feat(core): 为 LLM API 请求补齐 User-Agent以解决403问题(支持环境变量覆盖)

.env.example

@@ -13,6 +13,10 @@ INKOS_LLM_API_KEY=sk-your-key-here
 # Model name
 INKOS_LLM_MODEL=gpt-4o
 
+# Optional: set a custom User-Agent header for LLM HTTP requests
+# Some gateways/WAFs may block requests without a browser-like UA.
+INKOS_HTTP_USER_AGENT=Mozilla/5.0 (compatible; InkOS)
+
 # Notifications (optional)
 INKOS_TELEGRAM_BOT_TOKEN=
 INKOS_TELEGRAM_CHAT_ID=

packages/cli/src/commands/init.ts

@@ -71,10 +71,10 @@ export const initCommand = new Command("init")
             "# Global config at ~/.inkos/.env will be used by default.",
             "# Uncomment below to override for this project only:",
             "# INKOS_LLM_PROVIDER=openai",
-            "# INKOS_LLM_BASE_URL=",
-            "# INKOS_LLM_API_KEY=",
-            "# INKOS_LLM_MODEL=",
-            "",
+            "# INKOS_LLM_BASE_URL=https://api.openai.com/v1",
+            "# INKOS_LLM_API_KEY=your-api-key-here",
+            "# INKOS_LLM_MODEL=gpt-4o",
+            "# INKOS_HTTP_USER_AGENT=Mozilla/5.0 (compatible; InkOS)",
             "# Web search (optional):",
             "# TAVILY_API_KEY=tvly-xxxxx",
           ].join("\n"),
@@ -97,6 +97,7 @@ export const initCommand = new Command("init")
             "# INKOS_LLM_MAX_TOKENS=8192",
             "# INKOS_LLM_THINKING_BUDGET=0          # Anthropic extended thinking budget",
             "# INKOS_LLM_API_FORMAT=chat             # chat (default) or responses (OpenAI Responses API)",
+            "# INKOS_HTTP_USER_AGENT=Mozilla/5.0 (compatible; InkOS)",
             "",
             "# Web search (optional, for auditor era-research):",
             "# TAVILY_API_KEY=tvly-xxxxx              # Free at tavily.com (1000 searches/month)",

packages/core/src/llm/provider.ts

@@ -2,6 +2,10 @@ import OpenAI from "openai";
 import Anthropic from "@anthropic-ai/sdk";
 import type { LLMConfig } from "../models/project.js";
 
+function resolveUserAgent(): string {
+  const fromEnv = process.env.INKOS_HTTP_USER_AGENT?.trim();
+  if (fromEnv) return fromEnv;
+  return "Mozilla/5.0 (compatible; InkOS)";
 // === Streaming Monitor Types ===
 
 export interface StreamProgress {
@@ -118,6 +122,7 @@ export function createLLMClient(config: LLMConfig): LLMClient {
     extra: config.extra ?? {},
   };
 
+  const userAgent = resolveUserAgent();
   const apiFormat = config.apiFormat ?? "chat";
   const stream = config.stream ?? true;
 
@@ -128,7 +133,11 @@ export function createLLMClient(config: LLMConfig): LLMClient {
       provider: "anthropic",
       apiFormat,
       stream,
-      _anthropic: new Anthropic({ apiKey: config.apiKey, baseURL }),
+      _anthropic: new Anthropic({
+        apiKey: config.apiKey,
+        baseURL,
+        defaultHeaders: { "User-Agent": userAgent },
+      }),
       defaults,
     };
   }
@@ -137,7 +146,11 @@ export function createLLMClient(config: LLMConfig): LLMClient {
     provider: "openai",
     apiFormat,
     stream,
-    _openai: new OpenAI({ apiKey: config.apiKey, baseURL: config.baseUrl }),
+    _openai: new OpenAI({
+      apiKey: config.apiKey,
+      baseURL: config.baseUrl,
+      defaultHeaders: { "User-Agent": userAgent },
+    }),
     defaults,
   };
 }