No user-submitted related servers found.
Scored across 144 tools
Many tools have distinct, well-documented purposes, but the set contains numerous overlapping read tools for map state (known_world, map_window, revealed_map, map_index, tactical_view, explore_frontier) and status (turn_status, briefing, overview, turn_digest). An agent must carefully read descriptions to choose the right one, and some choices (e.g., available_unit_actions vs todo_actions vs units vs unit_mission_targets) could be confused.
Names are uniformly snake_case with predictable prefixes: read-only tools are nouns (units, cities, overview), actions are verbs (move_unit, set_production, choose_policy), option lists use available_* or *_options, and choosers use choose_*. A few generic names like do and noun-only reads deviate slightly from a strict verb_noun pattern, but the overall convention is clear and consistent.
144 tools is far beyond the 3–15 range considered well-scoped, and the rubric explicitly flags 50+ as an extreme mismatch. Even for a complex game like Civilization V, this volume imposes a heavy selection burden on the agent and suggests many niche or overlapping tools could be consolidated.
The surface covers nearly every game subsystem: units, combat, cities, production, research, policies, religion, espionage, diplomacy, trade, map reading, save/load, notebook, and conditional orders. There are no obvious gaps in CRUD or lifecycle coverage for the domain, and dead ends are explicitly handled with blocker-clearing tools.