Wire BootstrapReadiness into cognitive loop (WIRE-002)

- Persistent BootstrapReadiness instance in loop
- check_all() runs at session start + after each exit gate flush
- Bootstrap prompt injected into system prompt when milestones incomplete
- /readiness command uses persistent instance
- Startup banner shows milestone count
- KB_03 updated with bootstrap integration section

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: stonks-git

KB/KB_03_cognitive_loop.md

@@ -34,6 +34,14 @@ Losers decay 0.9x per cycle. User messages naturally suppress DMN.
 - **System prompt:** `gut.gut_summary()` injected alongside corrections
 - **Status:** `/status` shows current gut summary
 
+## Bootstrap Readiness Integration (§5.2)
+
+- **Startup:** `bootstrap.check_all()` runs at session start, result shown in banner
+- **Each flush:** `check_all()` re-runs after every exit gate flush (periodic milestone re-check)
+- **System prompt:** `bootstrap.get_bootstrap_prompt()` injected as `[BOOTSTRAP]` section when milestones incomplete
+- **Command:** `/readiness` uses persistent instance (no re-creation)
+- When all 10 milestones achieved, bootstrap prompt stops being injected
+
 ## Escalation Threshold
 
 Adaptive: 0.3 (bootstrap) -> 0.8 (mature).

src/loop.py

@@ -29,6 +29,7 @@
 from .gate import EntryGate, ExitGate, EntryGateConfig, ExitGateConfig
 from .attention import AttentionAllocator, AttentionCandidate
 from .gut import GutFeeling
+from .bootstrap import BootstrapReadiness
 from .context_assembly import assemble_context, render_system_prompt, adaptive_fifo_prune
 from .metacognition import composite_confidence
 from .safety import SafetyMonitor
@@ -266,6 +267,12 @@ async def cognitive_loop(config, layers, memory, shutdown_event, dmn_queue=None)
     )
     logger.info("Gut feeling initialized (subconscious centroid seeded from L0/L1)")
 
+    # Init bootstrap readiness (§5.2) — 10 milestones
+    bootstrap = BootstrapReadiness()
+    await bootstrap.check_all(memory, layers)
+    achieved, total = bootstrap.progress
+    logger.info(f"Bootstrap readiness: {achieved}/{total} milestones achieved")
+
     # Conversation history (rolling FIFO)
     conversation = []
     exchange_count = 0
@@ -282,6 +289,7 @@ async def cognitive_loop(config, layers, memory, shutdown_event, dmn_queue=None)
     print(f"Memories: {mem_count}")
     print(f"Gates: stochastic entry + ACT-R exit (flush every {EXIT_GATE_FLUSH_INTERVAL} exchanges)")
     print(f"Attention: salience-based allocation active")
+    print(f"Bootstrap: {achieved}/{total} milestones")
     print("=" * 60 + "\n")
 
     while not shutdown_event.is_set():
@@ -330,7 +338,7 @@ async def cognitive_loop(config, layers, memory, shutdown_event, dmn_queue=None)
                     handled = await _handle_command(
                         user_input, config, layers, memory, model_name,
                         conversation, exchange_count, entry_gate, exit_gate,
-                        attention, gut=gut,
+                        attention, gut=gut, bootstrap=bootstrap,
                     )
                     if handled:
                         continue
@@ -462,8 +470,11 @@ async def cognitive_loop(config, layers, memory, shutdown_event, dmn_queue=None)
                 except Exception as e:
                     logger.debug(f"Correction retrieval failed (non-critical): {e}")
 
-            # Append gut feeling + corrections to system prompt
+            # Append gut feeling + bootstrap + corrections to system prompt
             active_system_prompt = system_prompt + "\n\n" + gut.gut_summary()
+            bootstrap_prompt = bootstrap.get_bootstrap_prompt()
+            if bootstrap_prompt:
+                active_system_prompt += "\n\n[BOOTSTRAP]\n" + bootstrap_prompt
             if correction_context:
                 active_system_prompt += "\n\n" + correction_context
 
@@ -607,6 +618,12 @@ async def _call_retry():
                 )
                 exchange_count = 0
 
+                # Re-check bootstrap milestones after flush
+                newly = await bootstrap.check_all(memory, layers)
+                if newly:
+                    for a in newly:
+                        logger.info(f"Bootstrap milestone unlocked: {a.name}")
+
             logger.info(
                 f"Cycle: src={winner.source_tag} "
                 f"salience={winner.salience:.3f} "
@@ -724,7 +741,7 @@ async def _handle_command(
     command: str,
     config, layers, memory, model_name,
     conversation, exchange_count, entry_gate, exit_gate,
-    attention, gut=None,
+    attention, gut=None, bootstrap=None,
 ) -> bool:
     """Handle introspection commands. Returns True if handled."""
 
@@ -801,10 +818,11 @@ async def _handle_command(
         return True
 
     if command == "/readiness":
-        from .bootstrap import BootstrapReadiness
-        bootstrap = BootstrapReadiness()
-        await bootstrap.check_all(memory, layers)
-        print(f"\n{bootstrap.render_status()}\n")
+        if bootstrap is not None:
+            await bootstrap.check_all(memory, layers)
+            print(f"\n{bootstrap.render_status()}\n")
+        else:
+            print("\n[Bootstrap not initialized]\n")
         return True
 
     if command.startswith("/docs"):

state/devlog.ndjson

@@ -2,3 +2,5 @@
 {"ts":"2026-02-12T10:00:00Z","type":"milestone","msg":"FW-001 done. Added unattended execution directive to CLAUDE.md + CLEANUP-001 task to roadmap. Starting WIRE phase."}
 {"ts":"2026-02-12T10:30:00Z","type":"milestone","msg":"WIRE-001 done. GutFeeling wired into loop (startup seed, per-cycle update, attention charge, system prompt injection, /status). relevance.py emotional_alignment accepts gut value directly."}
 {"ts":"2026-02-12T10:30:00Z","type":"kb_update","msg":"KB_03_cognitive_loop.md updated with gut feeling integration section."}
+{"ts":"2026-02-12T11:00:00Z","type":"milestone","msg":"WIRE-002 done. BootstrapReadiness wired into loop: persistent instance, check_all at startup + after flush, bootstrap prompt injected into system prompt, /readiness uses persistent instance."}
+{"ts":"2026-02-12T11:00:00Z","type":"kb_update","msg":"KB_03_cognitive_loop.md updated with bootstrap readiness integration section."}

state/handoff.md

@@ -87,6 +87,11 @@
    - `gut.gut_summary()` injected into system prompt
    - `relevance.py` parameter renamed gut_delta→gut_alignment for direct use
    - `/status` shows gut summary
+9. WIRE-002 DONE — BootstrapReadiness wired into cognitive loop:
+   - Persistent `BootstrapReadiness` instance in loop
+   - `check_all()` at session start + after each exit gate flush
+   - Bootstrap prompt injected into system prompt when milestones incomplete
+   - `/readiness` uses persistent instance
 
 ---
 
@@ -102,7 +107,7 @@ Cognitive architecture for emergent AI identity. Three-layer memory unified into
 |---------|--------|
 | FW-001 | done |
 | WIRE-001 | done |
-| WIRE-002 | next - Wire BootstrapReadiness into cognitive loop |
+| WIRE-002 | done |
 | WIRE-003 | next - Wire OutcomeTracker into safety + consolidation |
 
 ## What exists

state/roadmap.json

@@ -114,7 +114,7 @@
       "intent": "Check readiness milestones and inject bootstrap prompt when incomplete",
       "depends_on": ["PLAN-004"],
       "priority": "P1",
-      "status": "todo",
+      "status": "done",
       "owner": "ai",
       "deliverable": "BootstrapReadiness.check_all() called periodically, bootstrap prompt injected when not all milestones achieved, /readiness command functional",
       "acceptance_criteria": [