#!/usr/bin/env python3 """Test the newly added MONIKA capabilities.""" import asyncio import json from pathlib import Path from mcp import ClientSession, StdioServerParameters from mcp.client.stdio import stdio_client async def test_new_capabilities(): """Test all newly wired capabilities.""" server_script = Path(__file__).parent / "start_mcp_server.py" server_params = StdioServerParameters( command="python", args=[str(server_script)], env=None, ) async with stdio_client(server_params) as (read, write): async with ClientSession(read, write) as session: await session.initialize() print("=" * 70) print("TESTING NEW MONIKA CAPABILITIES") print("=" * 70) # List tools to verify new ones are present tools_result = await session.list_tools() tools = {tool.name: tool for tool in tools_result.tools} print(f"\nTotal tools: {len(tools)}") new_tools = [ "scratchpad_read", "scratchpad_history", "scratchpad_4d_path", "sass_query", "meta_state_report", "verification_suite_run", "action_scores_detailed", ] print("\nNew tools status:") for tool in new_tools: status = "✓" if tool in tools else "✗" print(f" {status} {tool}") # Test 1: Execute a runtime step to populate scratchpad print("\n" + "=" * 70) print("TEST 1: Runtime Step + Scratchpad Population") print("=" * 70) result = await session.call_tool("runtime_step", { "text": "Let's reason through a complex problem involving recursive algorithms and memory optimization." }) step = json.loads(result.content[0].text) print(f"✓ Step {step['step']} executed") print(f" Budget left: {step['budget_left']}") # Test 2: Read scratchpad print("\n" + "=" * 70) print("TEST 2: Scratchpad Read") print("=" * 70) result = await session.call_tool("scratchpad_read", {}) scratch = json.loads(result.content[0].text) print(f"✓ Current trace steps: {len(scratch['current_trace'])}") print(f" Token usage: {scratch['current_tokens']}/{scratch['max_tokens']}") if scratch['current_trace']: print(f" Last thought: {scratch['current_trace'][-1][:80]}...") # Test 3: Scratchpad history print("\n" + "=" * 70) print("TEST 3: Scratchpad History") print("=" * 70) result = await session.call_tool("scratchpad_history", {"max_traces": 3}) history = json.loads(result.content[0].text) print(f"✓ Retrieved {len(history['traces'])} traces") print(f" Rolling token mean: {history['rolling_token_mean']:.1f}") for i, trace in enumerate(history['traces']): print(f" Trace {i+1}: {trace['outcome']} - {trace['summary']}") # Test 4: 4D Path visualization print("\n" + "=" * 70) print("TEST 4: 4D Reasoning Path") print("=" * 70) result = await session.call_tool("scratchpad_4d_path", {}) path_data = json.loads(result.content[0].text) if path_data.get('path'): print(f"✓ 4D Path found!") print(f" Summary: {path_data['summary']}") print("\n ASCII Visualization:") print(path_data['ascii_viz']) # Show path details path = path_data['path'] if 'points' in path and path['points']: print(f"\n Path coordinates ({len(path['points'])} points):") for i, pt in enumerate(path['points'][:3]): print(f" {i}: x={pt['x']:.2f} y={pt['y']:.2f} z={pt['z']:.2f} w={pt['w']:.2f}") print(f" Centroid: {path['centroid']}") else: print(" No 4D path yet:", path_data.get('message')) # Test 5: SASS Query print("\n" + "=" * 70) print("TEST 5: SASS Semantic Search") print("=" * 70) result = await session.call_tool("sass_query", { "query": "algorithms recursive optimization" }) sass_result = json.loads(result.content[0].text) print(f"✓ SASS query: '{sass_result['query']}'") print(f" Results: {len(sass_result['results'])}") for i, res in enumerate(sass_result['results'][:3]): print(f" {i+1}. [score={res['score']:.3f}] {res['text'][:80]}...") # Test 6: Meta State Report print("\n" + "=" * 70) print("TEST 6: Meta State Self-Report") print("=" * 70) result = await session.call_tool("meta_state_report", {}) meta = json.loads(result.content[0].text) print(f"✓ Meta state:") print(f" Confidence: {meta['confidence']:.3f}") print(f" ROI: {meta['roi']:.3f}") print(f" History length: {meta['history_length']}") print(f" Snapshot keys: {list(meta['snapshot'].keys())}") # Test 7: Verification Suite print("\n" + "=" * 70) print("TEST 7: Verification Suite") print("=" * 70) result = await session.call_tool("verification_suite_run", { "context": "Testing MONIKA verification systems" }) verify = json.loads(result.content[0].text) print(f"✓ Verification outcomes: {len(verify['outcomes'])}") passed = sum(1 for o in verify['outcomes'] if o['passed']) print(f" Passed: {passed}/{len(verify['outcomes'])}") for i, outcome in enumerate(verify['outcomes'][:3]): status = "PASS" if outcome['passed'] else "FAIL" print(f" {i+1}. [{status}] {outcome['evidence'][:60]}...") # Test 8: Detailed Action Scores print("\n" + "=" * 70) print("TEST 8: Detailed Action Scores & Salience") print("=" * 70) result = await session.call_tool("action_scores_detailed", {}) actions = json.loads(result.content[0].text) print(f"✓ Action candidates: {len(actions['scores'])}") if actions['scores']: print(" Top 5 actions:") for i, action in enumerate(actions['scores'][:5]): act = action['action'] print(f" {i+1}. {act['operator']} (depth={act['depth']}, patch={act['patch']})") print(f" Score: {action['score']:.3f}") print(f" Rationale: {action['rationale'][:60]}...") else: print(" No action scores yet (controller not initialized)") # Show top salience signals salience = actions.get('salience_vector', {}) if salience: sorted_sal = sorted(salience.items(), key=lambda x: abs(x[1]), reverse=True) print(f"\n Top 5 salience signals:") for key, val in sorted_sal[:5]: print(f" {key}: {val:.4f}") # Final summary print("\n" + "=" * 70) print("CAPABILITY TEST COMPLETE") print("=" * 70) working = sum(1 for tool in new_tools if tool in tools) print(f"✓ {working}/{len(new_tools)} new capabilities operational") print("\nMONIKA's internal state is now fully introspectable!") print("You can now:") print(" • View 4D reasoning paths in real-time") print(" • Execute SASS semantic queries") print(" • Monitor scratchpad working memory") print(" • Read meta-state reports") print(" • Run verification suites") print(" • Inspect detailed action scoring") if __name__ == "__main__": asyncio.run(test_new_capabilities())