@easynet/agent-llm 1.0.2 → 1.0.3

package/dist/cli.js

@@ -0,0 +1,120 @@
+#!/usr/bin/env node
+import {
+  createChatModelFromLlmConfig,
+  loadLlmConfig
+} from "./chunk-3Z3KXKNU.js";
+
+// src/cli.ts
+import { join } from "path";
+import { tool } from "@langchain/core/tools";
+import { z } from "zod";
+import { HumanMessage, ToolMessage } from "@langchain/core/messages";
+var defaultLlmSection = {
+  provider: "openai",
+  model: process.env.OPENAI_MODEL ?? "gpt-4o-mini",
+  temperature: 0,
+  apiKey: process.env.OPENAI_API_KEY,
+  ...process.env.OPENAI_BASE_URL && { baseURL: process.env.OPENAI_BASE_URL }
+};
+function parseArgv() {
+  const args = process.argv.slice(2);
+  let configPath;
+  const rest = [];
+  for (let i = 0; i < args.length; i++) {
+    if (args[i] === "--config" && args[i + 1]) {
+      configPath = args[i + 1];
+      i++;
+    } else {
+      rest.push(args[i]);
+    }
+  }
+  const query = rest.join(" ").trim() || "hi";
+  return { configPath, query };
+}
+function loadLlmSection(configPath) {
+  const path = configPath ?? join(process.cwd(), "config", "llm.yaml");
+  const llm = loadLlmConfig(path);
+  return llm ?? defaultLlmSection;
+}
+var getWeather = tool(
+  (input) => {
+    const loc = input.location.toLowerCase();
+    if (["sf", "san francisco"].includes(loc)) {
+      return "It's 60\xB0F and foggy in San Francisco.";
+    }
+    if (["ny", "new york"].includes(loc)) {
+      return "It's 72\xB0F and partly cloudy in New York.";
+    }
+    return `Weather for ${input.location}: 70\xB0F and sunny.`;
+  },
+  {
+    name: "get_weather",
+    description: "Get the current weather for a location.",
+    schema: z.object({
+      location: z.string().describe("City or place name (e.g. SF, New York)")
+    })
+  }
+);
+var addNumbers = tool(
+  (input) => String(input.a + input.b),
+  {
+    name: "add_numbers",
+    description: "Add two numbers.",
+    schema: z.object({
+      a: z.number().describe("First number"),
+      b: z.number().describe("Second number")
+    })
+  }
+);
+var tools = [getWeather, addNumbers];
+var toolsByName = new Map(tools.map((t) => [t.name, t]));
+var MAX_TURNS = 10;
+async function runAgent(model, query) {
+  const withTools = model.bindTools?.(tools, { tool_choice: "auto" });
+  if (!withTools) throw new Error("Model does not support bindTools");
+  const messages = [new HumanMessage(query)];
+  for (let turn = 0; turn < MAX_TURNS; turn++) {
+    const response = await withTools.invoke(messages);
+    const aiMessage = response;
+    if (!aiMessage.tool_calls?.length) {
+      const content = typeof aiMessage.content === "string" ? aiMessage.content : Array.isArray(aiMessage.content) ? aiMessage.content.map((c) => "text" in c ? c.text : "").join("") : String(aiMessage.content ?? "");
+      return content;
+    }
+    messages.push(aiMessage);
+    for (const tc of aiMessage.tool_calls) {
+      const id = tc.id ?? `call_${turn}_${tc.name}`;
+      const toolFn = toolsByName.get(tc.name);
+      let result;
+      if (toolFn) {
+        try {
+          const out = await toolFn.invoke(
+            tc.args
+          );
+          result = typeof out === "string" ? out : JSON.stringify(out);
+        } catch (err) {
+          result = `Error: ${err instanceof Error ? err.message : String(err)}`;
+        }
+      } else {
+        result = `Unknown tool: ${tc.name}`;
+      }
+      messages.push(new ToolMessage({ content: result, tool_call_id: id }));
+    }
+  }
+  return "Agent reached max turns without a final answer.";
+}
+async function main() {
+  const { configPath, query } = parseArgv();
+  const llmSection = loadLlmSection(configPath);
+  const model = createChatModelFromLlmConfig({ llmSection });
+  console.log("Query:", query);
+  console.log("---");
+  const answer = await runAgent(model, query);
+  console.log("Answer:", answer);
+  console.log("---");
+  console.log("Done.");
+}
+main().catch((err) => {
+  console.error(err);
+  process.exit(1);
+});
+//# sourceMappingURL=cli.js.map

package/dist/cli.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/cli.ts"],"sourcesContent":["#!/usr/bin/env node\n/**\n * CLI for @easynet/agent-llm: run the LangChain agent with a query.\n * Usage: agent-llm \"your question\"\n *   or: agent-llm --config ./config/llm.yaml \"hi\"\n *   or: npx @easynet/agent-llm \"hi\"\n */\n\nimport { join } from \"node:path\";\nimport { createChatModelFromLlmConfig, loadLlmConfig } from \"./index.js\";\nimport { tool } from \"@langchain/core/tools\";\nimport { z } from \"zod\";\nimport { HumanMessage, AIMessage, ToolMessage } from \"@langchain/core/messages\";\nimport type { BaseMessage } from \"@langchain/core/messages\";\n\nconst defaultLlmSection = {\n  provider: \"openai\" as const,\n  model: process.env.OPENAI_MODEL ?? \"gpt-4o-mini\",\n  temperature: 0,\n  apiKey: process.env.OPENAI_API_KEY,\n  ...(process.env.OPENAI_BASE_URL && { baseURL: process.env.OPENAI_BASE_URL }),\n};\n\nfunction parseArgv(): { configPath: string | undefined; query: string } {\n  const args = process.argv.slice(2);\n  let configPath: string | undefined;\n  const rest: string[] = [];\n  for (let i = 0; i < args.length; i++) {\n    if (args[i] === \"--config\" && args[i + 1]) {\n      configPath = args[i + 1];\n      i++;\n    } else {\n      rest.push(args[i]);\n    }\n  }\n  const query = rest.join(\" \").trim() || \"hi\";\n  return { configPath, query };\n}\n\nfunction loadLlmSection(configPath: string | undefined): unknown {\n  const path = configPath ?? join(process.cwd(), \"config\", \"llm.yaml\");\n  const llm = loadLlmConfig(path);\n  return llm ?? defaultLlmSection;\n}\n\nconst getWeather = tool(\n  (input: { location: string }) => {\n    const loc = input.location.toLowerCase();\n    if ([\"sf\", \"san francisco\"].includes(loc)) {\n      return \"It's 60°F and foggy in San Francisco.\";\n    }\n    if ([\"ny\", \"new york\"].includes(loc)) {\n      return \"It's 72°F and partly cloudy in New York.\";\n    }\n    return `Weather for ${input.location}: 70°F and sunny.`;\n  },\n  {\n    name: \"get_weather\",\n    description: \"Get the current weather for a location.\",\n    schema: z.object({\n      location: z.string().describe(\"City or place name (e.g. SF, New York)\"),\n    }),\n  }\n);\n\nconst addNumbers = tool(\n  (input: { a: number; b: number }) => String(input.a + input.b),\n  {\n    name: \"add_numbers\",\n    description: \"Add two numbers.\",\n    schema: z.object({\n      a: z.number().describe(\"First number\"),\n      b: z.number().describe(\"Second number\"),\n    }),\n  }\n);\n\nconst tools = [getWeather, addNumbers];\nconst toolsByName = new Map(tools.map((t) => [t.name, t]));\n\nconst MAX_TURNS = 10;\n\nasync function runAgent(\n  model: ReturnType<typeof createChatModelFromLlmConfig>,\n  query: string\n): Promise<string> {\n  const withTools = model.bindTools?.(tools, { tool_choice: \"auto\" });\n  if (!withTools) throw new Error(\"Model does not support bindTools\");\n  const messages: BaseMessage[] = [new HumanMessage(query)];\n\n  for (let turn = 0; turn < MAX_TURNS; turn++) {\n    const response = await withTools.invoke(messages);\n    const aiMessage = response as AIMessage;\n\n    if (!aiMessage.tool_calls?.length) {\n      const content =\n        typeof aiMessage.content === \"string\"\n          ? aiMessage.content\n          : Array.isArray(aiMessage.content)\n            ? (aiMessage.content as { type?: string; text?: string }[])\n                .map((c) => (\"text\" in c ? c.text : \"\"))\n                .join(\"\")\n            : String(aiMessage.content ?? \"\");\n      return content;\n    }\n\n    messages.push(aiMessage);\n\n    for (const tc of aiMessage.tool_calls) {\n      const id = tc.id ?? `call_${turn}_${tc.name}`;\n      const toolFn = toolsByName.get(tc.name as \"get_weather\" | \"add_numbers\");\n      let result: string;\n      if (toolFn) {\n        try {\n          const out = await (toolFn as { invoke: (args: Record<string, unknown>) => Promise<unknown> }).invoke(\n            tc.args as Record<string, unknown>\n          );\n          result = typeof out === \"string\" ? out : JSON.stringify(out);\n        } catch (err) {\n          result = `Error: ${err instanceof Error ? err.message : String(err)}`;\n        }\n      } else {\n        result = `Unknown tool: ${tc.name}`;\n      }\n      messages.push(new ToolMessage({ content: result, tool_call_id: id }));\n    }\n  }\n\n  return \"Agent reached max turns without a final answer.\";\n}\n\nasync function main() {\n  const { configPath, query } = parseArgv();\n  const llmSection = loadLlmSection(configPath);\n  const model = createChatModelFromLlmConfig({ llmSection });\n\n  console.log(\"Query:\", query);\n  console.log(\"---\");\n\n  const answer = await runAgent(model, query);\n  console.log(\"Answer:\", answer);\n  console.log(\"---\");\n  console.log(\"Done.\");\n}\n\nmain().catch((err) => {\n  console.error(err);\n  process.exit(1);\n});\n"],"mappings":";;;;;;;AAQA,SAAS,YAAY;AAErB,SAAS,YAAY;AACrB,SAAS,SAAS;AAClB,SAAS,cAAyB,mBAAmB;AAGrD,IAAM,oBAAoB;AAAA,EACxB,UAAU;AAAA,EACV,OAAO,QAAQ,IAAI,gBAAgB;AAAA,EACnC,aAAa;AAAA,EACb,QAAQ,QAAQ,IAAI;AAAA,EACpB,GAAI,QAAQ,IAAI,mBAAmB,EAAE,SAAS,QAAQ,IAAI,gBAAgB;AAC5E;AAEA,SAAS,YAA+D;AACtE,QAAM,OAAO,QAAQ,KAAK,MAAM,CAAC;AACjC,MAAI;AACJ,QAAM,OAAiB,CAAC;AACxB,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,QAAI,KAAK,CAAC,MAAM,cAAc,KAAK,IAAI,CAAC,GAAG;AACzC,mBAAa,KAAK,IAAI,CAAC;AACvB;AAAA,IACF,OAAO;AACL,WAAK,KAAK,KAAK,CAAC,CAAC;AAAA,IACnB;AAAA,EACF;AACA,QAAM,QAAQ,KAAK,KAAK,GAAG,EAAE,KAAK,KAAK;AACvC,SAAO,EAAE,YAAY,MAAM;AAC7B;AAEA,SAAS,eAAe,YAAyC;AAC/D,QAAM,OAAO,cAAc,KAAK,QAAQ,IAAI,GAAG,UAAU,UAAU;AACnE,QAAM,MAAM,cAAc,IAAI;AAC9B,SAAO,OAAO;AAChB;AAEA,IAAM,aAAa;AAAA,EACjB,CAAC,UAAgC;AAC/B,UAAM,MAAM,MAAM,SAAS,YAAY;AACvC,QAAI,CAAC,MAAM,eAAe,EAAE,SAAS,GAAG,GAAG;AACzC,aAAO;AAAA,IACT;AACA,QAAI,CAAC,MAAM,UAAU,EAAE,SAAS,GAAG,GAAG;AACpC,aAAO;AAAA,IACT;AACA,WAAO,eAAe,MAAM,QAAQ;AAAA,EACtC;AAAA,EACA;AAAA,IACE,MAAM;AAAA,IACN,aAAa;AAAA,IACb,QAAQ,EAAE,OAAO;AAAA,MACf,UAAU,EAAE,OAAO,EAAE,SAAS,wCAAwC;AAAA,IACxE,CAAC;AAAA,EACH;AACF;AAEA,IAAM,aAAa;AAAA,EACjB,CAAC,UAAoC,OAAO,MAAM,IAAI,MAAM,CAAC;AAAA,EAC7D;AAAA,IACE,MAAM;AAAA,IACN,aAAa;AAAA,IACb,QAAQ,EAAE,OAAO;AAAA,MACf,GAAG,EAAE,OAAO,EAAE,SAAS,cAAc;AAAA,MACrC,GAAG,EAAE,OAAO,EAAE,SAAS,eAAe;AAAA,IACxC,CAAC;AAAA,EACH;AACF;AAEA,IAAM,QAAQ,CAAC,YAAY,UAAU;AACrC,IAAM,cAAc,IAAI,IAAI,MAAM,IAAI,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC;AAEzD,IAAM,YAAY;AAElB,eAAe,SACb,OACA,OACiB;AACjB,QAAM,YAAY,MAAM,YAAY,OAAO,EAAE,aAAa,OAAO,CAAC;AAClE,MAAI,CAAC,UAAW,OAAM,IAAI,MAAM,kCAAkC;AAClE,QAAM,WAA0B,CAAC,IAAI,aAAa,KAAK,CAAC;AAExD,WAAS,OAAO,GAAG,OAAO,WAAW,QAAQ;AAC3C,UAAM,WAAW,MAAM,UAAU,OAAO,QAAQ;AAChD,UAAM,YAAY;AAElB,QAAI,CAAC,UAAU,YAAY,QAAQ;AACjC,YAAM,UACJ,OAAO,UAAU,YAAY,WACzB,UAAU,UACV,MAAM,QAAQ,UAAU,OAAO,IAC5B,UAAU,QACR,IAAI,CAAC,MAAO,UAAU,IAAI,EAAE,OAAO,EAAG,EACtC,KAAK,EAAE,IACV,OAAO,UAAU,WAAW,EAAE;AACtC,aAAO;AAAA,IACT;AAEA,aAAS,KAAK,SAAS;AAEvB,eAAW,MAAM,UAAU,YAAY;AACrC,YAAM,KAAK,GAAG,MAAM,QAAQ,IAAI,IAAI,GAAG,IAAI;AAC3C,YAAM,SAAS,YAAY,IAAI,GAAG,IAAqC;AACvE,UAAI;AACJ,UAAI,QAAQ;AACV,YAAI;AACF,gBAAM,MAAM,MAAO,OAA2E;AAAA,YAC5F,GAAG;AAAA,UACL;AACA,mBAAS,OAAO,QAAQ,WAAW,MAAM,KAAK,UAAU,GAAG;AAAA,QAC7D,SAAS,KAAK;AACZ,mBAAS,UAAU,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC;AAAA,QACrE;AAAA,MACF,OAAO;AACL,iBAAS,iBAAiB,GAAG,IAAI;AAAA,MACnC;AACA,eAAS,KAAK,IAAI,YAAY,EAAE,SAAS,QAAQ,cAAc,GAAG,CAAC,CAAC;AAAA,IACtE;AAAA,EACF;AAEA,SAAO;AACT;AAEA,eAAe,OAAO;AACpB,QAAM,EAAE,YAAY,MAAM,IAAI,UAAU;AACxC,QAAM,aAAa,eAAe,UAAU;AAC5C,QAAM,QAAQ,6BAA6B,EAAE,WAAW,CAAC;AAEzD,UAAQ,IAAI,UAAU,KAAK;AAC3B,UAAQ,IAAI,KAAK;AAEjB,QAAM,SAAS,MAAM,SAAS,OAAO,KAAK;AAC1C,UAAQ,IAAI,WAAW,MAAM;AAC7B,UAAQ,IAAI,KAAK;AACjB,UAAQ,IAAI,OAAO;AACrB;AAEA,KAAK,EAAE,MAAM,CAAC,QAAQ;AACpB,UAAQ,MAAM,GAAG;AACjB,UAAQ,KAAK,CAAC;AAChB,CAAC;","names":[]}

package/dist/config.d.ts

@@ -1,10 +1,10 @@
 /**
  * Parse agent.yaml llm section into normalized LLMConfig[] and default id.
- * Supports: flat (每个模型一个 name 为 key)、instances[]、或单对象。
+ * Supports: flat (each model keyed by name), instances[], or single object.
  */
 import type { LLMConfig } from "./types.js";
 /**
- * 解析 llm section：扁平（每个模型一个 name 为 key）或 default+instances 或单对象。
+ * Parse llm section: flat (each model keyed by name), default+instances, or single object.
  */
 export declare function parseLlmSection(section: unknown): {
     defaultId: string;

package/dist/createAgentLlM.d.ts

@@ -0,0 +1,10 @@
+export interface CreateAgentLlMOptions {
+    /** Path to YAML config file. If omitted, uses llm.yaml in cwd or config/llm.yaml in cwd/parent. */
+    configPath?: string;
+}
+/**
+ * Create a LangChain-formatted LLM from config.
+ * Pass configPath to use a specific YAML file; otherwise uses llm.yaml (cwd) or config/llm.yaml (cwd/parent).
+ */
+export declare function createAgentLlM(options?: CreateAgentLlMOptions): import("@langchain/core/language_models/chat_models").BaseChatModel<import("@langchain/core/language_models/chat_models").BaseChatModelCallOptions, import("@langchain/core/messages").AIMessageChunk>;
+//# sourceMappingURL=createAgentLlM.d.ts.map

package/dist/createAgentLlM.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"createAgentLlM.d.ts","sourceRoot":"","sources":["../src/createAgentLlM.ts"],"names":[],"mappings":"AASA,MAAM,WAAW,qBAAqB;IACpC,mGAAmG;IACnG,UAAU,CAAC,EAAE,MAAM,CAAC;CACrB;AAWD;;;GAGG;AACH,wBAAgB,cAAc,CAAC,OAAO,GAAE,qBAA0B,0MAOjE"}

package/dist/factory.d.ts

@@ -3,11 +3,11 @@
  */
 import type { AgentConfigLlmSection, ILLMRegistry } from "./types.js";
 export interface CreateLLMRegistryOptions {
-    /** 已解析的 llm section（来自 loadAgentConfig 的 config.llm） */
+    /** Parsed llm section (e.g. from loadAgentConfig's config.llm) */
     llmSection: AgentConfigLlmSection | null | undefined;
 }
 /**
- * 从 agent config 的 llm 段创建 LLM 注册表；支持多 provider、多模型、每 LLM 有 id 与 type。
+ * Create LLM registry from agent config llm section; supports multiple providers/models, each LLM has id and type.
  */
 export declare function createLLMRegistry(options: CreateLLMRegistryOptions): ILLMRegistry;
 //# sourceMappingURL=factory.d.ts.map

package/dist/factory.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"factory.d.ts","sourceRoot":"","sources":["../src/factory.ts"],"names":[],"mappings":"AAAA;;GAEG;AAIH,OAAO,KAAK,EAAE,qBAAqB,EAAc,YAAY,EAAE,MAAM,YAAY,CAAC;AAElF,MAAM,WAAW,wBAAwB;IACvC,wDAAwD;IACxD,UAAU,EAAE,qBAAqB,GAAG,IAAI,GAAG,SAAS,CAAC;CACtD;AAED;;GAEG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,wBAAwB,GAAG,YAAY,CAyBjF"}
+{"version":3,"file":"factory.d.ts","sourceRoot":"","sources":["../src/factory.ts"],"names":[],"mappings":"AAAA;;GAEG;AAIH,OAAO,KAAK,EAAE,qBAAqB,EAAc,YAAY,EAAE,MAAM,YAAY,CAAC;AAElF,MAAM,WAAW,wBAAwB;IACvC,kEAAkE;IAClE,UAAU,EAAE,qBAAqB,GAAG,IAAI,GAAG,SAAS,CAAC;CACtD;AAED;;GAEG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,wBAAwB,GAAG,YAAY,CAyBjF"}

package/dist/index.d.ts

@@ -4,6 +4,8 @@
  * Extensions register via registerProvider/registerChatModelProvider; call loadLLMExtensions() before using them.
  */
 export { parseLlmSection } from "./config.js";
+export { loadLlmConfig, parseLlmYaml, substituteEnv } from "./loadLlmConfig.js";
+export type { LoadLlmConfigOptions } from "./loadLlmConfig.js";
 export { createLLMRegistry } from "./factory.js";
 export type { CreateLLMRegistryOptions } from "./factory.js";
 export { createClient, registerProvider } from "./providers/index.js";
@@ -11,6 +13,8 @@ export { registerChatModelProvider, getChatModelFactory } from "./chatModelRegis
 export { createOpenAIClient, createOpenAIChatClient, createOpenAIImageClient } from "./providers/openai.js";
 export { createChatModelFromLlmConfig } from "./llmAdapter.js";
 export type { CreateChatModelFromLlmConfigOptions } from "./llmAdapter.js";
+export { createAgentLlM } from "./createAgentLlM.js";
+export type { CreateAgentLlMOptions } from "./createAgentLlM.js";
 export { loadLLMExtensions, resolveLLMExtensionPackages } from "./loadLLMExtensions.js";
 export type { LLMType, LLMConfig, AgentConfigLlmSection, ChatMessage, ChatResult, ImageResult, ToolDefinition, ChatWithToolsMessage, ChatWithToolsResult, ILLMClient, ILLMRegistry, } from "./types.js";
 //# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AAC9C,OAAO,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAC;AACjD,YAAY,EAAE,wBAAwB,EAAE,MAAM,cAAc,CAAC;AAC7D,OAAO,EAAE,YAAY,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAC;AACtE,OAAO,EAAE,yBAAyB,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAC;AACxF,OAAO,EAAE,kBAAkB,EAAE,sBAAsB,EAAE,uBAAuB,EAAE,MAAM,uBAAuB,CAAC;AAC5G,OAAO,EAAE,4BAA4B,EAAE,MAAM,iBAAiB,CAAC;AAC/D,YAAY,EAAE,mCAAmC,EAAE,MAAM,iBAAiB,CAAC;AAC3E,OAAO,EAAE,iBAAiB,EAAE,2BAA2B,EAAE,MAAM,wBAAwB,CAAC;AAExF,YAAY,EACV,OAAO,EACP,SAAS,EACT,qBAAqB,EACrB,WAAW,EACX,UAAU,EACV,WAAW,EACX,cAAc,EACd,oBAAoB,EACpB,mBAAmB,EACnB,UAAU,EACV,YAAY,GACb,MAAM,YAAY,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AAC9C,OAAO,EAAE,aAAa,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AAChF,YAAY,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC/D,OAAO,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAC;AACjD,YAAY,EAAE,wBAAwB,EAAE,MAAM,cAAc,CAAC;AAC7D,OAAO,EAAE,YAAY,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAC;AACtE,OAAO,EAAE,yBAAyB,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAC;AACxF,OAAO,EAAE,kBAAkB,EAAE,sBAAsB,EAAE,uBAAuB,EAAE,MAAM,uBAAuB,CAAC;AAC5G,OAAO,EAAE,4BAA4B,EAAE,MAAM,iBAAiB,CAAC;AAC/D,YAAY,EAAE,mCAAmC,EAAE,MAAM,iBAAiB,CAAC;AAC3E,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AACrD,YAAY,EAAE,qBAAqB,EAAE,MAAM,qBAAqB,CAAC;AACjE,OAAO,EAAE,iBAAiB,EAAE,2BAA2B,EAAE,MAAM,wBAAwB,CAAC;AAExF,YAAY,EACV,OAAO,EACP,SAAS,EACT,qBAAqB,EACrB,WAAW,EACX,UAAU,EACV,WAAW,EACX,cAAc,EACd,oBAAoB,EACpB,mBAAmB,EACnB,UAAU,EACV,YAAY,GACb,MAAM,YAAY,CAAC"}

package/dist/index.js

@@ -1,344 +1,23 @@
-// src/config.ts
-var DEFAULT_LLM_ID = "default";
-var RESERVED_KEYS = /* @__PURE__ */ new Set([
-  "default",
-  "instances",
-  "catalog",
-  "provider",
-  "model",
-  "temperature",
-  "apiKey",
-  "baseURL",
-  "base_url",
-  "type",
-  "id"
-]);
-function parseLlmSection(section) {
-  if (section == null || typeof section !== "object") {
-    return { defaultId: DEFAULT_LLM_ID, configs: [] };
-  }
-  if (Array.isArray(section)) {
-    const configs = section.filter((i) => i != null && typeof i === "object").map((item, i) => normalizeLlmConfig({ ...item, id: item.id ?? item.name ?? String(i) })).filter((c) => c != null);
-    const defaultId = configs.length > 0 ? configs[0].id : DEFAULT_LLM_ID;
-    return { defaultId, configs };
-  }
-  const s = section;
-  const flatEntries = Object.entries(s).filter(
-    ([k, v]) => !RESERVED_KEYS.has(k) && v != null && typeof v === "object" && !Array.isArray(v)
-  );
-  if (flatEntries.length > 0) {
-    const configs = [];
-    for (const [id, entry] of flatEntries) {
-      const c = entryToLlmConfig(id, entry);
-      if (c) configs.push(c);
-    }
-    const defaultId = typeof s.default === "string" && s.default && flatEntries.some(([k]) => k === s.default) ? s.default : configs.length > 0 ? configs[0].id : DEFAULT_LLM_ID;
-    return { defaultId, configs };
-  }
-  if (Array.isArray(s.instances)) {
-    const configs = s.instances.filter((i) => i != null && typeof i === "object").map((i) => normalizeLlmConfig(i)).filter((c) => c != null);
-    const defaultId = typeof s.default === "string" && s.default ? s.default : configs.length > 0 ? configs[0].id : DEFAULT_LLM_ID;
-    return { defaultId, configs };
-  }
-  if (typeof s.provider === "string" || typeof s.model === "string" || typeof s.name === "string") {
-    const one = singleObjectToLlmConfig(s);
-    return { defaultId: one.id, configs: [one] };
-  }
-  return { defaultId: DEFAULT_LLM_ID, configs: [] };
-}
-var EXTENSION_OPTION_KEYS = ["featureKey", "tenant", "authToken", "verifySSL", "bypassAuth", "host", "resolveHost", "timeoutMs", "options"];
-function entryToLlmConfig(id, entry) {
-  const opts = entry.options;
-  const baseURL = typeof entry.base_url === "string" ? entry.base_url : typeof entry.baseURL === "string" ? entry.baseURL : void 0;
-  const model = typeof entry.name === "string" ? entry.name : typeof entry.model === "string" ? entry.model : void 0;
-  const provider = typeof entry.provider === "string" && entry.provider ? entry.provider : "openai";
-  const config = {
-    id,
-    type: "chat",
-    provider,
-    model,
-    temperature: typeof opts?.temperature === "number" ? opts.temperature : typeof entry.temperature === "number" ? entry.temperature : void 0,
-    apiKey: typeof opts?.apiKey === "string" ? opts.apiKey : typeof entry.apiKey === "string" ? entry.apiKey : void 0,
-    baseURL
-  };
-  if (typeof entry.type === "string" && entry.type === "image") config.type = "image";
-  if (opts && typeof opts === "object") config.options = opts;
-  for (const k of EXTENSION_OPTION_KEYS) {
-    if (entry[k] !== void 0) config[k] = entry[k];
-    else if (opts && opts[k] !== void 0) config[k] = opts[k];
-  }
-  return config;
-}
-function singleObjectToLlmConfig(s) {
-  const one = {
-    id: DEFAULT_LLM_ID,
-    type: "chat",
-    provider: typeof s.provider === "string" ? s.provider : "openai",
-    model: typeof s.model === "string" ? s.model : typeof s.name === "string" ? s.name : void 0,
-    temperature: typeof s.temperature === "number" ? s.temperature : void 0,
-    apiKey: typeof s.apiKey === "string" ? s.apiKey : void 0,
-    baseURL: typeof s.baseURL === "string" ? s.baseURL : typeof s.base_url === "string" ? s.base_url : void 0
-  };
-  Object.keys(s).forEach((k) => {
-    if (!["id", "type", "provider", "model", "name", "temperature", "apiKey", "baseURL", "base_url", "default", "instances"].includes(k)) {
-      one[k] = s[k];
-    }
-  });
-  return one;
-}
-function normalizeLlmConfig(o) {
-  const id = typeof o.id === "string" && o.id ? o.id : DEFAULT_LLM_ID;
-  const type = o.type === "image" ? "image" : "chat";
-  const provider = typeof o.provider === "string" && o.provider ? o.provider : "openai";
-  const config = {
-    id,
-    type,
-    provider,
-    model: typeof o.model === "string" ? o.model : typeof o.name === "string" ? o.name : void 0,
-    temperature: typeof o.temperature === "number" ? o.temperature : void 0,
-    apiKey: typeof o.apiKey === "string" ? o.apiKey : void 0,
-    baseURL: typeof o.baseURL === "string" ? o.baseURL : typeof o.base_url === "string" ? o.base_url : void 0
-  };
-  Object.keys(o).forEach((k) => {
-    if (!["id", "type", "provider", "model", "name", "temperature", "apiKey", "baseURL", "base_url"].includes(k)) {
-      config[k] = o[k];
-    }
-  });
-  return config;
-}
-
-// src/providers/openai.ts
-import OpenAI from "openai";
-function getApiKey(config) {
-  const key = config.apiKey ?? process.env.OPENAI_API_KEY ?? "";
-  if (!key) throw new Error("OpenAI-compatible apiKey required (config.apiKey or OPENAI_API_KEY)");
-  return key;
-}
-function createOpenAIClientOptions(config) {
-  const opts = { apiKey: getApiKey(config) };
-  if (typeof config.baseURL === "string" && config.baseURL) opts.baseURL = config.baseURL;
-  return opts;
-}
-function serializeMessage(m) {
-  if (m.role === "tool")
-    return { role: "tool", content: m.content, tool_call_id: m.tool_call_id };
-  if (m.role === "assistant" && "tool_calls" in m && m.tool_calls?.length) {
-    return {
-      role: "assistant",
-      content: m.content ?? null,
-      tool_calls: m.tool_calls.map((tc) => ({
-        id: tc.id,
-        type: "function",
-        function: { name: tc.function.name, arguments: tc.function.arguments }
-      }))
-    };
-  }
-  return { role: m.role, content: m.content };
-}
-function createOpenAIChatClient(config) {
-  const client = new OpenAI(createOpenAIClientOptions(config));
-  const model = config.model ?? process.env.OPENAI_MODEL ?? "gpt-4o-mini";
-  const temperature = config.temperature ?? 0;
-  return {
-    id: config.id,
-    type: "chat",
-    async chat(messages) {
-      const resp = await client.chat.completions.create({
-        model,
-        temperature,
-        messages: messages.map((m) => ({ role: m.role, content: m.content }))
-      });
-      const content = resp.choices[0]?.message?.content ?? "";
-      const usage = resp.usage ? { promptTokens: resp.usage.prompt_tokens, completionTokens: resp.usage.completion_tokens } : void 0;
-      return { content, usage };
-    },
-    async chatWithTools(messages, tools, _options) {
-      const resp = await client.chat.completions.create({
-        model,
-        temperature,
-        messages: messages.map(serializeMessage),
-        tools: tools.map((t) => ({
-          type: "function",
-          function: {
-            name: t.function.name,
-            description: t.function.description,
-            parameters: t.function.parameters ?? void 0
-          }
-        }))
-      });
-      const msg = resp.choices[0]?.message;
-      const usage = resp.usage ? { promptTokens: resp.usage.prompt_tokens, completionTokens: resp.usage.completion_tokens } : void 0;
-      return {
-        message: {
-          role: "assistant",
-          content: msg?.content ?? null,
-          tool_calls: msg?.tool_calls?.map((tc) => ({
-            id: tc.id,
-            type: "function",
-            function: {
-              name: tc.function?.name ?? "",
-              arguments: tc.function?.arguments ?? ""
-            }
-          }))
-        },
-        usage
-      };
-    }
-  };
-}
-function createOpenAIImageClient(config) {
-  const client = new OpenAI(createOpenAIClientOptions(config));
-  const model = config.model ?? "dall-e-3";
-  return {
-    id: config.id,
-    type: "image",
-    async chat() {
-      throw new Error("OpenAI image model does not support chat; use generateImage()");
-    },
-    async generateImage(options) {
-      const resp = await client.images.generate({
-        model,
-        prompt: options.prompt,
-        size: options.size ?? "1024x1024",
-        n: options.n ?? 1,
-        response_format: "url"
-      });
-      const url = resp.data?.[0]?.url ?? void 0;
-      return { url };
-    }
-  };
-}
-function createOpenAIClient(config) {
-  if (config.type === "image") return createOpenAIImageClient(config);
-  return createOpenAIChatClient(config);
-}
-
-// src/providers/index.ts
-var OPENAI_COMPATIBLE = "openai-compatible";
-function createOpenAICompat(config) {
-  return createOpenAIClient(config);
-}
-var PROVIDERS = {
-  openai: createOpenAICompat,
-  [OPENAI_COMPATIBLE]: createOpenAICompat
-};
-function createClient(config) {
-  const p = (config.provider ?? "").toLowerCase();
-  const fn = PROVIDERS[p];
-  if (!fn) {
-    const supported = [.../* @__PURE__ */ new Set([...Object.keys(PROVIDERS), "extension providers"])].sort().join(", ");
-    throw new Error(
-      `Unsupported LLM provider: ${config.provider}. Supported: ${supported}.`
-    );
-  }
-  return fn(config);
-}
-function registerProvider(name, factory) {
-  PROVIDERS[name.toLowerCase()] = factory;
-}
-
-// src/factory.ts
-function createLLMRegistry(options) {
-  const { defaultId, configs } = parseLlmSection(options.llmSection);
-  const map = /* @__PURE__ */ new Map();
-  for (const config of configs) {
-    try {
-      const client = createClient(config);
-      map.set(config.id, client);
-    } catch (err) {
-      console.warn(`[agent-llm] Skip LLM "${config.id}": ${err instanceof Error ? err.message : String(err)}`);
-    }
-  }
-  return {
-    get(id) {
-      return map.get(id);
-    },
-    defaultId() {
-      if (map.has(defaultId)) return defaultId;
-      return map.size > 0 ? [...map.keys()][0] : void 0;
-    },
-    ids() {
-      return [...map.keys()];
-    }
-  };
-}
-
-// src/chatModelRegistry.ts
-var CHAT_MODEL_FACTORIES = /* @__PURE__ */ new Map();
-function registerChatModelProvider(providerName, factory) {
-  CHAT_MODEL_FACTORIES.set(providerName.toLowerCase(), factory);
-}
-function getChatModelFactory(providerName) {
-  return CHAT_MODEL_FACTORIES.get(providerName.toLowerCase());
-}
-
-// src/llmAdapter.ts
-import { ChatOpenAI } from "@langchain/openai";
-var DEFAULT_MODEL = "gpt-4o-mini";
-function createChatModelFromLlmConfig(options) {
-  const { llmSection, modelEnv, apiKeyEnv } = options;
-  const { defaultId, configs } = parseLlmSection(llmSection ?? null);
-  const defaultConfig = configs.find((c) => c.id === defaultId) ?? configs[0];
-  if (!defaultConfig) {
-    const model2 = modelEnv ?? process.env.OPENAI_MODEL ?? DEFAULT_MODEL;
-    const apiKey2 = apiKeyEnv ?? process.env.OPENAI_API_KEY;
-    return new ChatOpenAI({
-      model: model2,
-      temperature: 0,
-      ...apiKey2 ? { apiKey: apiKey2 } : {}
-    });
-  }
-  const provider = defaultConfig.provider ?? "openai";
-  const chatModelFactory = getChatModelFactory(provider);
-  if (chatModelFactory) {
-    const config = {
-      ...defaultConfig,
-      model: modelEnv ?? defaultConfig.model ?? (provider === "cis" ? process.env.CIS_MODEL ?? "gcp/gemini-2.5-pro" : defaultConfig.model),
-      temperature: typeof defaultConfig.temperature === "number" ? defaultConfig.temperature : 0
-    };
-    return chatModelFactory(config);
-  }
-  const model = modelEnv ?? defaultConfig?.model ?? process.env.OPENAI_MODEL ?? DEFAULT_MODEL;
-  const apiKey = apiKeyEnv ?? defaultConfig?.apiKey ?? process.env.OPENAI_API_KEY;
-  const temperature = typeof defaultConfig?.temperature === "number" ? defaultConfig.temperature : 0;
-  const baseURL = defaultConfig?.baseURL;
-  const constructorOptions = {
-    model,
-    temperature,
-    ...apiKey ? { apiKey } : {},
-    ...baseURL ? { configuration: { baseURL } } : {}
-  };
-  return new ChatOpenAI(constructorOptions);
-}
-
-// src/loadLLMExtensions.ts
-var loadedPackages = /* @__PURE__ */ new Set();
-var DEFAULT_EXTENSIONS = ["wallee-llm"];
-function resolveLLMExtensionPackages(types) {
-  const typeList = types == null ? [] : Array.isArray(types) ? types : [types];
-  const packages = typeList.filter(
-    (t) => typeof t === "string" && t.length > 0
-  );
-  return packages.length > 0 ? packages : DEFAULT_EXTENSIONS;
-}
-async function loadLLMExtensions(extensionPackages) {
-  const packages = extensionPackages ?? DEFAULT_EXTENSIONS;
-  for (const pkg of packages) {
-    if (loadedPackages.has(pkg)) continue;
-    loadedPackages.add(pkg);
-    try {
-      const m = await import(
-        /* @vite-ignore */
-        pkg
-      );
-      if (typeof m.registerLLMExtension === "function") {
-        m.registerLLMExtension();
-      }
-    } catch {
-    }
-  }
-}
+import {
+  createAgentLlM,
+  createChatModelFromLlmConfig,
+  createClient,
+  createLLMRegistry,
+  createOpenAIChatClient,
+  createOpenAIClient,
+  createOpenAIImageClient,
+  getChatModelFactory,
+  loadLLMExtensions,
+  loadLlmConfig,
+  parseLlmSection,
+  parseLlmYaml,
+  registerChatModelProvider,
+  registerProvider,
+  resolveLLMExtensionPackages,
+  substituteEnv
+} from "./chunk-3Z3KXKNU.js";
 export {
+  createAgentLlM,
   createChatModelFromLlmConfig,
   createClient,
   createLLMRegistry,
@@ -347,9 +26,12 @@ export {
   createOpenAIImageClient,
   getChatModelFactory,
   loadLLMExtensions,
+  loadLlmConfig,
   parseLlmSection,
+  parseLlmYaml,
   registerChatModelProvider,
   registerProvider,
-  resolveLLMExtensionPackages
+  resolveLLMExtensionPackages,
+  substituteEnv
 };
 //# sourceMappingURL=index.js.map