Abstract: The safety analysis of pebble-bed high-temperature gas-cooled reactor (PB-HTGR) relies heavily on high-fidelity system-level simulation tools. However, existing analysis codes face significant limitations in modeling the integrated transient response of the full primary circuit. Programs like TINTE, while well-validated for core physics, often simplify key primary loop components such as the main helium circulator and steam generator to fixed boundary conditions (source/sink terms). This simplification precludes the accurate simulation of dynamic interactions and feedback between these components and the core during transients. Other approaches, such as externally coupling detailed component models or adapting light water reactor system codes, introduce synchronization errors, architectural complexity, or fundamental model mismatches. To overcome these architectural limitations, this paper presented the independent development of ANCHOR, a novel system-level, high-fidelity transient safety analysis code for PB-HTGR. The code was written in C++ and adopted a modular architecture for flexibility. Its core advancement lied in two key aspects. First, it established a full primary circuit closed solving framework, integrating the core, hot/cold helium ducts, steam generator, and main helium circulator into a unified, dynamically coupled system model. Second, it introduced a distributed-parameter dynamic model for the main helium circulator, based on its characteristic performance curves. This model directly incorporated the circulator’s pressure rise and dissipation work into the momentum and energy conservation equations, replacing the traditional lumped-parameter source-term approach. This enables a physically realistic simulation of circulator transients (startup, shutdown, speed variation) and its coupling with the loop. The accuracy and reliability of ANCHOR were rigorously demonstrated through a hierarchical verification strategy. This included: benchmarking against the IAEA SANA experiment, confirming the accuracy of its packed-bed effective thermal conductivity model; comparison with the HCP code for a reactivity-induced transient, validating its neutronics and thermal-hydraulic coupling; and system-level validation against the established TINTE code for a depressurized loss of forced cooling (DLOFC) accident in HTR-PM, showing excellent agreement in fuel temperatures. To showcase its capability in analyzing complex beyond-design-basis events, ANCHOR was applied to simulate a challenging accident scenario for the HTR-PM, a loss of normal feedwater combined with a failure of the main circulator baffle to close. The ANCHOR successfully captures the complete transition mechanism from forced circulation to reverse natural circulation within the primary loop. It elucidates the dynamic evolution of this natural circulation flow and its critical effects on system pressure and core temperature redistribution. All calculated temperatures remain well within safety limits. In conclusion, the ANCHOR, with its integrated loop architecture and high-fidelity component modeling, provides a robust, self-developed, and high-precision tool for the safety design and analysis of PB-HTGRs, effectively addressing gaps left by previous analysis methodologies.