Phase 1 cleanup completed; large generated MLB CSVs were untracked while staying on disk.
Added WNBA namespace/system status into shared memory and documented current model readiness.
Added mlb/analytics build flow, DuckDB mirror, query script, and first analytics views.
Advisor workspace updated so DuckDB is an approved read-only analytics layer while canonical MLB DB remains source of truth.
Would make ChiueyBETS stronger on batter vs handedness, pitcher context, F5, and NRFI support questions.
Would shrink the repo substantially, but is not required for day-to-day hybrid backup correctness.