Merge pull request #30 from modelstudioai/feat/model-recommend

add model recommend cli commend

Author: XXPermanentXX

README.md

@@ -34,6 +34,7 @@ Equip your AI Agent out-of-the-box with these capabilities, composable across co
 - **App calls** — Invoke agents and workflows already published on Aliyun Model Studio
 - **MCP integration** — Orchestrate Bailian MCP servers: list services, inspect tools, and invoke any tool directly from the terminal
 - **Web search** — Real-time internet retrieval for up-to-date, accurate answers
+- **Model recommendation** — Describe your scenario and get best-fit model suggestions; supports scoped search, model comparison, and alternative discovery
 - **Console capabilities** — Browse Bailian apps (`app list`) and check free-tier quota (`usage free`)
 - **Local file auto-upload** — Every URL parameter accepts a local path; uploaded to free temp storage with 48-hour validity
 
@@ -98,6 +99,12 @@ bl image generate --prompt "A cat in a spacesuit" --out-dir ./images/
 # Generate a video from local image
 bl video generate --image ./cat.png --prompt "Make the cat move" --download cat.mp4
 
+# Model recommendation — find the best model for your use case
+bl advisor recommend --message "I need a visual-understanding chatbot"
+
+# Compare specific models
+bl advisor recommend --message "qwen-max vs deepseek-v3 for code generation"
+
 # Browser login (required for console capability commands)
 bl auth login --console
 

README_CN.md

@@ -34,6 +34,7 @@ _专为 AI Agent 打造，每个命令均可作为结构化工具调用。_
 - **应用调用** — 调用已发布在阿里云百炼平台上的智能体与工作流应用
 - **MCP 集成** — 统一调度百炼 MCP 服务：列出服务、查看工具、直接在终端调用任意工具
 - **联网搜索** — 实时互联网信息检索，提升回答准确性及时效性
+- **模型推荐** — 描述你的场景，智能推荐最适合的模型；支持限定范围搜索、模型对比和替代发现
 - **控制台能力** — 浏览百炼应用（`app list`），查询模型免费额度（`usage free`）
 - **本地文件自动上传** — 所有 URL 参数同时支持本地路径，免费临时存储 48 小时
 
@@ -93,6 +94,12 @@ bl image generate --prompt "一只穿太空服的猫在火星上" --out-dir ./im
 # 图生视频（本地文件自动上传）
 bl video generate --image ./cat.png --prompt "让画面中的猫动起来" --download cat.mp4
 
+# 模型推荐 — 根据场景推荐最适合的模型
+bl advisor recommend --message "我要做一个能理解图片的客服机器人"
+
+# 对比特定模型
+bl advisor recommend --message "qwen-max 和 deepseek-v3 哪个更适合做代码生成"
+
 # 浏览器登录（控制台能力相关命令需要）
 bl auth login --console
 

packages/cli/package.json

@@ -43,7 +43,9 @@
     "check": "vp check"
   },
   "dependencies": {
-    "bailian-cli-core": "workspace:*"
+    "bailian-cli-core": "workspace:*",
+    "boxen": "catalog:",
+    "chalk": "catalog:"
   },
   "devDependencies": {
     "@clack/prompts": "^0.7.0",

packages/cli/src/commands/advisor/recommend.ts

@@ -0,0 +1,321 @@
+import {
+  analyzeIntent,
+  buildDocLink,
+  type Config,
+  defineCommand,
+  detectOutputFormat,
+  type GetModelsOptions,
+  type GlobalFlags,
+  getModels,
+  type IntentProfile,
+  isInteractive,
+  type PipelineStep,
+  type RecommendedModel,
+  type RecommendResult,
+  rankModels,
+  recallSemantic,
+} from "bailian-cli-core";
+import boxen from "boxen";
+import chalk, { Chalk, type ChalkInstance } from "chalk";
+import { emitBare, emitResult } from "../../output/output.ts";
+import { createSpinner } from "../../output/progress.ts";
+import { failIfMissing, promptText } from "../../output/prompt.ts";
+
+function formatContextWindow(tokens: number): string {
+  if (tokens >= 1_000_000)
+    return `${(tokens / 1_000_000).toFixed(tokens % 1_000_000 === 0 ? 0 : 1)}M`;
+  if (tokens >= 1_000) return `${(tokens / 1_000).toFixed(tokens % 1_000 === 0 ? 0 : 1)}K`;
+  return String(tokens);
+}
+
+const MODALITY_LABELS: Record<string, string> = {
+  Text: "文本",
+  Image: "图片",
+  Video: "视频",
+  Audio: "音频",
+};
+const CAPABILITY_LABELS: Record<string, string> = {
+  TG: "文本生成",
+  VU: "视觉理解",
+  IG: "图像生成",
+  VG: "视频生成",
+  TTS: "语音合成",
+  ASR: "语音识别",
+  Reasoning: "推理",
+};
+const BUDGET_LABELS: Record<string, string> = {
+  low: "低成本优先",
+  medium: "适中",
+  high: "高投入",
+};
+const QUALITY_LABELS: Record<string, string> = {
+  flagship: "旗舰优先",
+  balanced: "均衡",
+  "cost-optimized": "性价比优先",
+};
+const PREFERENCE_MODE_LABELS: Record<string, string> = {
+  scoped: "限定范围",
+  comparison: "对比评估",
+  alternative: "替代推荐",
+};
+
+function formatIntentSummary(intent: IntentProfile, noColor: boolean): string {
+  const colorize = noColor ? new Chalk({ level: 0 }) : chalk;
+
+  const lines: string[] = [];
+  lines.push(colorize.cyan.bold("需求理解"));
+
+  if (intent.taskSummary) {
+    lines.push("");
+    lines.push(intent.taskSummary);
+  }
+
+  if (intent.scenarioHints.length) {
+    lines.push("");
+    lines.push(`${colorize.dim("场景特征")}  ${intent.scenarioHints.join(" · ")}`);
+  }
+
+  const inputLabels = intent.inputModality.map((mod) => MODALITY_LABELS[mod] ?? mod);
+  const outputLabels = intent.outputModality.map((mod) => MODALITY_LABELS[mod] ?? mod);
+  if (inputLabels.length || outputLabels.length) {
+    lines.push("");
+    const parts: string[] = [];
+    if (inputLabels.length) parts.push(`${colorize.dim("输入")} ${inputLabels.join(", ")}`);
+    if (outputLabels.length) parts.push(`${colorize.dim("输出")} ${outputLabels.join(", ")}`);
+    lines.push(parts.join("    "));
+  }
+
+  const capLabels = intent.requiredCapabilities.map((cap) => CAPABILITY_LABELS[cap] ?? cap);
+  if (capLabels.length) {
+    lines.push(`${colorize.dim("所需能力")}  ${capLabels.join(", ")}`);
+  }
+
+  const budgetLabel = BUDGET_LABELS[intent.budget] ?? intent.budget;
+  const qualityLabel = QUALITY_LABELS[intent.qualityPreference] ?? intent.qualityPreference;
+  lines.push("");
+  lines.push(
+    `${colorize.dim("预算倾向")} ${budgetLabel}    ${colorize.dim("质量偏好")} ${qualityLabel}`,
+  );
+
+  const preference = intent.modelPreference;
+  if (preference && preference.mode !== "unconstrained") {
+    lines.push("");
+    const modeLabel = PREFERENCE_MODE_LABELS[preference.mode] ?? preference.mode;
+    const prefParts = [colorize.dim("推荐模式") + ` ${colorize.yellow(modeLabel)}`];
+    if (preference.targets?.length) {
+      prefParts.push(colorize.dim("目标") + ` ${preference.targets.join(", ")}`);
+    }
+    if (preference.excludes?.length) {
+      prefParts.push(colorize.dim("排除") + ` ${preference.excludes.join(", ")}`);
+    }
+    lines.push(prefParts.join("    "));
+  }
+
+  if (intent.segments?.length) {
+    lines.push("");
+    lines.push(colorize.dim("任务拆解"));
+    for (const [idx, segment] of intent.segments.entries()) {
+      const outMods = segment.outputModality.map((mod) => MODALITY_LABELS[mod] ?? mod).join(", ");
+      lines.push(
+        `  ${colorize.dim(`${idx + 1}.`)} ${segment.step}${outMods ? colorize.dim(` → ${outMods}`) : ""}`,
+      );
+    }
+  }
+
+  return boxen(lines.join("\n"), {
+    padding: { top: 0, bottom: 0, left: 1, right: 1 },
+    margin: { top: 0, bottom: 0, left: 1, right: 0 },
+    borderColor: "cyan",
+    borderStyle: "round",
+    dimBorder: true,
+  });
+}
+
+const RECOMMEND_LABELS = ["最佳推荐", "次优选择", "备选参考"];
+
+function renderCard(rec: RecommendedModel, index: number, colorize: ChalkInstance): string {
+  const labelColors = [colorize.green.bold, colorize.blue.bold, colorize.magenta.bold];
+  const colorFn = labelColors[index] ?? colorize.white.bold;
+  const label = RECOMMEND_LABELS[index] ?? `推荐 #${index + 1}`;
+
+  const lines: string[] = [];
+  lines.push(colorFn(`⬢ 推荐 #${index + 1} — ${label}`));
+  lines.push("");
+  lines.push(`${colorize.bold(rec.name)}  ${colorize.dim(`(${rec.model})`)}`);
+  lines.push("");
+  lines.push(`${colorize.cyan("推荐理由")}  ${rec.reason}`);
+
+  if (rec.highlights.length) {
+    lines.push("");
+    lines.push(
+      rec.highlights.map((highlight) => colorize.bgGray.white(` ${highlight} `)).join(" "),
+    );
+  }
+
+  const meta: string[] = [];
+  if (rec.contextWindow) meta.push(`上下文 ${formatContextWindow(rec.contextWindow)}`);
+  if (rec.maxOutputTokens) meta.push(`最大输出 ${formatContextWindow(rec.maxOutputTokens)}`);
+  if (meta.length) {
+    lines.push("");
+    lines.push(colorize.dim(meta.join("  ·  ")));
+  }
+
+  const docLink = buildDocLink(rec.docUrl);
+  if (docLink) {
+    lines.push("");
+    lines.push(colorize.dim(`文档  ${docLink}`));
+  }
+
+  return boxen(lines.join("\n"), {
+    padding: { top: 0, bottom: 0, left: 1, right: 1 },
+    margin: { top: 0, bottom: 0, left: 1, right: 0 },
+    borderColor: "gray",
+    borderStyle: "round",
+    dimBorder: true,
+  });
+}
+
+function formatSingleResult(results: RecommendedModel[], noColor: boolean): string {
+  const colorize = noColor ? new Chalk({ level: 0 }) : chalk;
+  return results.map((rec, idx) => renderCard(rec, idx, colorize)).join("\n");
+}
+
+function formatPipelineResult(summary: string, steps: PipelineStep[], noColor: boolean): string {
+  const colorize = noColor ? new Chalk({ level: 0 }) : chalk;
+  const lines: string[] = [];
+  lines.push(`  ${colorize.yellow.bold("⚡ 组合方案")}  ${summary}`);
+
+  for (const [stepIdx, { step, recommendations, warnings }] of steps.entries()) {
+    lines.push("");
+    lines.push(colorize.bold(`  ━━━ Step ${stepIdx + 1}: ${step} ━━━`));
+
+    if (warnings?.length) {
+      for (const warning of warnings) {
+        lines.push(`  ${colorize.yellow("⚠")} ${colorize.yellow(warning)}`);
+      }
+    }
+
+    lines.push("");
+    lines.push(recommendations.map((rec, idx) => renderCard(rec, idx, colorize)).join("\n"));
+  }
+
+  return lines.join("\n");
+}
+
+function formatResult(result: RecommendResult, noColor: boolean): string {
+  if (result.type === "pipeline") {
+    return formatPipelineResult(result.summary, result.steps, noColor);
+  }
+  return formatSingleResult(result.recommendations, noColor);
+}
+
+function isEmptyResult(result: RecommendResult): boolean {
+  if (result.type === "pipeline") return result.steps.length === 0;
+  return result.recommendations.length === 0;
+}
+
+export default defineCommand({
+  name: "advisor recommend",
+  description:
+    "Recommend the best models for your use case (intent analysis → candidate recall → LLM ranking)",
+  usage: "bl advisor recommend <prompt> [flags]",
+  options: [
+    {
+      flag: "--message <text>",
+      description: "Describe your requirements (alternative to positional prompt)",
+    },
+    {
+      flag: "--dry-run",
+      description: "Show intent analysis and candidate list without LLM ranking",
+    },
+    {
+      flag: "--output <format>",
+      description: "Output format: text (default in TTY), json, yaml",
+    },
+  ],
+  examples: [
+    'bl advisor recommend --message "我要做一个能理解图片的客服机器人"',
+    'bl advisor recommend --message "做一个Agent自动根据用户意图生成动画片"',
+    'bl advisor recommend --message "法律合同审查，要求高精准度"',
+    'bl advisor recommend --message "做一个低成本高并发的在线客服" --output json',
+    'bl advisor recommend --message "长文本摘要" --dry-run',
+    "bl advisor recommend                                          # 交互式输入需求",
+  ],
+  async run(config: Config, flags: GlobalFlags) {
+    const positional = ((flags as Record<string, unknown>)._positional as string[]) ?? [];
+    let userInput = (flags.message as string) || positional.join(" ");
+
+    if (!userInput.trim()) {
+      if (isInteractive({ nonInteractive: config.nonInteractive })) {
+        const hint = await promptText({ message: "描述你的需求:" });
+        if (!hint) {
+          process.stderr.write("已取消。\n");
+          process.exit(1);
+        }
+        userInput = hint;
+      } else {
+        failIfMissing("message", 'bl advisor recommend "你的需求"');
+      }
+    }
+
+    const top = 3;
+    const format = detectOutputFormat(config.output);
+
+    const modelsOptions: GetModelsOptions = {
+      onPrepareStart: () => process.stderr.write("初始化中...\n"),
+    };
+    process.stderr.write("正在分析需求...\n");
+    const [allModels, intent] = await Promise.all([
+      getModels(config, modelsOptions),
+      analyzeIntent(config, userInput),
+    ]);
+
+    if (intent.confidence === 0) {
+      process.stderr.write("需求分析超时，使用默认参数继续...\n");
+    } else {
+      process.stderr.write("\n");
+    }
+
+    // Stage 2: Candidate Recall (semantic recall, auto-builds embeddings on first run)
+    const candidates = await recallSemantic(config, allModels, userInput, 50, intent);
+
+    if (config.dryRun) {
+      emitResult(
+        {
+          userInput,
+          intent,
+          candidateCount: candidates.length,
+          candidates: candidates.map(({ model, score }) => ({
+            model: model.model,
+            score,
+          })),
+          top,
+        },
+        format,
+      );
+      return;
+    }
+
+    // Stage 3: LLM Ranking
+    const spinner = createSpinner("正在推荐最佳模型...");
+    spinner.start();
+
+    const result = await rankModels(config, candidates, intent, userInput, top);
+
+    spinner.stop();
+
+    if (isEmptyResult(result)) {
+      emitBare("暂无满足该需求的模型。");
+      return;
+    }
+
+    if (format !== "text") {
+      emitResult(result, format);
+      return;
+    }
+
+    emitBare(formatIntentSummary(intent, config.noColor));
+    emitBare("");
+    emitBare(formatResult(result, config.noColor));
+  },
+});

packages/cli/src/commands/catalog.ts

@@ -38,6 +38,7 @@ import consoleCall from "./console/call.ts";
 import usageFree from "./usage/free.ts";
 import pipelineRun from "./pipeline/run.ts";
 import pipelineValidate from "./pipeline/validate.ts";
+import advisorRecommend from "./advisor/recommend.ts";
 
 /** Command registry map (no dependency on registry.ts — safe for build-time import). */
 export const commands: Record<string, Command> = {
@@ -78,5 +79,6 @@ export const commands: Record<string, Command> = {
   "config show": configShow,
   "config set": configSet,
   "config export-schema": configExportSchema,
+  "advisor recommend": advisorRecommend,
   update: update,
 };

packages/cli/src/main.ts

@@ -66,7 +66,8 @@ const NO_AUTH_SETUP = [
 ];
 
 async function main() {
-  const argv = process.argv.slice(2);
+  let argv = process.argv.slice(2);
+  if (argv[0] === "--") argv = argv.slice(1);
 
   if (argv.includes("--version") || argv.includes("-v")) {
     process.stdout.write(`bl ${CLI_VERSION}\n`);