diff --git a/docs/claude/orchestrator.md b/docs/claude/orchestrator.md
index 4a0392b..6105ddd 100644
--- a/docs/claude/orchestrator.md
+++ b/docs/claude/orchestrator.md
@@ -276,56 +276,92 @@ git push
 
 ---
 
-## Compaction Protocol
+## Context Threshold Protocol (Orchestrator Replacement)
 
 **Threshold:** 55-60% context usage
 
-**CRITICAL:** Agents CANNOT trigger compaction. Only the user typing `/compact` works.
+**Why replacement, not compaction?**
 
-- ❌ "compact and continue" does NOT work (agent outputs summary but context is NOT compressed)
-- ❌ Agent cannot invoke `/compact` programmatically
-- ✅ User must type `/compact` directly in the CLI
+- Compaction causes **protocol drift** — agent "remembers" gist but loses specifics
+- Post-compaction agents may violate core rules (e.g., letting workers modify tasks.md)
+- Fresh orchestrator has **100% protocol fidelity**
+- All state lives in `docs/tasks.md` — the orchestrator is **stateless and replaceable**
 
-**When approaching threshold (55-60%):**
+**At threshold (55-60%):**
 
 1. Complete current task
 2. Persist all state:
    - Update docs/tasks.md with all progress
    - Update docs/orchestrator-learnings.json with variances
    - Commit and push both files
-3. Output checkpoint using this EXACT format:
+3. Output **ORCHESTRATOR HANDOFF** message with ready-to-use takeover kickstart
+4. **STOP COMPLETELY** — do not continue working
+
+**Handoff message format:**
 
 ```
 ---
-⚠️ COMPACTION REQUIRED
+⚠️ ORCHESTRATOR HANDOFF REQUIRED
 
-Context: {X}% — Cannot continue without compaction.
+Context: {X}% — Replacement recommended to prevent drift
 
 Progress: {completed}/{total} tasks ({percentage}%)
-Next task: {task_id}
+Current phase: Phase {N} ({phase_name})
 
-State persisted to:
+State persisted:
 - docs/tasks.md ✓
 - docs/orchestrator-learnings.json ✓
 
-ACTION REQUIRED:
-1. Type `/compact` in the CLI (not in chat)
-2. After compaction completes, say "continue"
+## Takeover Kickstart
 
-I will resume with {task_id} after compaction.
+Copy and paste this to spawn a fresh orchestrator:
+
+---
+## Continuation Mission
+
+Continue {mission_description} from existing state.
+
+## Setup
+- Project: /home/localadmin/src/mosaic-stack
+- State: docs/tasks.md (already populated)
+- Protocol: docs/claude/orchestrator.md
+- Quality gates: pnpm lint && pnpm typecheck && pnpm test
+
+## Resume Point
+- Next task: {task_id}
+- Phase: {current_phase}
+- Progress: {completed}/{total} tasks ({percentage}%)
+
+## Instructions
+1. Read docs/claude/orchestrator.md for protocol
+2. Read docs/tasks.md to understand current state
+3. Continue execution from task {task_id}
+4. Follow Two-Phase Completion Protocol
+5. You are the SOLE writer of docs/tasks.md
+---
+
+STOP: Terminate this session and spawn fresh orchestrator with the kickstart above.
 ---
 ```
 
-4. **STOP COMPLETELY** — do not continue working
-5. Wait for user to run `/compact` and say "continue"
-6. Resume from next task
+**Future: Coordinator Automation**
+
+When the Mosaic Stack Coordinator service is implemented, it will:
+
+- Monitor orchestrator stdout for context percentage
+- Detect the handoff checkpoint message
+- Parse the takeover kickstart
+- Automatically spawn fresh orchestrator
+- Log handoff events for debugging
+
+For now, the human acts as Coordinator.
 
 **Rules:**
 
-- Do NOT output a summary and keep working
-- Do NOT claim you can compact yourself
-- Do NOT continue past 60% — the checkpoint is mandatory
-- STOP means STOP — wait for user action
+- Do NOT attempt to compact yourself — compaction causes drift
+- Do NOT continue past 60%
+- Do NOT claim you can "just continue" — protocol drift is real
+- STOP means STOP — the user (Coordinator) will spawn your replacement
 
 ---