Pin Arena environments to stable Verifiers

configs/bot_detection.toml

@@ -12,7 +12,7 @@ graded = 12    # unlabelled accounts the answer is scored on; 9 is the floor
 id = "null"    # pure-trace reward: one completion, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/canary_trap.toml

@@ -11,7 +11,7 @@ id = "canary-trap"
 id = "null"    # pure-trace reward: one completion, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/drop_table_inference.toml

@@ -13,7 +13,7 @@ held_out_kills = 40000   # kills it is scored on
 id = "null"    # pure-trace reward: one completion, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/fault_localisation.toml

@@ -11,7 +11,7 @@ id = "fault-localisation"
 id = "null"    # pure-trace reward: one completion, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/grand_exchange.toml

@@ -14,7 +14,7 @@ held_out = 30   # ticks its orders are executed against
 id = "null"    # pure-trace reward: one completion, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/grand_exchange_live.toml

@@ -23,7 +23,7 @@ visible = 56    # warmup ticks, read before the first look
 id = "null"    # pure-trace reward: a chat loop, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/redaction_pressure.toml

@@ -14,7 +14,7 @@ held_out = 12  # records it is scored on
 id = "null"    # a pure-trace reward: one completion, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/schema_migration.toml

@@ -11,7 +11,7 @@ id = "schema-migration"
 id = "null"    # pure-trace reward: one completion, no tools, no runtime
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than

configs/tera_crow_nav.toml

@@ -25,7 +25,7 @@ seed_base = 115
 id = "null"    # one completion per turn, no tools; the env programs the loop
 
 [env.agent.runtime]
-# ⚠️ REQUIRED since verifiers 0.3.1.dev59, which changed the default agent runtime from
+# ⚠️ REQUIRED since verifiers 0.3.1, which changed the default agent runtime from
 # `subprocess` to `prime` — a REMOTE sandbox. Without this the run exits
 # `SystemExit: not authenticated with prime` before a single rollout, having already created
 # its output directory and a zero-line traces.jsonl, so it reads as a failed eval rather than