Skip to content

Tools reference＃

The kluster.ai Code MCP server provides verification tools for checking AI-generated code quality and security. These tools enable real-time code verification directly within your IDE through MCP integration.

It includes:

  • kluster_code_review_auto: Verifies code quality and detects bugs, including logic errors, security issues, and performance problems.
  • kluster_dependency_validator: Validates the security and compliance of packages and dependencies in your code.

These tools share the same set of parameters. This page documents those parameters and the response formats you'll see when using these tools in Cursor, Claude Code, or any MCP-compatible client.

Parameters＃

These tools analyze AI-generated code and its dependencies to detect bugs, security vulnerabilities, and other quality issues.

Parameters

code_diff string required

Unified diff format showing the actual code changes.

user_requests string required

Chronological sequence of user messages with current request marked as >>> CURRENT REQUEST:.

modified_files_path string required

Full absolute paths of modified files separated by ;.

Response fields＃

All Code verification tools return the same response structure:

  • isCodeCorrect: Boolean indicating if the code has issues.
  • explanation: Summary of all issues found.
  • issues: Array of detected problems with:
  • type: Issue category (intent, semantic, knowledge, performance, quality, logical, security).
  • severity: Impact level (critical, high, medium, low).
  • priority: Execution priority (P0-P5).
  • description: Brief issue summary.
  • explanation: Detailed issue explanation.
  • actions: Recommended fixes.
  • priority_instructions: Execution rules for addressing issues.
  • agent_todo_list: Prioritized list of fixes to apply.

Example response＃

{  
    "isCodeCorrect": false,  
    "explanation": "Found 3 issues. 1 critical issue needs immediate attention.",  
    "issues": [  
        {  
            "type": "security",  
            "severity": "critical",  
            "priority": "P2",  
            "description": "SQL injection vulnerability",  
            "explanation": "User input is directly concatenated into SQL query without sanitization.",  
            "actions": "Use parameterized queries or prepared statements."  
        }  
    ],  
    "priority_instructions": "Fix P2 issues before deploying code.",  
    "agent_todo_list": [  
        "P2.1: Fix SQL injection vulnerability by using parameterized queries"  
    ]  
}

Priority system＃

Code verification assigns priority levels to detected issues, helping you focus on the most critical problems first. The system automatically prioritizes based on issue type and severity.

  • P0-P1: Intent issues (highest priority) - code doesn't match request.
  • P2: Critical severity - must fix immediately.
  • P3: High severity - should fix soon.
  • P4: Medium severity - nice to fix.
  • P5: Low severity - optional improvements.

Next steps＃

© 2025 kluster.ai. All rights reserved.
Made with 🩵 by PaperMoon