Skip to content

Analysis Tools

Analysis tools help AI assistants understand code structure without guessing. They provide accurate information about functions, classes, imports, and code organization.

Tools in This Category

Tool Description Tier
analyze_code Parse code structure (functions, classes, imports) Community

analyze_code

Parse code and extract its structure with zero hallucination.

What It Does

analyze_code parses source code using real AST (Abstract Syntax Tree) analysis—not regex pattern matching. It returns structured information about:

  • Functions: Names, parameters, return types, docstrings, line numbers
  • Classes: Names, methods, inheritance, attributes
  • Imports: All import statements with their targets
  • Complexity: Cyclomatic complexity metrics

When AI Agents Use This

  • Before making any code modifications
  • To understand file structure before extracting code
  • To find all functions/classes in a file
  • To check code complexity

Quick Reference

Property Value
Tier Community
Languages Python, JavaScript, TypeScript, Java, JSX, TSX
Token Cost ~100-500 tokens

Parameters

Parameter Type Required Description
code string ✓* Source code to analyze
file_path string ✓* Path to file to analyze
language string Language hint (auto-detected if omitted)

*Provide either code or file_path, not both.

Example Usage

Prompt:

"Analyze the structure of src/auth.py"

Tool Call: 

{
  "tool": "analyze_code",
  "parameters": {
    "file_path": "src/auth.py"
  }
}

Prompt:

"What functions are in this code?"

Tool Call: 

{
  "tool": "analyze_code",
  "parameters": {
    "code": "def hello(): pass\ndef world(): pass"
  }
}

Response Format

{
  "data": {
    "functions": [
      {
        "name": "authenticate_user",
        "line_start": 15,
        "line_end": 32,
        "parameters": [
          {"name": "username", "type": "str"},
          {"name": "password", "type": "str"}
        ],
        "return_type": "Optional[User]",
        "docstring": "Authenticate a user by username and password.",
        "complexity": 4,
        "is_async": false
      }
    ],
    "classes": [
      {
        "name": "AuthService",
        "line_start": 5,
        "line_end": 50,
        "docstring": "Service for authentication operations.",
        "methods": ["__init__", "authenticate_user", "logout"],
        "bases": ["BaseService"],
        "attributes": ["db", "cache"]
      }
    ],
    "imports": [
      {"module": "typing", "names": ["Optional"], "line": 1},
      {"module": "hashlib", "names": null, "line": 2}
    ],
    "global_variables": [],
    "total_lines": 50,
    "total_functions": 3,
    "total_classes": 1
  },
  "tier_applied": "community",
  "duration_ms": 45
}

Why This Matters

Without analyze_code With analyze_code
AI guesses function names Exact function list with signatures
"I think there's a class..." Precise class structure and methods
Reads entire file (~10K tokens) Returns only structure (~200 tokens)
May hallucinate non-existent code Zero hallucination—AST parsed

Tier Differences

Feature Community Pro Enterprise
Basic structure parsing ✅ ✅ ✅
Complexity metrics ✅ ✅ ✅
Docstring extraction ✅ ✅ ✅
Type hint extraction ✅ ✅ ✅
Max file size 1 MB 10 MB Unlimited

Common Patterns

Pattern: Analyze → Extract → Modify

1. analyze_code(file_path="src/auth.py")
   → Learn that authenticate_user exists at line 15

2. extract_code(file_path="src/auth.py", target_name="authenticate_user")
   → Get exact function code

3. update_symbol(file_path="src/auth.py", target_name="authenticate_user", new_code=...)
   → Replace with improved version

Troubleshooting

Returns empty results
  • Check that the file exists and is readable
  • Verify the file contains valid syntax for the language
  • Use the language parameter if auto-detection fails
Syntax error in response
  • The file may have invalid syntax
  • Try running a linter on the file first
  • Check for encoding issues (non-UTF-8 files)

Full deep dive