Strip explanatory comments from ladder/agent config additions

Author: john-b-yang

codeclash/agents/utils.py

@@ -26,8 +26,6 @@ class GameContext(BaseModel):
     round: int
     rounds: int
     working_dir: str
-    # Arena's static description (CodeArena.description), exposed to prompt templates as
-    # {{arena_description}}. Optional so non-pvp GameContext construction sites keep working.
     arena_description: str = ""
 
     def _render_prompt_templates(self) -> dict:

codeclash/arenas/battlesnake/battlesnake.py

@@ -203,8 +203,6 @@ def validate_code(self, agent: Player) -> tuple[bool, str | None]:
                 error_msg.append(f"There should be a `{func}` function implemented in `{self.submission}`")
         if len(error_msg) > 0:
             return False, "\n".join(error_msg + ["Don't change the function signatures!"])
-        
-	# `python main.py` boots the server via the __main__ entrypoint.
         if "__main__" not in bot_content or "run_server" not in bot_content:
             return False, (
                 f"`{self.submission}` must keep its server entrypoint so the bot starts: the "

configs/ablations/ladder/ladder_prompt.yaml

@@ -1,8 +1,3 @@
-# Shared task instructions for a model climbing a CC:Ladder run (all arenas). Each ladder run
-# config pulls this in on its `prompts:` line, so the framing lives in one place. Uses vars the
-# ladder already provides: player_id, arena_description, rounds, round, working_dir.
-# Deliberately says nothing about ladders/rungs/advancement — just: write code to beat one opponent.
-# NOTE: keep the include-directive token out of these comments; the resolver matches it on any line.
 game_description: |-
   You are a software developer ({{player_id}}) competing in a coding game. You are matched against a
   single opponent, and your job is to write and improve the code for your bot so that it defeats theirs.

configs/ablations/ladder/ladder_rules.yaml

@@ -1,5 +1,2 @@
-# Shared advancement rule for every CC:Ladder run config. Each <arena>[__<model>].yaml pulls
-# this in on its `ladder_rules:` line, so the win condition stays consistent across all
-# arenas/models in one place.
-min_round_wins: 2   # whole number of agent rounds to win to advance (1..tournament.rounds)
-win_last_k: 0       # must win the last K rounds (0 = off; must be <= min_round_wins)
+min_round_wins: 2
+win_last_k: 0

configs/mini/default.yaml

@@ -102,7 +102,4 @@ instance_template: |
   You cannot continue working in any way on this task after submitting.
 step_limit: 30
 cost_limit: 0
-# Global observation format for every model/tournament/arena. ClashAgent pushes this onto the
-# model config (mini-swe-agent renders observations model-side). The leading banner surfaces the
-# per-round step budget so the model can pace itself; the rest is mini's default observation body.
 observation_template: "[Step {{n_model_calls}}/{{step_limit}} used - {{step_limit - n_model_calls}} left]\n{% if output.exception_info %}<exception>{{output.exception_info}}</exception>\n{% endif %}<returncode>{{output.returncode}}</returncode>\n<output>\n{{output.output}}</output>"