The vulnerability allows code injection via crafted formulas using eval(). The analysis started by identifying the main entry point for formula evaluation in Pycel, which is ExcelCompiler.evaluate(), based on common usage patterns found in the library's README. This function processes the cell reference containing the malicious formula. The PoC specifically mentions that eval() bypasses checks when nested (e.g., within an IF function). This points to the mechanism responsible for evaluating function arguments and function calls within the formula AST. The FunctionExpression.evaluate() method is central to this process. If eval(...) is parsed as a function, this method would handle its evaluation. The vulnerability suggests that the sandboxing for eval fails here, leading to the execution of the Python eval function. The file paths and class/method names are based on the structure of the dgorissen/pycel repository, which is a public fork and likely representative of the codebase up to the vulnerable version 1.0b30.