dsh-livebench-panel 0.1.7 → 0.1.9

package/lib/client.js

@@ -98,10 +98,12 @@ window.__ModuleLoader__.load({
 			const [log, setLog] = useState("");
 			const [exitCode, setExitCode] = useState(null);
 			const [startError, setStartError] = useState(null);
+			const [homeInput, setHomeInput] = useState("");
+			const [homeBusy, setHomeBusy] = useState(false);
 			const [results, setResults] = useState(null);
-			const [pickedTasks, setPickedTasks] = useState(() => new Set());
-			const [taskOrder, setTaskOrder] = useState(() => {
-				try { return JSON.parse(localStorage.getItem("dlb_task_order_v1")) ?? []; } catch { return []; }
+			const [pickedModels, setPickedModels] = useState(() => new Set());
+			const [modelOrder, setModelOrder] = useState(() => {
+				try { return JSON.parse(localStorage.getItem("dlb_model_order_v1")) ?? []; } catch { return []; }
 			});
 			const dragKey = useRef(null);
 			const logRef = useRef(null);
@@ -253,6 +255,24 @@ window.__ModuleLoader__.load({
 				await refreshStatus();
 			};
 
+			const saveHome = async () => {
+				setHomeBusy(true);
+				setStartError(null);
+				try {
+					const payload = await api("/home", {
+						method: "POST",
+						headers: { "content-type": "application/json" },
+						body: JSON.stringify({ path: homeInput }),
+					});
+					if (!payload.ok) setStartError(payload.error ?? "设置失败");
+					else { setHomeInput(""); await loadConfig(); }
+				} catch (error) {
+					setStartError(String(error.message ?? error));
+				} finally {
+					setHomeBusy(false);
+				}
+			};
+
 			const setField = (key) => (event) => {
 				const value = event.target.value;
 				setSel((prev) => {
@@ -265,78 +285,75 @@ window.__ModuleLoader__.load({
 				});
 			};
 
-			// ---- 成绩矩阵：task 为行（含 category 行标）、model 为列 ----
-			// 附带：time 列（该任务评测开始时间 = 行内最早判分时间）、
+			// ---- 成绩矩阵：model 为行标识、category/task 为列标识 ----
+			// 附带：time 列（该模型评测开始时间 = 行内最早判分时间）、
 			// 行拖拽排序（localStorage 持久化）、行选择与单行/批量删除。
 			const taskKey = (category, task) => `${category}/${task}`;
 			const matrix = useMemo(() => {
 				if (!results) return null;
 				const models = [...new Set(results.rows.map((r) => r.model))].sort();
-				const byTask = new Map(); // `${cat}/${task}` -> {category, task, cells: Map(model->row), startTime}
+				const tasks = [...new Set(results.rows.map((r) => taskKey(r.category, r.task)))].sort();
+				const cells = new Map(); // `${model}\u0000${taskKey}` -> row
+				const startTimes = new Map(); // model -> min tstamp
 				for (const row of results.rows) {
-					const key = taskKey(row.category, row.task);
-					const entry = byTask.get(key) ?? { category: row.category, task: row.task, cells: new Map(), startTime: Infinity };
-					entry.cells.set(row.model, row);
-					if (row.time && row.time < entry.startTime) entry.startTime = row.time;
-					byTask.set(key, entry);
+					cells.set(`${row.model}\u0000${taskKey(row.category, row.task)}`, row);
+					const prev = startTimes.get(row.model);
+					if (row.time && (prev === undefined || row.time < prev)) startTimes.set(row.model, row.time);
 				}
-				return { models, tasks: [...byTask.entries()] };
+				return { models, tasks, cells, startTimes };
 			}, [results]);
-			const sortedTasks = useMemo(() => {
+			const sortedModels = useMemo(() => {
 				if (!matrix) return [];
-				const indexOf = new Map(taskOrder.map((key, index) => [key, index]));
-				return [...matrix.tasks].sort((a, b) => {
-					const ia = indexOf.has(a[0]) ? indexOf.get(a[0]) : Number.MAX_SAFE_INTEGER;
-					const ib = indexOf.has(b[0]) ? indexOf.get(b[0]) : Number.MAX_SAFE_INTEGER;
+				const indexOf = new Map(modelOrder.map((key, index) => [key, index]));
+				return [...matrix.models].sort((a, b) => {
+					const ia = indexOf.has(a) ? indexOf.get(a) : Number.MAX_SAFE_INTEGER;
+					const ib = indexOf.has(b) ? indexOf.get(b) : Number.MAX_SAFE_INTEGER;
 					if (ia !== ib) return ia - ib;
-					return a[0].localeCompare(b[0]);
+					return a.localeCompare(b);
 				});
-			}, [matrix, taskOrder]);
+			}, [matrix, modelOrder]);
 
-			const persistTaskOrder = (keys) => {
-				setTaskOrder(keys);
-				try { localStorage.setItem("dlb_task_order_v1", JSON.stringify(keys)); } catch { /* private mode */ }
+			const persistModelOrder = (keys) => {
+				setModelOrder(keys);
+				try { localStorage.setItem("dlb_model_order_v1", JSON.stringify(keys)); } catch { /* private mode */ }
 			};
 
-			const onRowDrop = (targetKey) => {
-				const sourceKey = dragKey.current;
+			const onRowDrop = (targetModel) => {
+				const sourceModel = dragKey.current;
 				dragKey.current = null;
-				if (!sourceKey || sourceKey === targetKey) return;
-				const keys = sortedTasks.map(([key]) => key);
-				const from = keys.indexOf(sourceKey);
-				const to = keys.indexOf(targetKey);
+				if (!sourceModel || sourceModel === targetModel) return;
+				const keys = sortedModels.slice();
+				const from = keys.indexOf(sourceModel);
+				const to = keys.indexOf(targetModel);
 				if (from < 0 || to < 0) return;
 				keys.splice(to, 0, keys.splice(from, 1)[0]);
-				persistTaskOrder(keys);
+				persistModelOrder(keys);
 			};
 
-			const togglePicked = (key) => setPickedTasks((prev) => {
+			const togglePicked = (model) => setPickedModels((prev) => {
 				const next = new Set(prev);
-				if (next.has(key)) next.delete(key); else next.add(key);
+				if (next.has(model)) next.delete(model); else next.add(model);
 				return next;
 			});
 
-			// 删除一个任务行 = 删除该任务下所有模型的判分与答案记录
-			const deleteTasks = async (keys) => {
-				if (!results || keys.length === 0) return;
+			// 删除一个模型行 = 删除该模型在所有任务下的判分与答案记录
+			const deleteModels = async (modelsToDelete) => {
+				if (!results || modelsToDelete.length === 0) return;
 				const serverRows = [];
-				for (const key of keys) {
-					const [category, task] = key.split("/");
-					for (const row of results.rows) {
-						if (row.category === category && row.task === task) {
-							serverRows.push({ model: row.model, category, task });
-						}
+				for (const row of results.rows) {
+					if (modelsToDelete.includes(row.model)) {
+						serverRows.push({ model: row.model, category: row.category, task: row.task });
 					}
 				}
 				if (serverRows.length === 0) return;
-				if (!window.confirm(`确认删除所选 ${keys.length} 个任务的全部评测成绩（共 ${serverRows.length} 条模型记录，含答案）？不可恢复。`)) return;
+				if (!window.confirm(`确认删除所选 ${modelsToDelete.length} 个模型的全部评测成绩（共 ${serverRows.length} 条记录，含答案）？不可恢复。`)) return;
 				await api("/delete", {
 					method: "POST",
 					headers: { "content-type": "application/json" },
 					body: JSON.stringify({ rows: serverRows }),
 				});
-				setPickedTasks(new Set());
-				persistTaskOrder(taskOrder.filter((key) => !keys.includes(key)));
+				setPickedModels(new Set());
+				persistModelOrder(modelOrder.filter((model) => !modelsToDelete.includes(model)));
 				await loadResults();
 			};
 
@@ -350,10 +367,21 @@ window.__ModuleLoader__.load({
 			return h("div", { className: c("root") },
 				h("div", { className: c("head") },
 					h("h2", null, "LiveBench"),
-					h("span", { className: c("sub") }, "LiveBench 评测面板 · 基于 V:\\PythonProject\\C_UtilizeSpace\\LiveBench"),
+					h("span", { className: c("sub") }, config?.root ? `LiveBench 评测面板 · ${config.root}` : "LiveBench 评测面板"),
 					h("span", { className: c("badge"), "data-ok": config?.available ? "1" : "0" },
 						config === null ? "检测中…" : config.available ? "LiveBench 就绪" : "未找到 LiveBench"),
 				),
+				h("div", { className: c("row") },
+					h("span", { className: c("label") }, "LiveBench 路径"),
+					h("input", {
+						className: c("input"), style: { flex: "1", minWidth: "240px" },
+						placeholder: config?.root ?? "V:\\…\\LiveBench 或 ~/LiveBench",
+						value: homeInput, onChange: (event) => setHomeInput(event.target.value),
+					}),
+					h("button", { className: c("btnGhost") + " " + c("btn"), disabled: homeBusy || homeInput.trim().length === 0, onClick: saveHome },
+						homeBusy ? "保存中…" : "设置路径"),
+					h("span", { className: c("hint") }, "需含 livebench\\run_livebench.py 与 .venv；也可用环境变量 DSH_LIVEBENCH_HOME"),
+				),
 				configError !== null && h("p", { className: c("error") }, `加载配置失败：${configError}`),
 				h("div", { className: c("card") },
 					h("div", { className: c("grid") },
@@ -420,7 +448,7 @@ window.__ModuleLoader__.load({
 						running && h("button", { className: c("btn") + " " + c("btnDanger"), onClick: onStop }, "停止"),
 						h("button", { className: c("btnGhost") + " " + c("btn"), onClick: () => { loadConfig(); loadResults(); } }, "刷新"),
 						selectedProvider && !selectedProvider.routable && h("span", { className: c("hint") },
-							"该 provider 未配置 openai-completions baseURL，LiveBench 将按模型名原生路由（未注册的模型名会失败）。"),
+							"该 provider 的协议或端点未知，无法自动路由：LiveBench 将按模型名原生尝试，未注册的模型名会失败。"),
 					),
 					startError !== null && h("p", { className: c("error") }, startError),
 				),
@@ -431,49 +459,53 @@ window.__ModuleLoader__.load({
 					),
 					h("pre", { className: c("log"), ref: logRef }, log || "（暂无输出）"),
 				),
-				sortedTasks.length > 0 && h("div", { className: c("card") },
+				sortedModels.length > 0 && h("div", { className: c("card") },
 					h("div", { className: c("row") },
-						h("span", { className: c("label") }, "评测成绩（分数 = 平均分 ×100，括号内为 已判/总题数；行首可拖动排序，顺序保存在本机）"),
+						h("span", { className: c("label") }, "评测成绩（model 为行；分数 = 平均分 ×100；行首可拖动排序，顺序保存在本机）"),
 						h("button", {
 							className: c("btn") + " " + c("btnDanger"),
-							disabled: pickedTasks.size === 0,
-							onClick: () => deleteTasks([...pickedTasks]),
-						}, `删除所选（${pickedTasks.size}）`),
+							disabled: pickedModels.size === 0,
+							onClick: () => deleteModels([...pickedModels]),
+						}, `删除所选（${pickedModels.size}）`),
 					),
 					h("table", { className: c("table") },
 						h("thead", null, h("tr", null,
 							h("th", { style: { width: "26px" } }, "⠿"),
 							h("th", { style: { width: "30px" } },
-								h("input", { type: "checkbox", checked: pickedTasks.size > 0 && pickedTasks.size === sortedTasks.length,
-									onChange: (event) => setPickedTasks(event.target.checked ? new Set(sortedTasks.map(([key]) => key)) : new Set()) })),
-							h("th", null, "task"),
-							matrix.models.map((model) => h("th", { key: model }, model)),
+								h("input", { type: "checkbox", checked: pickedModels.size > 0 && pickedModels.size === sortedModels.length,
+									onChange: (event) => setPickedModels(event.target.checked ? new Set(sortedModels) : new Set()) })),
+							h("th", null, "model"),
+							matrix.tasks.map((taskKeyCol) => {
+								const [cat, task] = taskKeyCol.split("/");
+								return h("th", { key: taskKeyCol, title: `${cat} / ${task}` }, task);
+							}),
 							h("th", null, "删除"),
 							h("th", null, "time"),
 						)),
 						h("tbody", null,
-							sortedTasks.map(([key, entry]) => {
-								const checked = pickedTasks.has(key);
+							sortedModels.map((model) => {
+								const checked = pickedModels.has(model);
+								const startTime = matrix.startTimes.get(model);
 								return h("tr", {
-									key,
+									key: model,
 									draggable: true,
-									onDragStart: () => { dragKey.current = key; },
+									onDragStart: () => { dragKey.current = model; },
 									onDragOver: (event) => event.preventDefault(),
-									onDrop: () => onRowDrop(key),
+									onDrop: () => onRowDrop(model),
 								},
 									h("td", { className: c("drag"), title: "拖动排序" }, "⠿"),
-									h("td", null, h("input", { type: "checkbox", checked, onChange: () => togglePicked(key) })),
-									h("td", null, h("span", { title: `${entry.category} / ${entry.task}` }, `${entry.category} / ${entry.task}`)),
-									matrix.models.map((model) => {
-										const row = entry.cells.get(model);
-										return h("td", { key: model, className: c("score") },
+									h("td", null, h("input", { type: "checkbox", checked, onChange: () => togglePicked(model) })),
+									h("td", null, model),
+									matrix.tasks.map((taskKeyCol) => {
+										const row = matrix.cells.get(`${model}\u0000${taskKeyCol}`);
+										return h("td", { key: taskKeyCol, className: c("score") },
 											row ? `${row.score.toFixed(1)} (${row.judged}/${row.total})` : "—");
 									}),
 									h("td", null, h("button", {
-										className: c("btnGhost") + " " + c("btn"), title: "删除该任务下所有模型的评测成绩",
-										onClick: () => deleteTasks([key]),
+										className: c("btnGhost") + " " + c("btn"), title: "删除该模型的所有评测成绩",
+										onClick: () => deleteModels([model]),
 									}, "删除")),
-									h("td", null, fmtTime(entry.startTime === Infinity ? null : entry.startTime)),
+									h("td", null, fmtTime(startTime)),
 								);
 							})),
 					),

package/lib/index.js

@@ -39,8 +39,61 @@ const inject = ["webServer"];
 const API = "/dsh-livebench-panel/api";
 /** Profile whose settings.yaml holds the harness provider config. */
 const PROFILE = "web";
-/** LiveBench checkout used unless DSH_LIVEBENCH_HOME overrides it. */
-const DEFAULT_LIVEBENCH_HOME = "V:\\PythonProject\\C_UtilizeSpace\\LiveBench";
+/** LiveBench location resolution: saved user path > env var > user home > legacy. */
+const CONFIG_FILE = join(process.env.DSH_HOME ?? join(homedir(), ".dsh"), "livebench-panel.json");
+const LEGACY_LIVEBENCH_HOME = "V:\\PythonProject\\C_UtilizeSpace\\LiveBench";
+
+/** Read the user-saved LiveBench path (set from the panel UI). */
+function savedLivebenchHome() {
+	try {
+		const path = JSON.parse(readFileSync(CONFIG_FILE, "utf8")).livebenchHome;
+		return typeof path === "string" && path.trim().length > 0 ? path.trim() : null;
+	} catch {
+		return null;
+	}
+}
+
+/** Candidate roots, most explicit first. Windows and Linux both covered. */
+function livebenchCandidates() {
+	return [
+		savedLivebenchHome(),
+		process.env.DSH_LIVEBENCH_HOME ?? null,
+		join(homedir(), "LiveBench"),
+		LEGACY_LIVEBENCH_HOME,
+	].filter((candidate) => typeof candidate === "string" && candidate.length > 0);
+}
+
+/** A root counts as a LiveBench checkout when the venv python and entry script exist. */
+function isLivebenchRoot(root) {
+	return existsSync(join(root, ".venv", "Scripts", "python.exe")) || existsSync(join(root, ".venv", "bin", "python"))
+		? existsSync(join(root, "livebench", "run_livebench.py"))
+		: false;
+}
+
+/** Resolve the LiveBench layout from the first valid candidate. */
+function livebenchLayout() {
+	const candidates = livebenchCandidates();
+	for (const root of candidates) {
+		const pythonExe = join(root, ".venv", "Scripts", process.platform === "win32" ? "python.exe" : "python");
+		if (existsSync(pythonExe) && existsSync(join(root, "livebench", "run_livebench.py"))) {
+			return {
+				root,
+				livebenchDir: join(root, "livebench"),
+				dataDir: join(root, "livebench", "data", "live_bench"),
+				pythonExe,
+				available: true,
+			};
+		}
+	}
+	const root = candidates[0] ?? join(homedir(), "LiveBench");
+	return {
+		root,
+		livebenchDir: join(root, "livebench"),
+		dataDir: join(root, "livebench", "data", "live_bench"),
+		pythonExe: join(root, ".venv", process.platform === "win32" ? join("Scripts", "python.exe") : join("bin", "python")),
+		available: false,
+	};
+}
 /** Question-set releases LiveBench accepts (mirrors livebench/common.py). */
 const RELEASES = [
 	"2024-06-24", "2024-07-26", "2024-08-31", "2024-11-25",
@@ -123,10 +176,10 @@ function readProviders(profileDir) {
 	if (piAiData !== null) {
 		for (const provider of providers) {
 			if (provider.baseURL === null) {
-				const base = builtinBaseUrl(piAiData, provider.id);
-				if (base !== null) {
-					provider.baseURL = base;
-					provider.api = provider.api ?? "openai-completions";
+				const info = builtinProtocolInfo(piAiData, provider.id);
+				if (info) {
+					provider.baseURL = info.baseURL;
+					provider.api = info.api;
 				}
 			}
 		}
@@ -162,17 +215,25 @@ function resolvePiAiDataDir(profileDir) {
 	return null;
 }
 
-/** Read one provider's built-in OpenAI-compatible baseUrl from pi-ai data. */
-function builtinBaseUrl(piAiDataDir, providerId) {
+/**
+ * Read one provider's built-in protocol + baseUrl from pi-ai data.
+ * Prefers OpenAI Chat Completions; Anthropic Messages is also routable
+ * (LiveBench talks it natively and honors ANTHROPIC_BASE_URL/ANTHROPIC_API_KEY).
+ */
+function builtinProtocolInfo(piAiDataDir, providerId) {
 	if (!/^[a-z0-9-]+$/.test(providerId)) return null;
 	const file = join(piAiDataDir, `${providerId}.json`);
 	if (!existsSync(file)) return null;
 	try {
 		const data = JSON.parse(readFileSync(file, "utf8"));
-		const openai = data?.["openai-completions"];
-		if (openai && typeof openai === "object") {
-			for (const entry of Object.values(openai)) {
-				if (entry && typeof entry.baseUrl === "string" && entry.baseUrl.length > 0) return entry.baseUrl;
+		for (const api of ["openai-completions", "anthropic-messages"]) {
+			const group = data?.[api];
+			if (group && typeof group === "object") {
+				for (const entry of Object.values(group)) {
+					if (entry && typeof entry.baseUrl === "string" && entry.baseUrl.length > 0) {
+						return { api, baseURL: entry.baseUrl };
+					}
+				}
 			}
 		}
 	} catch { /* unreadable data file — treat as unknown */ }
@@ -210,19 +271,6 @@ function effortsOfModel(model) {
 	return efforts;
 }
 
-/** LiveBench layout on disk. */
-function livebenchLayout() {
-	const root = process.env.DSH_LIVEBENCH_HOME ?? DEFAULT_LIVEBENCH_HOME;
-	const venvPython = join(root, ".venv", "Scripts", "python.exe");
-	return {
-		root,
-		livebenchDir: join(root, "livebench"),
-		dataDir: join(root, "livebench", "data", "live_bench"),
-		pythonExe: venvPython,
-		available: existsSync(venvPython) && existsSync(join(root, "livebench", "run_livebench.py")),
-	};
-}
-
 /** Scan data/live_bench/<category>/<task>/question.jsonl into a category→tasks map. */
 function scanCategoryTasks(dataDir) {
 	const tasks = {};
@@ -315,21 +363,32 @@ function displayModelName(providerId, modelId) {
  * The file is regenerated on every start; secrets never go in here.
  * @returns {string|null} error message, or null on success.
  */
-function writeGeneratedModelConfig(layout, YAML, { displayName, modelId, reasoningEffort }) {
+function writeGeneratedModelConfig(layout, YAML, { displayName, modelId, reasoningEffort, protocol }) {
 	const configDir = join(layout.livebenchDir, "model", "model_configs");
 	const target = join(configDir, "dsh_panel_generated.yaml");
 	void YAML;
-	const doc = [
+	const lines = [
 		"# Generated by dsh-livebench-panel — regenerated on every evaluation start.",
 		"---",
 		`display_name: ${displayName}`,
 		"api_name:",
-		"  local: " + modelId,
-		"api_kwargs:",
-		"  default:",
-		`    reasoning_effort: ${reasoningEffort}`,
-		"",
-	].join("\n");
+	];
+	if (protocol === "anthropic") {
+		// Route through LiveBench's native anthropic client; the endpoint is
+		// pointed at the provider proxy via ANTHROPIC_BASE_URL (spawn env).
+		lines.push(`  anthropic: ${modelId}`, "default_provider: anthropic");
+		if (reasoningEffort) {
+			// note: reasoning_effort is an OpenAI-style knob and is intentionally
+			// not forwarded on the anthropic protocol path.
+		}
+	} else {
+		lines.push(`  local: ${modelId}`);
+		if (reasoningEffort) {
+			lines.push("api_kwargs:", "  default:", `    reasoning_effort: ${reasoningEffort}`);
+		}
+	}
+	lines.push("");
+	const doc = lines.join("\n");
 	try {
 		if (!existsSync(configDir)) return `model_configs directory not found: ${configDir}`;
 		writeFileSync(target, doc, "utf8");
@@ -492,8 +551,19 @@ function apply(ctx) {
 		const hasEffortSuffix = reasoningEffort !== null && reasoningEffort !== "off";
 		const displayName = displayModelName(providerId || "direct", modelId) + (hasEffortSuffix ? "@" + reasoningEffort : "");
 		let cliModel = modelId;
-		if (hasEffortSuffix) {
-			const writeError = writeGeneratedModelConfig(layout, loadYaml(profileDir), { displayName, modelId, reasoningEffort });
+		let writeError = null;
+		// Anthropic-protocol proxies cannot go through --api-base (that path
+		// speaks OpenAI Chat Completions). Instead the generated model config
+		// selects LiveBench's native anthropic client and the spawn env points
+		// the SDK at the proxy (ANTHROPIC_BASE_URL / ANTHROPIC_API_KEY).
+		const isAnthropicRoute = provider && provider.api === "anthropic-messages" && provider.baseURL;
+		if (isAnthropicRoute || hasEffortSuffix) {
+			writeError = writeGeneratedModelConfig(layout, loadYaml(profileDir), {
+				displayName,
+				modelId,
+				reasoningEffort: isAnthropicRoute ? null : reasoningEffort,
+				protocol: isAnthropicRoute ? "anthropic" : "openai",
+			});
 			if (writeError) return { status: 500, payload: { ok: false, error: writeError } };
 			cliModel = displayName;
 		}
@@ -527,15 +597,13 @@ function apply(ctx) {
 			// livebench import (shortuuid etc.).
 			PATH: `${join(layout.root, ".venv", "Scripts")}${delimiter}${process.env.PATH ?? ""}`,
 		};
-		// Only OpenAI-compatible providers can be routed with --api-base; for
-		// those, hand the key over via env (never the command line). The key is
-		// resolved through the harness credential seam when available (values may
-		// live encrypted in .credentials.yaml rather than in the process env),
-		// falling back to the plain environment variable.
-		if (provider && provider.api === "openai-completions" && provider.baseURL) {
-			args.push("--api-base", provider.baseURL);
+		// Provider routing: keys are resolved through the harness credential seam
+		// when available (values may live encrypted in .credentials.yaml rather
+		// than in the process env), falling back to the plain environment
+		// variable. Secrets travel via env only — never the command line.
+		if (provider && provider.baseURL) {
+			let key;
 			if (provider.keyEnv) {
-				let key;
 				try {
 					const credentials = ctx.get ? ctx.get("credentials") : undefined;
 					if (credentials && typeof credentials.resolve === "function") {
@@ -544,7 +612,15 @@ function apply(ctx) {
 					}
 				} catch { /* credential seam unavailable — fall through to env */ }
 				if (!key && process.env[provider.keyEnv]) key = process.env[provider.keyEnv];
+			}
+			if (provider.api === "openai-completions") {
+				args.push("--api-base", provider.baseURL);
 				if (key) env.LIVEBENCH_API_KEY = key;
+			} else if (provider.api === "anthropic-messages") {
+				// The generated model config selects the native anthropic client;
+				// the Anthropic SDK picks endpoint+key up from these env vars.
+				if (key) env.ANTHROPIC_API_KEY = key;
+				env.ANTHROPIC_BASE_URL = provider.baseURL;
 			}
 		}
 
@@ -617,7 +693,7 @@ function apply(ctx) {
 				providers: providers.map(({ id, name: pname, models, api, baseURL }) => ({
 					id,
 					name: pname,
-					routable: api === "openai-completions" && typeof baseURL === "string" && baseURL.length > 0,
+					routable: (api === "openai-completions" || api === "anthropic-messages") && typeof baseURL === "string" && baseURL.length > 0,
 					baseURL: baseURL ?? null,
 					models,
 				})),
@@ -695,7 +771,14 @@ function apply(ctx) {
 				return;
 			}
 			try {
-				run.proc.kill();
+				if (process.platform === "win32") {
+					// proc.kill() only terminates run_livebench.py itself; its child
+					// (cmd → gen_api_answer.py → …) would survive. taskkill /T /F
+					// takes down the whole tree.
+					spawn("taskkill", ["/PID", String(run.proc.pid), "/T", "/F"], { windowsHide: true });
+				} else {
+					run.proc.kill();
+				}
 			} catch (error) {
 				sendJson(res, 500, { ok: false, error: String(error.message ?? error) });
 				return;
@@ -718,6 +801,45 @@ function apply(ctx) {
 		},
 	}), `${name}: results route`);
 
+	ctx.effect(() => ctx.webServer.register({
+		kind: "exact",
+		path: `${API}/home`,
+		handler: async (req, res) => {
+			if (req.method !== "POST") {
+				sendJson(res, 405, { ok: false, error: "method not allowed" });
+				return;
+			}
+			if (!originAllowed(req)) {
+				sendJson(res, 403, { ok: false, error: "forbidden origin" });
+				return;
+			}
+			let body;
+			try {
+				body = JSON.parse(await readBody(req));
+			} catch {
+				sendJson(res, 400, { ok: false, error: "invalid request body" });
+				return;
+			}
+			const path = typeof body.path === "string" ? body.path.trim() : "";
+			if (path.length === 0 || path.length > 260 || /[<>:"|?*\0]/.test(path.replace(/^[A-Za-z]:/, ""))) {
+				sendJson(res, 400, { ok: false, error: "invalid path" });
+				return;
+			}
+			if (!existsSync(join(path, "livebench", "run_livebench.py"))) {
+				sendJson(res, 400, { ok: false, error: `该目录下没有 livebench\\run_livebench.py：${path}` });
+				return;
+			}
+			try {
+				writeFileSync(CONFIG_FILE, JSON.stringify({ livebenchHome: path }, null, 2), "utf8");
+			} catch (error) {
+				sendJson(res, 500, { ok: false, error: `cannot save config: ${error.message}` });
+				return;
+			}
+			const layout = livebenchLayout();
+			sendJson(res, 200, { ok: true, root: layout.root, available: layout.available });
+		},
+	}), `${name}: home route`);
+
 	ctx.effect(() => ctx.webServer.register({
 		kind: "exact",
 		path: `${API}/delete`,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-livebench-panel",
-  "version": "0.1.7",
+  "version": "0.1.9",
   "description": "DSH web plugin: a LiveBench tab in the Trajectory view (right of 对话/轨迹). Run LiveBench evaluations against every model configured in the DeepSeek Harness — pick provider/model, category, task, release and question range from dropdowns, watch progress, and read scores in place.",
   "license": "MIT",
   "type": "module",