fix: DeepSeek /anthropic endpoint support + model config overhaul

- Route DeepSeek /anthropic endpoint through AnthropicProvider (Anthropic SDK
  format) since it does not accept OpenAI Chat Completions format. The
  engine-factory now auto-detects anthropic in the base URL and uses the
  correct provider.

- Fix --model CLI flag by making KODE_MODEL the highest priority in
  resolveModelConfig() and the KodeGatewayClient constructor.

- Make /model command persistent by updating settings.json on disk via
  config.set RPC handler (writes env.ANTHROPIC_MODEL/BASE_URL/AUTH_TOKEN
  and default_model).

- Add protected buildStreamUrl() in OpenAICompatProvider for subclasses
  to override URL path construction. DeepSeekProvider overrides to append
  /v1/chat/completions to the base URL.

- Change DeepSeek default base URL to https://api.deepseek.com/anthropic.

- Remove dead code: shared/src/config/loader.ts (201 lines, never imported
  by CLI — settings.json is the sole config source).

- Update configuration priority comment to reflect actual flow:
  CLI args > Env vars > ~/.kode/settings.json > Defaults.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: jimgqyu

README.md

@@ -2,11 +2,11 @@
 
 <p align="center">
   <b>An open-source, enterprise-grade CLI agent coding tool</b><br/>
-  <em>Multi-provider LLM support, Agent Teams, and a powerful hook system</em>
+  <em>A Claude Code alternative with multi-provider support, Agent Teams, and a powerful hook system</em>
 </p>
 
 <p align="center">
-  <img src="https://img.shields.io/badge/license-Apache%202.0-blue.svg" alt="License: Apache 2.0" />
+  <img src="https://img.shields.io/badge/license-MIT-blue.svg" alt="License: MIT" />
   <img src="https://img.shields.io/badge/TypeScript-5.7-blue" alt="TypeScript" />
   <img src="https://img.shields.io/badge/node-%3E%3D18.0.0-brightgreen" alt="Node.js >= 18.0.0" />
   <img src="https://img.shields.io/badge/tests-~500%20passed-success" alt="Tests" />
@@ -38,9 +38,9 @@ cat > ~/.kode/settings.json << 'EOF'
   "theme": "dark",
   "model_list": [
     {
-      "name": "deepseek-v4-pro",
-      "model": "deepseek-v4-pro",
-      "base_url": "https://api.deepseek.com/v1",
+      "name": "deepseek/deepseek-v4-pro",
+      "model": "deepseek-chat",
+      "base_url": "https://api.deepseek.com/anthropic",
       "auth_token_env": "DEEPSEEK_API_KEY",
       "provider": "deepseek"
     },
@@ -52,7 +52,7 @@ cat > ~/.kode/settings.json << 'EOF'
       "provider": "anthropic"
     }
   ],
-  "default_model": "deepseek-v4-pro"
+  "default_model": "deepseek/deepseek-v4-pro"
 }
 EOF
 
@@ -67,7 +67,7 @@ kode
 For a guided setup, run the installer:
 
 ```bash
-curl -fsSL https://raw.githubusercontent.com/AgenticMatrix/Kode-Agent/main/install.sh | bash
+curl -fsSL https://raw.githubusercontent.com/kode-agent/kode-agent/main/install.sh | bash
 ```
 
 ---
@@ -235,7 +235,7 @@ skills:
 
 | Feature | Kode Agent | Claude Code | Hermes-Agent |
 |---------|-----------|-------------|--------------|
-| **License** | Apache 2.0 | Proprietary | MIT |
+| **License** | MIT | Proprietary | MIT |
 | **Multi-Provider** | ✅ Anthropic + OpenAI + DeepSeek + Auto | ❌ Anthropic only | ✅ Multi-provider |
 | **Agent Teams** | ✅ Coordinator/Worker + SubagentBus | ❌ Single agent | ❌ Single agent |
 | **Hook System** | ✅ 27 events, Shell + JS | ⚠️ Limited hooks | ❌ No hooks |
@@ -283,18 +283,19 @@ export KODE_PROVIDER=openai
 export OPENAI_API_KEY="sk-..."
 
 # Use DeepSeek with Anthropic-compatible endpoint
-export KODE_PROVIDER=deepseek
-export DEEPSEEK_API_KEY="sk-..."
-export DEEPSEEK_BASE_URL="https://api.deepseek.com/v1"
+export KODE_PROVIDER=anthropic
+export ANTHROPIC_API_KEY="sk-..."
+export ANTHROPIC_BASE_URL="https://api.deepseek.com/anthropic"
+export ANTHROPIC_MODEL="deepseek-chat"
 ```
 
 ### Configuration File (~/.kode/config.yaml)
 
 For persistent configuration, use `~/.kode/config.yaml`:
 
 ```yaml
-provider: deepseek
-model: deepseek-v4-pro
+provider: anthropic
+model: claude-sonnet-4-20250514
 
 providers:
   anthropic:
@@ -306,7 +307,7 @@ providers:
     maxRetries: 3
     timeoutMs: 120000
   deepseek:
-    baseUrl: https://api.deepseek.com/v1
+    baseUrl: https://api.deepseek.com/anthropic
     maxRetries: 3
     timeoutMs: 120000
 
@@ -326,21 +327,21 @@ You can also use `~/.kode/settings.json` (JSON format) for environment variables
   "theme": "dark",
   "model_list": [
     {
-      "name": "deepseek-v4-pro",
-      "model": "deepseek-v4-pro",
-      "base_url": "https://api.deepseek.com/v1",
-      "auth_token_env": "DEEPSEEK_API_KEY",
-      "provider": "deepseek"
-    },
-    {
-      "name": "anthropic/claude-sonnet-4-6",
+      "name": "claude-sonnet-4-6",
       "model": "claude-sonnet-4-6",
       "base_url": "https://api.anthropic.com/v1",
       "auth_token_env": "ANTHROPIC_API_KEY",
       "provider": "anthropic"
+    },
+    {
+      "name": "deepseek/deepseek-v4-pro",
+      "model": "deepseek-chat",
+      "base_url": "https://api.deepseek.com/anthropic",
+      "auth_token_env": "DEEPSEEK_API_KEY",
+      "provider": "deepseek"
     }
   ],
-  "default_model": "deepseek-v4-pro"
+  "default_model": "claude-sonnet-4-6"
 }
 ```
 
@@ -363,9 +364,9 @@ Place hook definitions in `~/.kode/hooks/*.json`:
 
 Supported events: `SessionStart`, `UserPromptSubmit`, `PreMessage`, `PostMessage`, `PreToolUse`, `PostToolUse`, `PostToolBatch`, `Stop`, `StopFailure`, `PreCompact`, `PostCompact`, `PermissionRequest`, `PermissionDenied`, `Notification`, `SubagentStart`, `SubagentStop`, and more.
 
-### Project Configuration (KODE.md)
+### Project Configuration (CLAUDE.md)
 
-Place a `KODE.md` file in your project root for project-specific instructions. Kode Agent automatically loads this as context at the start of every session (also supports `CLAUDE.md` and `CODEBUDDY.md` for compatibility):
+Place a `CLAUDE.md` file in your project root for project-specific instructions. Kode Agent automatically loads this as context at the start of every session:
 
 ```markdown
 # Project Overview
@@ -384,7 +385,7 @@ This is a Next.js e-commerce application with Prisma ORM and PostgreSQL.
 - API routes follow RESTful conventions
 ```
 
-Use `/init` in an interactive session to have Kode Agent auto-generate a `KODE.md` for your project.
+Use `/init` in an interactive session to have Kode Agent auto-generate a `CLAUDE.md` for your project.
 
 ---
 
@@ -442,7 +443,7 @@ pnpm ci
 
 ## License
 
-[Apache 2.0](LICENSE) © Kode Agent Contributors
+MIT © Kode Agent Contributors
 
 ---
 

docs/configuration.md

@@ -12,8 +12,8 @@ automatically on first run if it doesn't exist.
 
 ```yaml
 # ── Provider ──────────────────────────────────────────────────────────────
-provider: deepseek             # deepseek | anthropic | openai
-model: deepseek-v4-pro
+provider: anthropic          # anthropic | deepseek | openai
+model: claude-sonnet-4-20250514
 
 # ── Provider-specific settings ────────────────────────────────────────────
 providers:
@@ -22,7 +22,7 @@ providers:
     maxRetries: 3
     timeoutMs: 120000
   deepseek:
-    baseUrl: https://api.deepseek.com/v1
+    baseUrl: https://api.deepseek.com/anthropic
     maxRetries: 3
     timeoutMs: 120000
   openai:
@@ -304,5 +304,5 @@ compactThreshold: 0.7      # Start compacting at 126K tokens
 ### Multiple Projects
 
 Configuration is global (`~/.kode/config.yaml`), but project-specific
-instructions go in `KODE.md` at the project root. The agent reads this
+instructions go in `CLAUDE.md` at the project root. The agent reads this
 file at the start of every session.

packages/cli/src/gateway/engine-factory.ts

@@ -83,7 +83,7 @@ export interface EngineFactoryOptions {
   apiKey?: string;
   /** Base URL override for the provider */
   baseUrl?: string;
-  /** Model identifier (default: from env or 'deepseek-v4-pro') */
+  /** Model identifier (default: from env or 'claude-sonnet-4-6') */
   model?: string;
   /** Provider name: "anthropic" | "openai" | "deepseek" | "auto" (default: "anthropic") */
   providerName?: string;
@@ -240,8 +240,8 @@ export function createQueryEngine(
   const cwd = opts.cwd ?? process.cwd();
   const apiKey = opts.apiKey ?? process.env.ANTHROPIC_API_KEY ?? '';
   const baseUrl = opts.baseUrl ?? process.env.ANTHROPIC_BASE_URL;
-  const model = opts.model ?? process.env.KODE_MODEL ?? process.env.ANTHROPIC_MODEL ?? 'deepseek-v4-pro';
-  const providerName = opts.providerName ?? process.env.KODE_PROVIDER ?? 'anthropic';
+  const model = (opts.model && opts.model !== 'claude-sonnet-4-6') ? opts.model : process.env.KODE_MODEL ?? process.env.ANTHROPIC_MODEL ?? opts.model ?? 'claude-sonnet-4-6';
+  const providerName = opts.providerName ?? (model.toLowerCase().includes('deepseek') ? 'deepseek' : process.env.KODE_PROVIDER ?? 'anthropic');
 
   // ── Determine agent role ─────────────────────────────────────────
   const coordinatorMode = opts.coordinatorMode ?? false;
@@ -371,11 +371,21 @@ export function createQueryEngine(
  * @returns A Provider instance
  */
 function createProvider(name: string, config: ProviderConfig): Provider {
+  const isAnthropicEndpoint = config.baseUrl?.includes('/anthropic');
+
   switch (name) {
     case 'openai':
       return new OpenAICompatProvider(config, 'openai');
-    case 'deepseek':
+    case 'deepseek': {
+      // DeepSeek's /anthropic endpoint uses Anthropic Messages API format, not
+      // OpenAI Chat Completions.  When the base URL targets the /anthropic
+      // endpoint we must use AnthropicProvider so the SDK sends the correct
+      // request shape.
+      if (isAnthropicEndpoint) {
+        return new AnthropicProvider(config);
+      }
       return new DeepSeekProvider(config);
+    }
     case 'auto': {
       // Auto mode: create a router with all providers that have API keys configured
       const router = new ProviderRouter();
@@ -386,10 +396,14 @@ function createProvider(name: string, config: ProviderConfig): Provider {
       if (openaiKey) {
         router.register('openai', new OpenAICompatProvider({ ...config, apiKey: openaiKey }, 'openai'), ['gpt-4o', 'gpt-4o-mini']);
       }
-      // Register DeepSeek if key available
+      // Register DeepSeek if key available — use the right provider per endpoint
       const deepseekKey = process.env.DEEPSEEK_API_KEY;
       if (deepseekKey) {
-        router.register('deepseek', new DeepSeekProvider({ ...config, apiKey: deepseekKey }), ['deepseek-chat', 'deepseek-reasoner']);
+        const deepseekCfg = { ...config, apiKey: deepseekKey };
+        const deepseekProvider = isAnthropicEndpoint
+          ? new AnthropicProvider(deepseekCfg)
+          : new DeepSeekProvider(deepseekCfg);
+        router.register('deepseek', deepseekProvider, ['deepseek-chat', 'deepseek-reasoner']);
       }
       // Return a proxy provider that delegates to the router
       return createRouterProxy(router, config.apiKey ? 'anthropic' : undefined);
@@ -407,7 +421,7 @@ function createProvider(name: string, config: ProviderConfig): Provider {
 function createRouterProxy(router: ProviderRouter, _defaultProvider?: string): Provider {
   const routerRef = { current: router };
   const providerRef: { current: Provider | null } = { current: null };
-  const modelRef: { current: string } = { current: 'deepseek-v4-pro' };
+  const modelRef: { current: string } = { current: 'claude-sonnet-4-6' };
 
   return {
     async stream(modelConfig, system, messages, tools, onEvent) {
@@ -455,5 +469,5 @@ export function hasApiKey(): boolean {
  * Get the configured model name.
  */
 export function getConfiguredModel(): string {
-  return process.env.KODE_MODEL ?? process.env.ANTHROPIC_MODEL ?? 'deepseek-v4-pro';
+  return process.env.KODE_MODEL ?? process.env.ANTHROPIC_MODEL ?? 'claude-sonnet-4-6';
 }