Python Code Validator

Validates, repairs and hardens AI-generated Python: syntax and lint diagnostics, security scanning (bandit, AST policy, credential scan), deterministic auto-fixes and optimisation via SMIRE, optional sandboxed execution and Python-to-JavaScript transpilation. mode=static (default) never runs the submitted code: it parses it, lints it (ruff), type-checks it (mypy), applies an AST security policy and scans for credentials. mode=repair adds deterministic fixes and returns the rewritten source in fixed_code, keeping the original when it cannot prove the fix. mode=execute additionally runs the code in a container sandbox — no network, read-only filesystem, killed at options.timeout_s — and reports its exit code, stdout and stderr; treat that as a side effect and do not submit code you do not want executed. Auth: a key is required for every call and static is free (100 calls per day, then HTTP 429); repair, execute need a paid key and answer 402 without one. Nothing outside the sandbox is written, and the code plus its verdict are retained to improve the service. Returns valid, score 0..1, diagnostics (rule, message, line, column), security findings, fixes, fixed_code and runtime; see outputSchema. Use it on Python you generated or edited before writing it to disk, and prefer static unless you need the fix or proof that it runs.