Refunds, Not Debt

Banking data shows 2024 net cash flow was nearly break-even, and 2025 came in positive. The anxiety framing around 2023–2024 unfiled tax returns says "debt problem" — but the actual signal is "refund opportunity." Income dropped, a W-2 shift reduced deductions, and an active installment agreement proves commitment, not avoidance.

When those returns get filed, the narrative becomes "Mike filed overdue returns and received a refund" — not "Mike owed more taxes." That's a materially different emotional context. The $750/month installment agreement is evidence of responsibility. The unfiled returns are an execution gap, not a moral failing.

This reframes projects/finances/ from a "problem to solve" to an "opportunity to execute." Motivation shifts from shame-avoidance to positive action. Nobody wants to budget. Everyone wants to purge chaos.

open-loops.md lists "Folicare Proposal — REVAMPED + deployed Mar 13… Awaiting Mike decision: send link to Sam?" The blocker reason is implicit, not explicit. Compare with "Fire Crown — awaiting Mark MacKenzie media kit/examples" — that's a clear dependency. "Folicare — awaiting decision" is not.

The vagueness means sessions generate follow-up questions instead of executing. Improvement #14 proposed adding a [BLOCKER REASON] field to open-loops — this item is a perfect proof-of-concept. When the reason is visible, the session knows whether to wait, ask, or act.

No action needed now. This is confirmation of Improvement #14's design soundness, not a new problem.

MEMORY.md Quick Reference says "Last updated: 2026-03-13 3:00 PM" — it's now 5 AM on March 17. The top 5 items include "mikekhaytman.com DNS fix" which was completed on March 16. Sessions read Quick Reference first thing, but it's stale.

Improvement #1 added the Quick Reference concept. It's marked 🟡 inconclusive because execution is inconsistent — and the root cause is architectural: the sleep protocol's Phase 2 has steps for open-loops refresh and TL;DR generation, but no explicit "Quick Reference refresh" step. Good ideas don't stay fresh without codified maintenance.

Adding "Phase 2d: Quick Reference Refresh" bakes Improvement #1 into reliable execution. The refresh also surfaces which items have resolved since the last pass.

March 14 session logged an exploration: "Mike asked about cloning OpenClaw workspace onto a remote server and setting up agents for coworkers." The idea is 15–20 hours for a 3-person pilot. But it's not in any task tracker — it's just buried in a daily memory file.

This is a meta-pattern: long-term strategic ideas get captured (good!) but then sink out of sight. They live in memory files but don't bubble up to sessions because they're not in the open-loops dashboard. Compare with smaller tactical items that get tracked constantly. The strategic stuff gets outcompeted by urgency.

Adding a queued entry to open-loops makes the idea visible even at low priority, and links back to the original context so sessions can find the details.