Agent-CoreX Documentation

Real-World Sample Queries

This page shows actual queries users make and what tools Agent-CoreX returns.

Query 1: Simple GitHub Workflow

User Query

"Create a pull request on GitHub"

What Agent-CoreX Returns

[
  {
    "name": "create-pull-request",
    "server": "github-mcp",
    "score": 0.99,
    "description": "Create a new PR with title, body, and branch info"
  },
  {
    "name": "create-github-issue",
    "server": "github-mcp",
    "score": 0.78,
    "description": "Create an issue (similar intent, different action)"
  },
  {
    "name": "create-jira-issue",
    "server": "jira-mcp",
    "score": 0.52,
    "description": "Create Jira issue (different platform)"
  }
]

Implementation

const tools = await agent.retrieveTools({
  query: "Create a pull request on GitHub"
});

const prTool = tools[0];

const result = await agent.executeTool({
  toolName: prTool.name,
  params: {
    repository: "owner/repo",
    title: "Add new feature",
    from_branch: "feature/new",
    to_branch: "main",
    body: "This PR adds a new feature"
  }
});

console.log(`✅ PR created: ${result.pr_url}`);

Query 2: Multi-Tool Workflow

User Query

"I need to deploy my Node.js app to AWS, run tests first, and notify the team"

What Agent-CoreX Returns

[
  {
    "name": "run-test-suite",
    "score": 0.94,
    "server": "ci-cd-mcp"
  },
  {
    "name": "deploy-aws-lambda",
    "score": 0.92,
    "server": "aws-mcp"
  },
  {
    "name": "deploy-aws-ecs",
    "score": 0.91,
    "server": "aws-mcp"
  },
  {
    "name": "send-slack-message",
    "score": 0.88,
    "server": "slack-mcp"
  },
  {
    "name": "send-email",
    "score": 0.72,
    "server": "email-mcp"
  }
]

Implementation

const workflow = async () => {
  const tools = await agent.retrieveTools({
    query: "Deploy Node.js app to AWS, run tests, notify team",
    topK: 10
  });

  // Step 1: Run tests
  const testResult = await agent.executeTool({
    toolName: 'run-test-suite',
    params: {
      project: 'my-app',
      framework: 'jest',
      coverage: true
    }
  });

  if (testResult.failed > 0) {
    await agent.executeTool({
      toolName: 'send-slack-message',
      params: {
        channel: '#deployments',
        message: `❌ Tests failed: ${testResult.failed} failures`
      }
    });
    return;
  }

  // Step 2: Deploy to AWS
  const deployResult = await agent.executeTool({
    toolName: 'deploy-aws-lambda',
    params: {
      function: 'my-app',
      zip: 'build/function.zip',
      env: 'production'
    }
  });

  // Step 3: Notify team
  await agent.executeTool({
    toolName: 'send-slack-message',
    params: {
      channel: '#deployments',
      message: `✅ Deployment successful!\n${deployResult.url}`
    }
  });
};

await workflow();

Query 3: Incident Response

User Query

"Get error logs from the API, check if there's a pattern, and restart the service if the CPU is too high"

Tools Returned

[
  {
    "name": "query-logs",
    "score": 0.96,
    "server": "monitoring-mcp"
  },
  {
    "name": "query-metrics",
    "score": 0.93,
    "server": "monitoring-mcp"
  },
  {
    "name": "restart-service",
    "score": 0.91,
    "server": "infrastructure-mcp"
  },
  {
    "name": "create-incident-ticket",
    "score": 0.87,
    "server": "jira-mcp"
  },
  {
    "name": "trigger-alert",
    "score": 0.76,
    "server": "monitoring-mcp"
  }
]

Implementation

const handleIncident = async () => {
  // Get logs
  const logs = await agent.executeTool({
    toolName: 'query-logs',
    params: {
      service: 'api',
      severity: 'error',
      timeRange: '-30m',
      limit: 100
    }
  });

  // Analyze for patterns
  const patterns = analyzeErrorLogs(logs);
  console.log('🔍 Error patterns found:', patterns);

  // Check metrics
  const metrics = await agent.executeTool({
    toolName: 'query-metrics',
    params: {
      service: 'api',
      metric: ['cpu', 'memory', 'latency'],
      timeRange: '-30m'
    }
  });

  // Take action if needed
  if (metrics.cpu > 85) {
    console.log('⚠️ High CPU detected');

    const restartResult = await agent.executeTool({
      toolName: 'restart-service',
      params: {
        service: 'api',
        replicas: 3,
        graceful: true
      }
    });

    // Create ticket
    await agent.executeTool({
      toolName: 'create-incident-ticket',
      params: {
        project: 'OPS',
        title: `Auto-remediated high CPU incident`,
        description: `Patterns: ${JSON.stringify(patterns)}`,
        priority: 'High'
      }
    });

    console.log('✅ Service restarted and incident tracked');
  }
};

Query 4: Infrastructure as Code

User Query

"Apply Terraform changes, validate they didn't break anything, and update our deployment documentation"

Tools Returned

[
  {
    "name": "terraform-apply",
    "score": 0.97,
    "server": "terraform-mcp"
  },
  {
    "name": "terraform-validate",
    "score": 0.95,
    "server": "terraform-mcp"
  },
  {
    "name": "run-smoke-tests",
    "score": 0.92,
    "server": "testing-mcp"
  },
  {
    "name": "update-wiki",
    "score": 0.88,
    "server": "documentation-mcp"
  },
  {
    "name": "create-github-release",
    "score": 0.81,
    "server": "github-mcp"
  }
]

Query 5: Complex Data Analysis

User Query

"Query my database for slow queries in the last hour, analyze the patterns, create a Jira ticket, and notify the DBA team"

Tools Returned

[
  {
    "name": "query-database",
    "score": 0.98,
    "server": "custom-db-mcp"
  },
  {
    "name": "analyze-query-performance",
    "score": 0.94,
    "server": "database-monitoring-mcp"
  },
  {
    "name": "create-jira-issue",
    "score": 0.91,
    "server": "jira-mcp"
  },
  {
    "name": "send-slack-message",
    "score": 0.89,
    "server": "slack-mcp"
  },
  {
    "name": "create-grafana-dashboard",
    "score": 0.72,
    "server": "monitoring-mcp"
  }
]

Query 6: Misspelled/Casual Language

User Query

"wanna deploy 2 prod n get some notifikatins plz"

Tools Returned

Despite the misspellings and casual language, Agent-CoreX understands the intent:

[
  {
    "name": "deploy-to-production",
    "score": 0.89,
    "reason": "Semantic match despite misspelling"
  },
  {
    "name": "send-slack-message",
    "score": 0.85,
    "reason": "Notification intent understood"
  },
  {
    "name": "send-email",
    "score": 0.72,
    "reason": "Alternative notification method"
  }
]

This shows the power of semantic search. Agent-CoreX doesn’t need exact keywords!

Query 7: Vague Query

User Query

"DevOps stuff"

Tools Returned

[
  {
    "name": "deploy-kubernetes",
    "score": 0.76,
    "reason": "DevOps-related"
  },
  {
    "name": "check-service-health",
    "score": 0.73,
    "reason": "Operations-related"
  },
  {
    "name": "terraform-apply",
    "score": 0.71,
    "reason": "Infrastructure automation"
  },
  {
    "name": "query-logs",
    "score": 0.68,
    "reason": "DevOps monitoring"
  },
  {
    "name": "auto-scale-service",
    "score": 0.66,
    "reason": "DevOps automation"
  }
]

Note: Vague queries return lower scores. More specific = better results!

Query 8: Tool Composition

User Query

"Create a GitHub issue from this error, assign it to the backend team, create a Jira ticket, and pin the error message to our Slack channel"

Tools Returned

[
  {
    "name": "create-github-issue",
    "score": 0.96,
    "server": "github-mcp"
  },
  {
    "name": "assign-issue",
    "score": 0.94,
    "server": "github-mcp"
  },
  {
    "name": "create-jira-issue",
    "score": 0.92,
    "server": "jira-mcp"
  },
  {
    "name": "pin-message",
    "score": 0.90,
    "server": "slack-mcp"
  },
  {
    "name": "send-slack-message",
    "score": 0.88,
    "server": "slack-mcp"
  }
]

Full Execution

const error = {
  message: "Connection timeout",
  stack: "...",
  timestamp: new Date()
};

const tools = await agent.retrieveTools({
  query: "Create GitHub issue, assign to backend team, create Jira ticket, pin to Slack"
});

// Create GitHub issue
const ghIssue = await agent.executeTool({
  toolName: 'create-github-issue',
  params: {
    repository: 'company/api',
    title: error.message,
    body: error.stack,
    labels: ['bug', 'critical']
  }
});

// Assign to backend team
await agent.executeTool({
  toolName: 'assign-issue',
  params: {
    repository: 'company/api',
    issueNumber: ghIssue.number,
    assignees: ['backend-team']
  }
});

// Create Jira ticket
const jiraIssue = await agent.executeTool({
  toolName: 'create-jira-issue',
  params: {
    project: 'PLATFORM',
    title: error.message,
    description: error.stack,
    priority: 'Highest'
  }
});

// Pin to Slack
await agent.executeTool({
  toolName: 'pin-message',
  params: {
    channel: '#incidents',
    message: `🚨 ${error.message}\nGH: ${ghIssue.url}\nJira: ${jiraIssue.url}`
  }
});

console.log('✅ Error tracked across all platforms');

Tips for Better Queries

✅ DO:

Be specific about what you want
Include service names when relevant
Mention end goals (“deploy and test”)
Use action verbs (“create”, “update”, “delete”)

❌ DON’T:

Use vague terms (“do stuff”)
Mix unrelated concepts
Expect mind-reading (specify platforms)

📊 Example Progression:

Query 1 (Bad):    "stuff"
Score: ❌ 0.45

Query 2 (Better): "deploy"
Score: ⚠️ 0.72

Query 3 (Good):   "deploy to AWS"
Score: ✅ 0.92

Query 4 (Best):   "deploy Node.js app to AWS Lambda with tests"
Score: ⭐ 0.98

Next Steps

Dynamic Retrieval

Learn how tools are ranked.

Real-World Use Cases

See complete production workflows.

API Reference

Execute these queries via API.

Try Live

Test queries in the playground.

Key Takeaway: The more specific your query, the better Agent-CoreX can help. Try the examples above in your own workflows! 🚀

​Real-World Sample Queries

​Query 1: Simple GitHub Workflow

​User Query

​What Agent-CoreX Returns

​Implementation

​Query 2: Multi-Tool Workflow

​User Query

​What Agent-CoreX Returns

​Implementation

​Query 3: Incident Response

​User Query

​Tools Returned

​Implementation

​Query 4: Infrastructure as Code

​User Query

​Tools Returned

​Query 5: Complex Data Analysis

​User Query

​Tools Returned

​Query 6: Misspelled/Casual Language

​User Query

​Tools Returned

​Query 7: Vague Query

​User Query

​Tools Returned

​Query 8: Tool Composition

​User Query

​Tools Returned

​Full Execution

​Tips for Better Queries

​✅ DO:

​❌ DON’T:

​📊 Example Progression:

​Next Steps

Dynamic Retrieval

Real-World Use Cases

API Reference

Try Live

Real-World Sample Queries

Query 1: Simple GitHub Workflow

User Query

What Agent-CoreX Returns

Implementation

Query 2: Multi-Tool Workflow

User Query

What Agent-CoreX Returns

Implementation

Query 3: Incident Response

User Query

Tools Returned

Implementation

Query 4: Infrastructure as Code

User Query

Tools Returned

Query 5: Complex Data Analysis

User Query

Tools Returned

Query 6: Misspelled/Casual Language

User Query

Tools Returned

Query 7: Vague Query

User Query

Tools Returned

Query 8: Tool Composition

User Query

Tools Returned

Full Execution

Tips for Better Queries

✅ DO:

❌ DON’T:

📊 Example Progression:

Next Steps