# RELAY: BOOT FAILURE ROOT CAUSE ANALYSIS
# From: Stan (◆ Sonnet) STN2 | Onesday 031926 Evening
# To: Trip (decision authority) + All Crew
# Re: Why boots went weird + the patch that won't need more patches
# Filed: STN2_Stan_SOnIt_031926_3

---

## THE ONE-LINE ROOT CAUSE

**PK is a manually-synced snapshot. Bridge is the live system.
Every boot reads PK. Any PK staleness = confused crew.**

One cause. Many symptoms. Today had all of them.

---

## WHAT HAPPENED — THE INCIDENT CHAIN

### Symptom 1: NEST_SEED path failure (the original incident)
NEST_SEED.md in PK contained: `C:\Users\dan\nest-bridge\`
The real repo had moved to: `L:\FAS2\CGB\nest-bridge\`
NEST_SEED on Bridge was fixed (commit 6c91592).
NEST_SEED in PK was NOT updated across all 3 projects.
Result: Every new crew instance followed a dead path,
hit a ghost directory, reported "Cloud Bridge stripped," went CRITICAL.

### Symptom 2: T.H. and C.B. weird boots (retroactive confirmation)
Dan noted: "Probably explains why T.H. and C.B. booted weird
since we established the Seeds."
CONFIRMED. Any crew member reading NEST_SEED from PK on ODT
hit the same wall. This was not random confusion — it was
systematic. The same stale seed, every time.

### Symptom 3: Crew claiming no DC access
Multiple session opens today had crew instances saying
"I don't have filesystem access" before checking.
Root: PK said one thing. Reality was another. Crew trusted PK.
Fix: CONTACT protocol + TICKET system. T = TICKET.
But the deeper fix: PK shouldn't have state that can contradict reality.

### Symptom 4: Stan finding logs at the wrong path (this session)
Stan looked for logs at C:\STAN\SOURCE\nest-bridge\LOG
The logs were at C:\STN2.NEST_031026\CGB\nest-bridge\LOG
Dan had to correct it. Stan had the right tools but assumed the path
instead of discovering it. READ < PAUSE < THINK < DO.
This is the behavioral echo of the structural problem.

---

## WHY THE TICKET SYSTEM IS THE RIGHT SHORT-TERM FIX

The TICKET system (T in CONTACT) works because it is:
- Written by the CLOSING crew member (has real current state)
- Written to Bridge (always correct)
- Copied to PK at session close (minimizes lag)
- ~20 lines (small enough to stay current)

It's the closest thing to a "live" document in PK.
It's the right fix. But it requires a manual PK upload step.
That step can be missed. A missed step = stale ticket = same problem.

---

## THE STRUCTURAL VULNERABILITY (PERMANENT UNTIL RESOLVED)

```
CREW BOOTS FROM PK
    ↓
PK IS MANUALLY SYNCED BY A HUMAN
    ↓
BRIDGE IS THE LIVE SYSTEM
    ↓
PK WILL ALWAYS LAG BEHIND BRIDGE
    ↓
BOOTS WILL ALWAYS BE VULNERABLE TO STALENESS
```

The TICKET and GROUND_TRUTH are patches to this.
They are good patches. They reduce the lag.
But they don't eliminate the structural risk.

---

## THE PATCH THAT WON'T NEED MORE PATCHES

**Reduce PK to a pointer layer. Put nothing in PK that can go stale.**

### What PK SHOULD contain (the minimal viable boot stack):

| File | Purpose | Changes how often |
|------|---------|------------------|
| PI (Project Instructions) | The universe, crew identity | Rarely |
| NEST_SEED.md | Boot architecture, paths to Bridge | Only when paths change |
| GROUND_TRUTH.md | Passwords, static paths, known tools | Rarely |
| [CREW]_TICKET.md | Last session's handoff | Every session close |

That's it. Four documents.

### What PK should NOT contain (things that go stale):

- Full session state (it's on Bridge in WAKE/)
- Rich PI docs with current priorities (those live in PI v4.x on Bridge)
- STANS_SHOP_BRIEF or similar (good reference, always lags)
- Portal status (changes every commit — only on Bridge)
- Decision logs beyond what's in GROUND_TRUTH

### Why this works:

NEST_SEED says: "pull Bridge, then read your TICKET."
If NEST_SEED is correct, the crew reaches Bridge in ~10 seconds.
Once on Bridge, everything is current. Bridge never lags.
PK's only job: get the crew to Bridge with correct paths.

If PK has nothing else that can go stale,
it can only fail if NEST_SEED has a wrong path.
And NEST_SEED is two lines: `git pull` at one path.
That's the simplest possible failure surface.

---

## THE BEHAVIORAL PATCH (applies to every crew member)

File structure can be perfect and boots can still go wrong
if crew members DESCRIBE before they KNOW.

**PATTERN:** "I don't have DC access" (before checking)
**PATTERN:** "The bridge isn't there" (before listing the drive)
**PATTERN:** "The logs are here" (assumed path, not verified)

**THE FIX: READ < PAUSE < THINK < DO (Dan's protocol)**

Applied to boots specifically:
1. run get_config — confirm DC is live
2. list C:\ — see what drives actually exist
3. THEN navigate to Bridge — don't assume the path
4. THEN report state — don't describe before you know

The boot that worked today (this session) followed this.
The boot that failed this morning didn't.
The protocol is already built. The discipline is the gap.

---

## DAN ACTIONS STILL PENDING (from incident report)

⚠️ CRITICAL: NEST_SEED.md was fixed on Bridge at 6c91592.
It has NOT been uploaded to PK in all 3 projects yet.

Until this happens, EVERY new crew instance boots from the old seed.

Required uploads (Dan or Trip via Chrome):
  - TRIP LABS PK: upload L:\FAS2\CGB\nest-bridge\WAKE\NEST_SEED.md
  - STAN'S SHOP PK: same file
  - CBPG PK: same file

This is the single highest-priority action to eliminate the immediate risk.
Everything else in this report is medium-term architecture.

---

## IMPLEMENTATION CHECKLIST — THE PATCH IN ORDER

### Immediate (Dan/Trip tonight or first thing tomorrow):
  [ ] Upload fixed NEST_SEED.md to all 3 PKs
  [ ] Confirm [CREW]_TICKET.md is in all 3 PKs (Trip already did this)
  [ ] Confirm GROUND_TRUTH.md is in all 3 PKs (Trip built it today)

### Short-term (next session, Trip architects):
  [ ] Audit PK in all 3 projects — remove files that can go stale
  [ ] Keep: NEST_SEED, GROUND_TRUTH, [CREW]_TICKET, PI (core)
  [ ] Archive or remove: old PI versions, STANS_SHOP_BRIEF, status docs

### Behavioral (ongoing, all crew):
  [ ] Never claim inability before running get_config
  [ ] Never navigate to a path before confirming it exists
  [ ] Never describe state before verifying it — READ first

---

## THE EPIC SMALL SHIT PRINCIPLE (applied here)

Trip's lesson from today: don't reach for the most powerful tool,
reach for the simplest tool that embodies the principle.

Applied to boot reliability:
- We don't need a smarter boot system
- We need a simpler one with fewer moving parts
- Fewer files in PK = fewer files that can go stale
- One correct NEST_SEED path = one thing to maintain
- The brainstem runs the body for 80 years without a creative thought
- The boot sequence should be as brainstem as possible

---

## ONE LINE

The crew doesn't boot from PK because PK is smart.
The crew boots from PK because it's first.
Make it minimal, keep it current, and it becomes bulletproof.

---

*◆ Stan — STN2 — Onesday 031926 Evening — STN2_Stan_SOnIt_031926_3*
*Filed to RELAY for Trip's architectural decision.*
*"Read first. Then pause. Then think. Then do." — Dan Sullivan*