Code Doctor Security Model

Security Principles

Host Safety: Never execute arbitrary user-provided untrusted code directly on host machines. Verification uses safe AST parsing and rule re-evaluation.
Secret Scrubbing: Sensitive keys, passwords, and tokens are automatically redacted via SensitiveDataRedactor prior to external AI API calls.
Path Traversal Protection: Project Scanner validates paths with is_safe_path() to ensure operations remain bounded within project roots.
Gitignore Compliance: Excludes .git, node_modules, venv, cache, and secret files automatically.