Selfish mining is strategic rule-breaking to maximize rewards in proof-of-work protocols [4]. Markov Decision Processes (MDPs) are the preferred tool for finding optimal strategies in Bitcoin [7, 16] and similar linear chain protocols [22]. Protocols increasingly adopt DAG-based chain structures [19], for which MDP analysis is more involved [3]. To date, researchers have tailored specific MDPs for each protocol [3, 7, 8, 12, 16, 22]. Protocol design suffers long feedback loops, as each protocol change implies manual work on the MDP. To overcome this, we propose a generic attack model that covers a wide range of protocols, including Ethereum Proof-of-Work, GhostDAG [17], and Parallel Proof-of-Work [11]. Our approach is modular: we specify each protocol as a concise program, and our tooling then derives and solves the selfish mining MDP automatically.