The universal method is to diagonalize matrix T under standard basis (1,x,x^2), and then take the eigenvector as the new basis such that [T]_beta is diagonal. Specially for this question, you can try assigning a=1, b=1, c=1 respectively and solve the other two variables.