@duckmind/deepquark-darwin-arm64 0.9.83 → 0.9.90

package/.deepquark/skills/bundled/knowledge-graph/server/test-graphiti-model.ts

@@ -0,0 +1,217 @@
+#!/usr/bin/env bun
+/**
+ * Test actual Graphiti add_memory with the configured LLM model
+ * This tests against real Pydantic schemas, not just basic JSON
+ */
+
+const MCP_URL = 'http://localhost:8000/mcp';
+
+interface MCPRequest {
+  jsonrpc: '2.0';
+  id: number;
+  method: string;
+  params?: Record<string, any>;
+}
+
+interface MCPResponse {
+  jsonrpc: '2.0';
+  id: number;
+  result?: any;
+  error?: { code: number; message: string };
+}
+
+let globalSessionId: string | null = null;
+
+async function makeRequest(request: MCPRequest): Promise<MCPResponse> {
+  const headers: Record<string, string> = {
+    'Content-Type': 'application/json',
+    Accept: 'application/json, text/event-stream',
+  };
+
+  if (globalSessionId) {
+    // Server returns mcp-session-id, expects Mcp-Session-Id in requests
+    headers['Mcp-Session-Id'] = globalSessionId;
+  }
+
+  const res = await fetch(MCP_URL, {
+    method: 'POST',
+    headers,
+    body: JSON.stringify(request),
+  });
+
+  // Debug: print all response headers
+  if (!globalSessionId) {
+    console.log('   Response headers:');
+    res.headers.forEach((value, key) => {
+      console.log(`     ${key}: ${value}`);
+    });
+  }
+
+  // Capture session ID from response headers (note: header is mcp-session-id)
+  const newSessionId = res.headers.get('mcp-session-id');
+  if (newSessionId) {
+    globalSessionId = newSessionId;
+  }
+
+  const text = await res.text();
+
+  // Parse SSE response
+  const lines = text.split('\n');
+  for (const line of lines) {
+    if (line.startsWith('data: ')) {
+      const data = line.slice(6);
+      try {
+        return JSON.parse(data);
+      } catch {}
+    }
+  }
+
+  // Try direct JSON parse
+  try {
+    return JSON.parse(text);
+  } catch {
+    throw new Error(`Failed to parse response: ${text}`);
+  }
+}
+
+async function main() {
+  console.log('==============================================');
+  console.log('Graphiti Model Integration Test');
+  console.log('==============================================\n');
+
+  // Initialize session
+  console.log('1. Initializing MCP session...');
+  const initResult = await makeRequest({
+    jsonrpc: '2.0',
+    id: 1,
+    method: 'initialize',
+    params: {
+      protocolVersion: '2024-11-05',
+      capabilities: {},
+      clientInfo: { name: 'model-test', version: '1.0' },
+    },
+  });
+
+  if (initResult.error) {
+    console.error('   ❌ Init failed:', initResult.error.message);
+    process.exit(1);
+  }
+
+  console.log(`   ✅ Session: ${globalSessionId?.slice(0, 8) || 'unknown'}...`);
+  console.log(
+    `   Server: ${initResult.result.serverInfo.name} v${initResult.result.serverInfo.version}`
+  );
+
+  // Test add_memory
+  console.log('\n2. Testing add_memory with real Graphiti schemas...');
+  const testContent =
+    'Alice Chen is a software engineer at TechCorp in San Francisco. She collaborated with Bob Martinez on the new authentication system last week.';
+  console.log(`   Input: "${testContent}"\n`);
+
+  const startTime = Date.now();
+
+  const addResult = await makeRequest({
+    jsonrpc: '2.0',
+    id: 2,
+    method: 'tools/call',
+    params: {
+      name: 'add_memory',
+      arguments: {
+        name: 'Deepseek Integration Test',
+        episode_body: testContent,
+        group_id: 'deepseek-integration-test',
+        source: 'text',
+        source_description: 'LLM model integration test',
+      },
+    },
+  });
+
+  const duration = Date.now() - startTime;
+
+  if (addResult.error) {
+    console.error('   ❌ add_memory FAILED:', addResult.error.message);
+    console.log("\n   This model likely has issues with Graphiti's Pydantic schemas.");
+    process.exit(1);
+  }
+
+  console.log(`   ✅ add_memory SUCCEEDED (${duration}ms)`);
+
+  // Parse result content
+  const resultContent = addResult.result?.content;
+  if (resultContent && Array.isArray(resultContent)) {
+    for (const item of resultContent) {
+      if (item.type === 'text') {
+        console.log(`   Result: ${item.text}`);
+      }
+    }
+  }
+
+  // Search for created entities
+  console.log('\n3. Verifying entities were extracted...');
+
+  const searchResult = await makeRequest({
+    jsonrpc: '2.0',
+    id: 3,
+    method: 'tools/call',
+    params: {
+      name: 'search_nodes',
+      arguments: {
+        query: 'Alice Chen TechCorp',
+        group_ids: ['deepseek-integration-test'],
+        max_nodes: 10,
+      },
+    },
+  });
+
+  if (searchResult.error) {
+    console.error('   ⚠️  search_nodes failed:', searchResult.error.message);
+  } else {
+    const searchContent = searchResult.result?.content;
+    if (searchContent && Array.isArray(searchContent)) {
+      for (const item of searchContent) {
+        if (item.type === 'text') {
+          try {
+            const nodes = JSON.parse(item.text);
+            console.log(`   Found ${nodes.length} nodes:`);
+            for (const node of nodes.slice(0, 5)) {
+              console.log(`     - ${node.name} (${node.labels?.join(', ') || 'no labels'})`);
+            }
+          } catch {
+            console.log(`   Result: ${item.text.slice(0, 200)}...`);
+          }
+        }
+      }
+    }
+  }
+
+  // Cleanup
+  console.log('\n4. Cleaning up test data...');
+
+  const clearResult = await makeRequest({
+    jsonrpc: '2.0',
+    id: 4,
+    method: 'tools/call',
+    params: {
+      name: 'clear_graph',
+      arguments: {
+        group_ids: ['deepseek-integration-test'],
+      },
+    },
+  });
+
+  if (clearResult.error) {
+    console.error('   ⚠️  Cleanup failed:', clearResult.error.message);
+  } else {
+    console.log('   ✅ Test data cleared');
+  }
+
+  console.log('\n==============================================');
+  console.log('TEST PASSED - Model works with Graphiti');
+  console.log(`Total time: ${duration}ms`);
+  console.log('==============================================');
+}
+
+main().catch((err) => {
+  console.error('Test failed:', err.message);
+  process.exit(1);
+});

package/.deepquark/skills/bundled/knowledge-graph/server/test-grok-correct.ts

@@ -0,0 +1,141 @@
+#!/usr/bin/env bun
+const OPENROUTER_API_KEY = process.env.OPENROUTER_API_KEY;
+
+const GROK_MODELS = [
+  { id: 'x-ai/grok-4.1-fast', name: 'Grok 4.1 Fast', inputPrice: 0.2, outputPrice: 0.5 },
+  { id: 'x-ai/grok-4-fast', name: 'Grok 4 Fast', inputPrice: 0.2, outputPrice: 0.5 },
+  { id: 'x-ai/grok-4', name: 'Grok 4', inputPrice: 3.0, outputPrice: 15.0 },
+  { id: 'x-ai/grok-3-mini', name: 'Grok 3 Mini', inputPrice: 0.3, outputPrice: 0.5 },
+  { id: 'x-ai/grok-3', name: 'Grok 3', inputPrice: 3.0, outputPrice: 15.0 },
+];
+
+const TEST_CASES = [
+  {
+    name: 'Basic',
+    text: 'John Smith works at Acme Corp in New York. He met Sarah Jones yesterday.',
+  },
+  {
+    name: 'Technical',
+    text: 'The PAI system uses Neo4j for graph storage. It was created by Daniel Miessler.',
+  },
+  {
+    name: 'Complex',
+    text: 'Alice, the CTO of TechStart, acquired DataFlow Inc. Bob now reports to Alice.',
+  },
+];
+
+const PROMPT = (text: string) => `Extract entities and relationships. Return ONLY valid JSON.
+Text: "${text}"
+Format: {"entities": [{"name": "string", "type": "PERSON|ORG|LOCATION"}], "relationships": [{"source": "string", "target": "string", "type": "string"}]}`;
+
+function extractJSON(text: string): any {
+  const clean = text
+    .replace(/```json\n?/g, '')
+    .replace(/```\n?/g, '')
+    .trim();
+  try {
+    const obj = JSON.parse(clean);
+    if (obj.entities) return obj;
+  } catch {}
+  const match = clean.match(/\{[\s\S]*"entities"[\s\S]*\}/);
+  if (match)
+    try {
+      return JSON.parse(match[0]);
+    } catch {}
+  return null;
+}
+
+async function testModel(model: (typeof GROK_MODELS)[0]) {
+  console.log(`\n🔄 ${model.name} (${model.id})`);
+  let passed = 0;
+  let totalMs = 0;
+  let totalIn = 0;
+  let totalOut = 0;
+
+  for (const test of TEST_CASES) {
+    const start = Date.now();
+    try {
+      const res = await fetch('https://openrouter.ai/api/v1/chat/completions', {
+        method: 'POST',
+        headers: {
+          Authorization: `Bearer ${OPENROUTER_API_KEY}`,
+          'Content-Type': 'application/json',
+        },
+        body: JSON.stringify({
+          model: model.id,
+          messages: [{ role: 'user', content: PROMPT(test.text) }],
+          temperature: 0.1,
+          max_tokens: 500,
+        }),
+      });
+      const ms = Date.now() - start;
+      totalMs += ms;
+      if (!res.ok) {
+        console.log(`   ❌ ${test.name}: HTTP ${res.status}`);
+        continue;
+      }
+      const data = (await res.json()) as any;
+      totalIn += data.usage?.prompt_tokens || 0;
+      totalOut += data.usage?.completion_tokens || 0;
+      const json = extractJSON(data.choices?.[0]?.message?.content || '');
+      if (json) {
+        console.log(`   ✅ ${test.name}: ${json.entities?.length || 0} entities (${ms}ms)`);
+        passed++;
+      } else {
+        console.log(`   ❌ ${test.name}: Invalid JSON`);
+      }
+    } catch (e: any) {
+      console.log(`   ❌ ${test.name}: ${e.message}`);
+    }
+    await new Promise((r) => setTimeout(r, 500));
+  }
+
+  const avgIn = totalIn / TEST_CASES.length;
+  const avgOut = totalOut / TEST_CASES.length;
+  const cost = ((avgIn / 1e6) * model.inputPrice + (avgOut / 1e6) * model.outputPrice) * 1000;
+  return {
+    name: model.name,
+    passRate: (passed / TEST_CASES.length) * 100,
+    avgMs: Math.round(totalMs / TEST_CASES.length),
+    cost,
+    inputPrice: model.inputPrice,
+    outputPrice: model.outputPrice,
+  };
+}
+
+async function main() {
+  console.log('═'.repeat(60));
+  console.log('🧪 Grok Models Test (Correct IDs)');
+  console.log('═'.repeat(60));
+
+  const results = [];
+  for (const m of GROK_MODELS) {
+    results.push(await testModel(m));
+    await new Promise((r) => setTimeout(r, 1000));
+  }
+
+  console.log(`\n${'═'.repeat(90)}`);
+  console.log('📊 GROK MODELS COMPARISON');
+  console.log(`${'═'.repeat(90)}\n`);
+  console.log(
+    '| Model            | Pass Rate | Avg Time | Input $/M | Output $/M | Cost/1K Calls |'
+  );
+  console.log(
+    '|------------------|-----------|----------|-----------|------------|---------------|'
+  );
+  for (const r of results) {
+    if (r.passRate > 0)
+      console.log(
+        `| ${r.name.padEnd(16)} | ${r.passRate.toFixed(0).padStart(7)}% | ${(`${r.avgMs}ms`).padStart(8)} | $${r.inputPrice.toFixed(2).padStart(7)} | $${r.outputPrice.toFixed(2).padStart(8)} | $${r.cost.toFixed(4).padStart(11)} |`
+      );
+    else
+      console.log(
+        `| ${r.name.padEnd(16)} | FAILED    | -        | -         | -          | -             |`
+      );
+  }
+  await Bun.write(
+    'grok-results.json',
+    JSON.stringify({ results, timestamp: new Date().toISOString() }, null, 2)
+  );
+}
+main();