Add --judge-provider flag; document key resolution rules (env per host, judge included)

Co-Authored-By: Claude <noreply@anthropic.com>

README.md

@@ -69,6 +69,7 @@ evalharness eval run gsm8k \
 | `--model NAME` | 服务端模型名（配合 `--api-url`）；或直接给完整 spec：`openai/http://h:8000/v1?qwen3-8b` |
 | `--provider {openai-chat,openai-pool}` | 协议/提供方，默认 `openai-chat`；端点池用 `openai-pool` |
 | `--judge NAME` + `--judge-api-url URL` | LLM-judge 模型（hle / simple_qa / imo 需要）；也可单给完整 spec `--judge openai/http://...?m` |
+| `--judge-provider {openai-chat,openai-pool}` | judge 协议；多 judge 端点负载均衡用 `openai-pool` |
 | `--profile NAME` | 命名生成参数集（内置 `dp4-nothink`、`qwen3-es-parity`、`t1-short`，或任意 `@register_gen_profile` 名）；优先级：插件默认 < profile 默认 < profile 单 bench 覆盖 < 显式参数 |
 | `--disable-thinking` | 发送 `enable_thinking=false`（Qwen3 类思考模型推荐；带 tools 的请求自动退回模板安全的软开关） |
 | `--textools` | 工具以文本形式随 prompt 下发，而非原生 tool_calls |
@@ -160,7 +161,9 @@ evalharness viz show report.json --style errors            # 失败样本下钻
 | `mock` / `mock-boxed` / `mock-fc` | 离线适配器（管线自检：回声 / 回放金标 / 回放工具调用） |
 | `!nothink` `!perf` `!textools` 后缀 | spec 内联开关（与 CLI 参数等价） |
 
-API key 按端点从环境变量读取（`OPENAI_API_KEY`、`ANTHROPIC_API_KEY` …）。
+API key（含 judge）按端点域名自动从环境变量读取：`api.openai.com`→`OPENAI_API_KEY`、
+`anthropic.com`→`ANTHROPIC_API_KEY`、`dashscope`→`DASHSCOPE_API_KEY`、`bigmodel`→`ZAI_API_KEY`；
+其余域名回退 `OPENAI_API_KEY`。自建端点无鉴权时无需设置。
 
 ## 内置 benchmark
 

evalharness/cli.py

@@ -233,8 +233,10 @@ def _compose_judge_spec(args):
     legacy spec; keep both working like the main model flags."""
     judge = getattr(args, 'judge', '') or ''
     url = getattr(args, 'judge_api_url', '') or ''
+    provider = getattr(args, 'judge_provider', 'openai-chat') or 'openai-chat'
+    internal = {'openai-chat': 'openai', 'openai-pool': 'openai-pool'}.get(provider, provider)
     if url and judge and '/' not in judge:
-        judge = f'openai/{url.rstrip("/")}?{judge}'
+        judge = f'{internal}/{url.rstrip("/")}?{judge}'
     return judge or None
 
 
@@ -243,8 +245,10 @@ def _compose_judge_spec(args):
     legacy spec; both keep working, mirroring the main model flags."""
     judge = getattr(args, 'judge', '') or ''
     url = getattr(args, 'judge_api_url', '') or ''
+    provider = getattr(args, 'judge_provider', 'openai-chat') or 'openai-chat'
+    internal = {'openai-chat': 'openai', 'openai-pool': 'openai-pool'}.get(provider, provider)
     if url and judge and '/' not in judge:
-        judge = f'openai/{url.rstrip("/")}?{judge}'
+        judge = f'{internal}/{url.rstrip("/")}?{judge}'
     return judge or None
 
 
@@ -520,6 +524,9 @@ def build_parser() -> argparse.ArgumentParser:
                    help='judge model name with --judge-api-url, or full spec')
     p.add_argument('--judge-api-url', default='',
                    help='judge API base URL when --judge is only the model name')
+    p.add_argument('--judge-provider', default='openai-chat',
+                   help='judge protocol/provider (default openai-chat; '
+                        'openai-pool for multi-endpoint judges)')
     p.add_argument('--profile', default='',
                    help='named gen-params profile (dp4-nothink | qwen3-es-parity | t1-short '
                         'or any @register_gen_profile name); layers: plugin default < '