fix(eval): require finite gateway startup budgets

eval/tests/test_model_gateway.py

@@ -358,12 +358,23 @@ def test_gateway_startup_budget_outlives_a_cold_litellm_import(monkeypatch, tmp_
     monkeypatch.setenv(GATEWAY_READY_TIMEOUT_ENV, "42.5")
     assert gateway_ready_timeout_s() == 42.5
 
-    for bad in ("0", "-1", "soon"):
+    for bad in ("0", "-1", "soon", "nan", "inf", "-inf", "1e999"):
         monkeypatch.setenv(GATEWAY_READY_TIMEOUT_ENV, bad)
         with pytest.raises(ValueError, match=GATEWAY_READY_TIMEOUT_ENV):
             gateway_ready_timeout_s()
 
 
+@pytest.mark.parametrize("timeout", [0.0, -1.0, float("nan"), float("inf"), float("-inf")])
+def test_gateway_rejects_invalid_explicit_readiness_budgets(tmp_path: Path, timeout: float) -> None:
+    with pytest.raises(ValueError, match="finite and positive"):
+        OpenAIGateway(
+            openai_api_key="sk-offline-test",
+            model_names=["gpt-4.1"],
+            work_dir=tmp_path / "gw",
+            ready_timeout_s=timeout,
+        )
+
+
 def test_gateway_readiness_timeout_reports_the_proxy_log_and_the_override(tmp_path: Path) -> None:
     gateway = OpenAIGateway(
         openai_api_key="sk-openai-secret",

eval/workflow_bench/model_gateway.py

@@ -11,6 +11,7 @@ proxy process — never in the sandboxed agent environment.
 from __future__ import annotations
 
 import argparse
+import math
 import os
 import re
 import secrets
@@ -60,8 +61,8 @@ def gateway_ready_timeout_s() -> float:
         value = float(raw)
     except ValueError as exc:
         raise ValueError(f"{GATEWAY_READY_TIMEOUT_ENV} must be a number of seconds, not {raw!r}") from exc
-    if value <= 0:
-        raise ValueError(f"{GATEWAY_READY_TIMEOUT_ENV} must be positive, not {raw!r}")
+    if not math.isfinite(value) or value <= 0:
+        raise ValueError(f"{GATEWAY_READY_TIMEOUT_ENV} must be finite and positive, not {raw!r}")
     return value
 
 
@@ -268,6 +269,8 @@ class OpenAIGateway(AbstractContextManager["OpenAIGateway"]):
         self.model_names = tuple(model_names)
         self.work_dir = work_dir
         self.ready_timeout_s = gateway_ready_timeout_s() if ready_timeout_s is None else ready_timeout_s
+        if not math.isfinite(self.ready_timeout_s) or self.ready_timeout_s <= 0:
+            raise ValueError("gateway readiness timeout must be finite and positive")
         self.auth_token = secrets.token_hex(16)
         self.port = _free_loopback_port()
         self.base_url = f"http://127.0.0.1:{self.port}"

eval/workflow_bench/runner.py

@@ -5,7 +5,7 @@ Usage:
         --tasks workflow_bench/tasks.scenarios.yaml --runs 3 \
         --model claude-sonnet-4-20250514
 
-Each task runs in a fresh detached git worktree of the target repo, once per
+Each task runs in a fresh self-contained clone of the target repo, once per
 arm per run:
 
 * ``workflow`` — two headless Claude Code sessions: gitnexus-plan, then