Boss fights expose weaknesses that normal mountains can hide. This guide keeps the repeatable lessons from a full run and a Dragon build test, while treating recorded unit counts and damage values as observations—not universal requirements for every patch.
1. Prepare the economy before the damage check
A boss build begins with logistics. If ore is stranded, the gold needed for a last-minute upgrade never arrives. Before the checkpoint:
- clear persistent ore piles;
- improve Runner capacity and uptime;
- reserve housing for the units the fight actually needs;
- clear Dens and valuable wall targets on the preceding mountains;
- keep scarce resources for a tactical ability or decisive upgrade.
The cleanest test is the previous mountain: if it takes longer every level while your bank is empty, you are not boss-ready yet.
2. Separate primary DPS from target coverage
The source tests found that a focused Laser build supported by Ballistas performed better than spreading resources across every weapon family. The lesson is broader than “always use Lasers”:
- choose one primary damage engine and reach its important upgrades;
- add a second unit only when it solves a coverage problem;
- keep support units for economy, research, and resource discovery;
- stop funding a unit that does not improve boss time or pre-boss rewards.
Lasers can scale strongly with Scientists and gem-related upgrades. Ballistas complement them by covering Dens and reliable lines of fire. In another version or artifact set, a different primary engine may win; the focus-and-support structure still applies.
3. Use a readiness test, not a copied DPM number
Recorded Dragon attempts reported very different displayed damage values, and game speed or version behavior can affect that display. A safer readiness test is operational:
- time a normal high-HP target at 1× speed;
- confirm that the main damage engine remains active throughout;
- check that the treasury can fund one emergency upgrade;
- verify that important abilities will be ready at the start;
- enter with a safety margin rather than a theoretical exact kill.
If a normal target barely dies inside the boss timer, do not expect a clean boss clear.
4. Position before the fight starts
Straight-line weapons and elevated targets make layout part of DPS. Before committing:
- move Ballistas where their firing lines reach the intended target;
- avoid placing support structures where they block or waste a useful angle;
- ensure burst damage is aimed at the boss, not an irrelevant wall section;
- check whether the fight relocates units into a separate arena.
Positioning is a free multiplier. Use it before buying another expensive unit.
5. Plan burst abilities around the timer
If a high-impact ability has a long cooldown, using the first charge early may create time for another charge near the end. Keep the final use for the point where it changes the outcome, not simply when it becomes available.
Do not spend the required Mithril or other scarce resource on a minor upgrade immediately before the fight. A reserve has option value: it can become burst damage, a key weapon tier, or an emergency logistics fix.
6. Dragon case study: why the hybrid worked
In the analyzed run, a mixed “little of everything” setup stalled because gold, Mithril, and upgrade depth were split across too many families. The successful Dragon attempt concentrated resources into:
- Lasers as the main scaling and gem engine;
- Ballistas for consistent focused damage and Den coverage before the fight;
- sufficient Runners to keep purchases flowing;
- a small number of Demolishers and exploration units for utility rather than primary DPS.
The result should be read as evidence for specialization, not a permanent tier list. Artifact choices and future balance changes can shift the best unit.
7. Recover from a failed attempt methodically
After a failure, identify which of three things happened:
- Damage failure: the boss stayed targetable, but HP remained. Deepen the main engine.
- Uptime failure: units were disabled, badly positioned, or starved of support. Fix operation before buying damage.
- Resource failure: the fight began without the burst charge or upgrade you planned. Rebuild the reserve.
Changing five systems at once hides the cause. Make one meaningful correction, then retest on a comparable target.
Boss checklist
- The previous mountain clears with time to spare at 1× speed.
- The ground is not holding a large unpaid ore balance.
- One primary damage engine has reached its key upgrades.
- Coverage units can hit Dens and the boss from their current positions.
- Tactical resources and abilities are reserved.
- A failed attempt has a specific diagnosis, not a complete rebuild by reflex.
首領戰會暴露普通山體能夠掩蓋的問題。這篇攻略保留了全流程建議與龍戰構築測試中可複用的部分；視頻裡的單位數量與面板傷害只視為一次運行記錄，不當作所有版本的硬門檻。
1. 先準備經濟，再接受傷害檢查
首領戰構築的起點仍然是物流。礦石運不回來，臨戰升級所需金幣就不會到賬。進入節點前應做到：
- 清除持續堆積的礦石；
- 提高搬運工負重與在線率；
- 為戰鬥真正需要的單位預留住房；
- 在前幾座山清完巢穴與高價值目標；
- 為戰術技能或關鍵升級保留稀缺資源。
最直接的測試就是上一座山：如果每層耗時持續上升，而且賬戶總是空的，當前還不適合挑戰首領。
2. 把主輸出與目標覆蓋拆開考慮
來源測試中，以激光炮為主力、魚叉弩炮為輔助的集中構築，優於平均投資所有武器。更通用的結論不是“永遠用激光炮”，而是：
- 選擇一個主傷害引擎，併到達它的關鍵升級；
- 第二類單位只在能解決覆蓋問題時加入；
- 保留經濟、研究與資源探索所需的支援單位；
- 某個單位既不能縮短首領戰時間，也不能增加戰前收益時，停止繼續投資。
激光炮可通過科學家與寶石相關升級成長，魚叉弩炮則補足拆巢和穩定火線。即使未來版本或神器組合改變最強兵種，“主力 + 補位”的結構仍然有效。
3. 用實戰測試代替照抄每分鐘傷害
不同龍戰記錄裡的面板傷害差異很大，遊戲速度與版本行為也可能影響顯示。更可靠的是運行狀態測試：
- 在 1× 速度下計時擊殺一個高血量普通目標；
- 確認主傷害引擎全程正常工作；
- 確認賬戶還能承擔一次緊急升級；
- 檢查關鍵技能是否能在開戰時使用；
- 留出安全餘量，不要只滿足理論上的極限擊殺。
如果普通目標都只能卡著計時結束，就不要期待首領戰能穩定通過。
4. 開戰前完成站位
直線武器與高處目標會讓佈局直接影響實際傷害輸出。進入戰鬥前應：
- 把魚叉弩炮移到能覆蓋目標的火線上；
- 避免支援建築佔據或浪費關鍵角度；
- 確保爆發技能瞄準首領，而不是無關牆面；
- 確認首領是否會把單位轉移到獨立戰場。
站位是免費的倍率，應在購買昂貴新單位之前先優化。
5. 圍繞計時器安排爆發技能
如果高傷技能冷卻很長，開場儘早使用第一發，可能讓結束前多出第二次機會。最後一次使用應留在真正能改變結果的時點，而不是冷卻一結束就隨手放掉。
戰前不要把所需秘銀或其他稀缺資源花在邊緣升級上。儲備代表選擇權：它可以轉化成爆發傷害、關鍵武器等級，也可以用來緊急修復物流。
6. 龍戰案例：為什麼集中混合構築有效
測試中的“每種都來一點”會停滯，因為金幣、秘銀與升級深度被過多兵種分散。成功的龍戰把資源集中在：
- 激光炮：主成長與寶石引擎；
- 魚叉弩炮：穩定集火，並負責戰前拆巢；
- 足量搬運工：保證購買能力持續運轉；
- 少量異化矮人與探索單位：提供功能，而不是承擔主要傷害輸出。
這個結果證明的是“專注投資”，而不是一張永久兵種榜。神器選擇和後續平衡都可能改變最優單位。
7. 失敗後按原因修復
失敗後先判斷屬於哪一類：
- 傷害不足： 首領全程可被攻擊，但剩餘血量明顯。繼續深化主引擎。
- 在線率不足： 單位被控、站位錯誤或支援斷檔。先修運行狀態。
- 資源不足： 開戰時缺少計劃中的技能次數或升級。重建儲備。
同時改動五個系統，會讓你無法確認真正原因。每次只做一項有意義的修正，再用相近目標複測。
首領戰前清單
- 上一座山在 1× 速度下仍能留出明顯餘量。
- 地面沒有大量尚未入賬的礦石。
- 一個主傷害引擎已經到達關鍵升級。
- 覆蓋單位從當前站位可以打到巢穴與首領。
- 戰術資源和技能次數已預留。
- 失敗後先有明確診斷，而不是條件反射地推倒重來。