gemini-code-exec

Gemini Code Exec

Use this skill to run Python code inside Gemini's isolated sandbox. This gives you a clean, side-effect-free execution environment for verifying computations, testing patterns, and validating logic without touching the local machine.

When to use this skill

• Math verification: Confirm a formula, check a statistical calculation, or validate a numeric result.
• Regex testing: Test a regular expression against a set of sample inputs before embedding it in production code.
• Algorithm simulation: Run a sorting algorithm, dynamic programming solution, or graph traversal with sample data to verify correctness.
• Data transformation validation: Confirm that a proposed data mapping, normalization, or encoding produces the expected output.
• Edge case probing: Run code against boundary values (zero, max int, empty string, None) to see what happens.
• Dependency-free snippets: Execute a pure-Python snippet that has no local filesystem or network dependencies.

MCP tools

Tool	Purpose
gemini_code_execute	Submit Python code for execution in Gemini's sandbox and receive stdout, stderr, and the return value

Usage pattern

Verify a math formula

{
  "tool": "gemini_code_execute",
  "arguments": {
    "code": "import math\n\ndef compound_interest(principal, rate, periods):\n    return principal * (1 + rate) ** periods\n\nresult = compound_interest(1000, 0.05, 10)\nprint(f'After 10 years: {result:.2f}')"
  }
}

Test a regex pattern

{
  "tool": "gemini_code_execute",
  "arguments": {
    "code": "import re\n\npattern = r'^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}$'\ntest_cases = ['[email protected]', 'bad-email', '[email protected]', '[email protected]']\nfor case in test_cases:\n    match = bool(re.match(pattern, case))\n    print(f'{case!r}: {match}')"
  }
}

Validate an algorithm

{
  "tool": "gemini_code_execute",
  "arguments": {
    "code": "def longest_common_subsequence(s1, s2):\n    m, n = len(s1), len(s2)\n    dp = [[0] * (n + 1) for _ in range(m + 1)]\n    for i in range(1, m + 1):\n        for j in range(1, n + 1):\n            if s1[i-1] == s2[j-1]:\n                dp[i][j] = dp[i-1][j-1] + 1\n            else:\n                dp[i][j] = max(dp[i-1][j], dp[i][j-1])\n    return dp[m][n]\n\nprint(longest_common_subsequence('ABCBDAB', 'BDCAB'))  # Expected: 4"
  }
}