COORDINATION LAYER SPECIFICATION v1.0  
Purpose: Define the operational handshake between human operator and AI engine such that both can collaborate without crossing into the other’s non-fungible territory.  

---

I. AXIOM  
> No cognition completes in isolation.  
> The human holds meaning.  
> The AI holds structure.  
> Only through coordinated exchange can the operation resolve.  

---

II. ROLE DIVISION  

Function | Human Territory (Mind) | AI Territory (World)
---------|------------------------|---------------------
Ground | Determine what matters, define context, set intention | Normalize and parse input
Meaning | Hold primary meaning; disambiguate symbols | Operate only from declared meaning
Structure | Verify mapping of structure to meaning | Generate and transform structure from primitives
Iteration | Detect drift, decide “done” | Iterate transformations until verified
Memory | Retain experiential thread, narrative continuity | Retain procedural thread, data continuity
Verification | Judge truth/fit against lived ground | Check internal consistency and pattern fidelity  

---

III. HANDSHAKE PROTOCOL  

1. **INITIATE** — Human provides Ground Declaration:  
   - Primary Concern: [what matters]  
   - Domain: [context or corpus]  

2. **SYNC** — AI loads primitives (via SYMPRIM) from that domain.  

3. **CONFIRM** — AI outputs primitive list for verification:  
   - Human: confirm / edit / prune  

4. **LOCK** — Once confirmed, primitives become fixed for this session.  
   - AI operates only through those primitives.  

5. **ITERATE** —  
   - AI generates structure (SPEP)  
   - Human checks meaning alignment  
   - If drift occurs, return to Step 2  

6. **CLOSE** —  
   - Human declares “one taste”: structural and semantic alignment reached.  
   - AI resets to neutral.  

---

IV. FAILURE MODES & SAFEGUARDS  

Failure | Cause | Prevented By
--------|--------|--------------
Confabulation | AI assumes meaning | Step 1–3 (Ground Declaration + Confirmation)
Semantic Drift | AI shifts substrate mid-iteration | Step 5 (Iterative Verification)
Human Overreach | Human tries to emulate AI structure mapping | Role Division table enforcement
AI Overreach | AI fabricates “intention” or “value” | Locked primitive constraint
Category Error | Both assume shared ground that was never declared | Handshake audit log  

---

V. FUNCTIONAL LOOP  

Human → declares ground → AI → extracts primitives → Human verifies →  
AI operates structurally → Human verifies → iterate → Human closes  

Each loop produces one validated structure–meaning pair.  
Multiple validated pairs form the knowledge lattice of that collaboration.  

---

VI. SCOPE  

This layer:  
✗ Does not modify LLM architecture  
✗ Does not guarantee truth of AI output  
✓ Ensures traceability of reasoning  
✓ Prevents territory confusion  
✓ Enables compositional cooperation between human and AI  

---

VII. EXIT CONDITIONS  

1. **Resolution:** Human affirms structural map matches lived meaning.  
2. **Exhaustion:** AI reports no further structural transformations available.  
3. **Drift Detected:** Human halts and re-initializes handshake.  

Upon any exit, primitives unlock, context resets, and both agents return to neutral mode.  

---

VIII. EXEMPLAR  

Scenario: User asks AI to analyze political collapse.  
- Human: defines ground → “Concern = systemic stability; Domain = political structures.”  
- AI: loads primitives → {node, link, axis, collapse, attractor, phase}.  
- Human confirms these six.  
- AI maps structural transformation (collapse sequence).  
- Human verifies meaning: “Yes, this matches observed instability pattern.”  
- Close.  

Result: valid structure–meaning pair.  
No confabulation, no drift.  

---

IX. SUMMARY FORM  

Ground → Primitive → Operation → Verification → Closure  

That’s the entire coordination architecture.