Tuberculosis (TB) remains a pressing global health threat, while the emergence of drug-resistant TB adds complexity to clinical treatment strategies, underscoring the need for rapid and accurate diagnosis. In this study, we present an innovative identification and antibiotic susceptibility testing system, termed Assembled PCR & CRISPR for Ultra-sensitive Rifampin-resistant Analysis Testing (ACURAT). ACURAT employed a nested polymerase chain reaction (PCR) scheme to sensitively detect as few as 10 copies of the TB genome, more importantly, addressing the clustered regularly interspaced short palindromic repeats/associated protein 9 (CRISPR-Cas9) system’s requirement for a protospacer adjacent motif (PAM) sequence by introducing with the designed primers. Recognizing the limitation of fully-matched guide ribonucleic acid (RNA) in distinguishing one-base single nucleotide polymorphisms (SNPs), we designed a micro-mismatch CRISPR-Cas9 system with significant selectivity, featuring a single-base mismatch with wildtype genotypes and at least two base mismatches with drug-resistant genotypes. The systematic optimization of ACURAT parameters revealed that a 19 nt guide RNA length, a 1 bp distance to the PAM sequence, and the use of a high-fidelity Cas9 nuclease significantly enhanced the efficiency. Finally, ACURAT successfully distinguished fifteen drug-resistance-related genotypes in codons 516, 526, 531, and 533 of the polymerase beta subunit gene. ACURAT demonstrated substantial clinical potential for efficient TB diagnosis and exhibited the capability to address other sequence recognition scenarios.