The anti-$k_t$ jet clustering algorithm is a new member of the class of sequential recombination algorithms, characterized by a negative power in the distance measure. Unlike traditional algorithms, it behaves like an idealized cone algorithm, producing conical jets with soft fragmentation, equal active and passive areas, zero anomalous dimensions, and non-global logarithms similar to a rigid boundary. It is infrared and collinear safe, making it suitable for experimental calibration and theoretical calculations. The algorithm is based on generalizing the $k_t$ and Cambridge/Aachen algorithms, using a distance measure defined by $d_{ij} = \min(k_{ti}^{2p}, k_{tj}^{2p}) \frac{\Delta_{ij}^2}{R^2}$ and $d_{iB} = k_{ti}^{2p}$. For $p = -1$, it is called the "anti-$k_t$" algorithm. It is resilient to soft radiation, maintaining jet boundaries unaffected by soft particles, while being flexible to hard radiation. This makes it ideal for applications where jet shapes are sensitive to soft effects, such as top mass reconstruction. The algorithm is also computationally efficient, with a time complexity of $O(N^{3/2})$ for clustering N particles. It is compared to other algorithms in terms of jet area, back-reaction, and non-global logarithms, showing superior performance in maintaining jet area and reducing fluctuations. The anti-$k_t$ algorithm is a natural replacement for iterative cone algorithms, offering a safe, efficient, and accurate method for jet clustering.The anti-$k_t$ jet clustering algorithm is a new member of the class of sequential recombination algorithms, characterized by a negative power in the distance measure. Unlike traditional algorithms, it behaves like an idealized cone algorithm, producing conical jets with soft fragmentation, equal active and passive areas, zero anomalous dimensions, and non-global logarithms similar to a rigid boundary. It is infrared and collinear safe, making it suitable for experimental calibration and theoretical calculations. The algorithm is based on generalizing the $k_t$ and Cambridge/Aachen algorithms, using a distance measure defined by $d_{ij} = \min(k_{ti}^{2p}, k_{tj}^{2p}) \frac{\Delta_{ij}^2}{R^2}$ and $d_{iB} = k_{ti}^{2p}$. For $p = -1$, it is called the "anti-$k_t$" algorithm. It is resilient to soft radiation, maintaining jet boundaries unaffected by soft particles, while being flexible to hard radiation. This makes it ideal for applications where jet shapes are sensitive to soft effects, such as top mass reconstruction. The algorithm is also computationally efficient, with a time complexity of $O(N^{3/2})$ for clustering N particles. It is compared to other algorithms in terms of jet area, back-reaction, and non-global logarithms, showing superior performance in maintaining jet area and reducing fluctuations. The anti-$k_t$ algorithm is a natural replacement for iterative cone algorithms, offering a safe, efficient, and accurate method for jet clustering.