local-openai2anthropic 0.2.4__tar.gz → 0.3.6__tar.gz

local_openai2anthropic-0.3.6/.claude/CLAUDE.md

@@ -0,0 +1,163 @@
+# Claude 开发指南
+
+## Configuration File
+
+### Config File Location
+
+- **Linux/macOS**: `~/.oa2a/config.toml`
+- **Windows**: `%USERPROFILE%\.oa2a\config.toml`
+
+### First Run
+
+A default config file will be created automatically on first startup. Edit this file to add your API keys.
+
+### Configuration Example
+
+```toml
+# OA2A Configuration File
+# Place this file at ~/.oa2a/config.toml
+
+# OpenAI API Configuration
+openai_api_key = "your-openai-api-key"
+openai_base_url = "https://api.openai.com/v1"
+openai_org_id = ""
+openai_project_id = ""
+
+# Server Configuration
+host = "0.0.0.0"
+port = 8080
+request_timeout = 300.0
+
+# API Key for authenticating requests to this server (optional)
+api_key = ""
+
+# CORS settings
+cors_origins = ["*"]
+cors_credentials = true
+cors_methods = ["*"]
+cors_headers = ["*"]
+
+# Logging
+log_level = "DEBUG"
+log_dir = ""  # Empty uses platform-specific default
+
+# Tavily Web Search Configuration
+tavily_api_key = ""
+tavily_timeout = 30.0
+tavily_max_results = 5
+websearch_max_uses = 5
+```
+
+## 版本号更新清单
+
+当修改项目版本号时，需要同步更新以下所有位置：
+
+### 1. pyproject.toml
+- **位置**: 第3行
+- **格式**: `version = "x.y.z"`
+- **示例**: `version = "0.3.5"`
+
+### 2. src/local_openai2anthropic/__init__.py
+- **位置**: 第6行
+- **格式**: `__version__ = "x.y.z"`
+- **示例**: `__version__ = "0.3.5"`
+
+### 3. src/local_openai2anthropic/main.py (FastAPI 应用版本)
+- **位置**: 第104行
+- **格式**: `version="x.y.z"`
+- **示例**: `version="0.3.5"`
+
+### 4. src/local_openai2anthropic/main.py (命令行 --version)
+- **位置**: 第254行
+- **格式**: `version="%(prog)s x.y.z"`
+- **示例**: `version="%(prog)s 0.3.5"`
+
+### 5. tests/test_main.py (测试断言)
+- **位置**: 第26行
+- **格式**: `assert app.version == "x.y.z"`
+- **示例**: `assert app.version == "0.3.5"`
+
+### 6. Git Tag
+- **格式**: `vx.y.z`
+- **示例**: `v0.3.5`
+- **命令**:
+  ```bash
+  git tag v0.3.5
+  git push origin v0.3.5
+  ```
+
+## 版本号格式
+
+使用语义化版本控制 (Semantic Versioning)：
+- **MAJOR**: 不兼容的 API 修改
+- **MINOR**: 向下兼容的功能新增
+- **PATCH**: 向下兼容的问题修复
+
+## 发布流程
+
+1. 更新上述所有文件中的版本号
+2. 运行测试确保通过: `pytest`
+3. 提交更改: `git commit -m "chore(release): bump version to x.y.z"`
+4. 创建标签: `git tag vx.y.z`
+5. 推送代码和标签: `git push && git push origin vx.y.z`
+6. GitHub Actions 将自动发布到 PyPI
+
+## 代码提交规范
+
+### 测试覆盖率要求
+
+在提交任何新代码之前，必须满足以下测试要求：
+
+#### 1. 新代码单元测试
+- **必须**为所有未提交的新代码编写单元测试
+- 新代码的测试覆盖率**必须 > 90%**
+- 测试文件命名规范: `test_<module_name>.py`
+- 测试函数命名规范: `test_<function_name>_<scenario>`
+
+#### 2. 总体覆盖率检查
+- 运行 `/everything-claude-code:test-coverage` 检查总体覆盖率
+- **总体覆盖率必须 ≥ 80%**
+- 如果总体覆盖率低于 80%，需要补充测试或优化现有代码
+
+#### 3. 代码审查
+
+在提交代码前，必须进行代码审查：
+
+```bash
+# 运行代码审查（检查安全漏洞、代码质量、最佳实践）
+/code-review:code-review
+
+# 根据审查结果修复问题
+# - CRITICAL/HIGH 级别问题必须修复
+# - MEDIUM/LOW 级别问题根据情况处理
+```
+
+代码审查将检查：
+- **安全问题**: 硬编码凭证、SQL注入、XSS、路径遍历等
+- **代码质量**: 函数长度、嵌套深度、错误处理、console.log等
+- **最佳实践**: 不可变模式、emoji使用、测试覆盖、可访问性等
+
+#### 4. 提交流程
+```bash
+# 1. 编写新代码
+# 2. 编写对应的单元测试
+# 3. 运行测试并检查覆盖率
+pytest --cov=src/local_openai2anthropic --cov-report=term-missing
+
+# 4. 运行 Claude Code 覆盖率检查
+/everything-claude-code:test-coverage
+
+# 5. 运行代码审查
+/code-review:code-review
+
+# 6. 确认新代码覆盖率 > 90% 且总体覆盖率 ≥ 80%，且无 CRITICAL/HIGH 问题
+# 7. 提交代码
+git add .
+git commit -m "feat: your commit message"
+```
+
+#### 4. 测试质量标准
+- 测试用例应覆盖正常路径、边界条件和异常情况
+- 使用 `pytest` 作为测试框架
+- 异步代码使用 `pytest-asyncio`
+- 适当的测试夹具 (fixtures) 和参数化测试

local_openai2anthropic-0.3.6/.reports/dead-code-analysis.md

@@ -0,0 +1,282 @@
+# Dead Code Analysis Report
+
+**Project:** local-openai2anthropic
+**Analysis Date:** 2026-02-03
+**Tools Used:** vulture, pylint, ruff
+
+---
+
+## Executive Summary
+
+This report identifies potentially unused code in the codebase. The analysis found **47 issues** across the source code, categorized by severity level. No code has been deleted - this is for review purposes only.
+
+### Summary by Severity
+
+| Severity | Count | Description |
+|----------|-------|-------------|
+| SAFE | 23 | Unused imports, variables, and f-strings - safe to remove |
+| CAUTION | 15 | Unused functions/methods - verify before removal |
+| DANGER | 9 | Model fields and properties - may be used externally |
+
+---
+
+## Detailed Findings
+
+### SEVERITY: SAFE (High Confidence for Removal)
+
+These items are safe to remove with minimal risk.
+
+#### 1. Unused Imports
+
+| File | Line | Item | Tool | Notes |
+|------|------|------|------|-------|
+| `src/local_openai2anthropic/converter.py` | 20 | `ChatCompletionChunk` from `openai.types.chat` | vulture (90%), ruff, pylint | Imported but never referenced |
+| `src/local_openai2anthropic/converter.py` | 26 | `ServerToolRegistry` from `local_openai2anthropic.server_tools` | vulture (90%), ruff, pylint | Imported but never referenced |
+| `src/local_openai2anthropic/protocol.py` | 9 | `Field` from `pydantic` | vulture (90%), ruff, pylint | Imported but never used in file |
+| `tests/test_daemon_runner.py` | 8 | `datetime` | vulture (90%) | Unused import in test file |
+| `tests/test_router_comprehensive.py` | 7 | `Mock` from `unittest.mock` | vulture (90%) | Unused import |
+| `tests/test_router_comprehensive.py` | 9 | `Request` from `fastapi` | vulture (90%) | Unused import |
+| `tests/test_router_comprehensive.py` | 10 | `StreamingResponse` from `fastapi.responses` | vulture (90%) | Unused import |
+| `tests/test_router_comprehensive.py` | 12 | `_handle_with_server_tools` from `local_openai2anthropic.router` | vulture (90%) | Unused import |
+| `tests/test_router_streaming.py` | 7 | `Mock` from `unittest.mock` | vulture (90%) | Unused import |
+| `tests/test_router_streaming.py` | 9 | `StreamingResponse` from `fastapi.responses` | vulture (90%) | Unused import |
+| `tests/test_router_streaming.py` | 11 | `_handle_with_server_tools` from `local_openai2anthropic.router` | vulture (90%) | Unused import |
+
+#### 2. Unused Local Variables
+
+| File | Line | Variable | Tool | Notes |
+|------|------|----------|------|-------|
+| `src/local_openai2anthropic/converter.py` | 297 | `is_error` | ruff (F841) | Assigned but never used in tool_result handling |
+| `src/local_openai2anthropic/daemon.py` | 274 | `e` (exception) | ruff (F841) | Exception variable not used in except block |
+| `src/local_openai2anthropic/main.py` | 240 | `status_parser` | vulture (60%), ruff | Assigned but status command parser not used |
+| `src/local_openai2anthropic/router.py` | 483 | `openai_stop_reason` | vulture (60%), ruff | Variable assigned but never used |
+| `src/local_openai2anthropic/router.py` | 1182 | `model` | ruff (F841) | Extracted from request but never used |
+
+#### 3. F-strings Without Placeholders
+
+| File | Line | Content | Tool | Notes |
+|------|------|---------|------|-------|
+| `src/local_openai2anthropic/daemon.py` | 146 | `"Use 'oa2a logs' to view output"` | ruff (F541) | No placeholders in f-string |
+| `src/local_openai2anthropic/daemon.py` | 152 | `"Another process may be listening on this port"` | ruff (F541) | No placeholders in f-string |
+| `src/local_openai2anthropic/daemon.py` | 263 | `"Server did not stop gracefully, use -f to force kill"` | ruff (F541) | No placeholders in f-string |
+| `src/local_openai2anthropic/daemon_runner.py` | 84 | `"Configuration loaded:"` | ruff (F541) | No placeholders in f-string |
+
+#### 4. Unused Arguments
+
+| File | Line | Function/Method | Argument | Tool | Notes |
+|------|------|-----------------|----------|------|-------|
+| `src/local_openai2anthropic/daemon_runner.py` | 45 | signal handler | `frame` | vulture (100%), pylint | Signal handler frame argument not used |
+| `src/local_openai2anthropic/main.py` | 95 | `http_exception_handler` | `request` | pylint | FastAPI exception handler receives request but doesn't use it |
+| `src/local_openai2anthropic/main.py` | 116 | `general_exception_handler` | `request` | pylint | FastAPI exception handler receives request but doesn't use it |
+| `src/local_openai2anthropic/router.py` | 860 | (inline function) | `handler` | pylint | Exception handler argument not used |
+| `src/local_openai2anthropic/router.py` | 1138 | `count_tokens` | `settings` | pylint | Settings dependency injected but not used |
+
+---
+
+### SEVERITY: CAUTION (Verify Before Removal)
+
+These items appear unused but may be called dynamically or through external APIs.
+
+#### 1. Unused Functions
+
+| File | Line | Function | Tool | Risk Assessment |
+|------|------|----------|------|-----------------|
+| `src/local_openai2anthropic/converter.py` | 348 | `_build_usage_with_cache` | vulture (60%) | **LOW RISK** - Private function, appears truly unused. Could be removed or used for future cache token support. |
+| `src/local_openai2anthropic/main.py` | 55 | `auth_middleware` | vulture (60%) | **MEDIUM RISK** - Middleware that may be registered dynamically. Verify if used in app setup. |
+| `src/local_openai2anthropic/main.py` | 94 | `http_exception_handler` | vulture (60%) | **LOW RISK** - Exception handler, may be registered via decorator. Check FastAPI app setup. |
+| `src/local_openai2anthropic/main.py` | 115 | `general_exception_handler` | vulture (60%) | **LOW RISK** - Exception handler, may be registered via decorator. Check FastAPI app setup. |
+| `src/local_openai2anthropic/router.py` | 898 | `create_message` | vulture (60%) | **MEDIUM RISK** - API endpoint handler. May be registered as route. Verify router setup. |
+| `src/local_openai2anthropic/router.py` | 1106 | `list_models` | vulture (60%) | **MEDIUM RISK** - API endpoint handler. May be registered as route. Verify router setup. |
+| `src/local_openai2anthropic/router.py` | 1135 | `count_tokens` | vulture (60%) | **MEDIUM RISK** - API endpoint handler. May be registered as route. Verify router setup. |
+| `src/local_openai2anthropic/router.py` | 1241 | `health_check` | vulture (60%) | **MEDIUM RISK** - API endpoint handler. May be registered as route. Verify router setup. |
+
+#### 2. Unused Methods
+
+| File | Line | Class | Method | Tool | Risk Assessment |
+|------|------|-------|--------|------|-----------------|
+| `src/local_openai2anthropic/server_tools/base.py` | 167 | `ServerToolRegistry` | `all_tools` | vulture (60%) | **LOW RISK** - Method that returns all registered tools. May be useful for debugging/admin. |
+
+#### 3. Unused Test Functions/Fixtures
+
+| File | Line | Function | Tool | Notes |
+|------|------|----------|------|-------|
+| `tests/test_router_comprehensive.py` | 31 | `async_iter` | vulture (60%) | Helper function, may be used in tests |
+| `tests/test_router_streaming.py` | 25 | `async_iter` | vulture (60%) | Helper function, may be used in tests |
+| `tests/test_router_streaming.py` | 614 | `async_iter` | vulture (60%) | Duplicate helper function |
+
+---
+
+### SEVERITY: DANGER (Do Not Remove Without Careful Review)
+
+These items are model fields, properties, or public API components that may be used externally even if not referenced internally.
+
+#### 1. Model Configuration and Properties
+
+| File | Line | Item | Type | Tool | Risk Assessment |
+|------|------|------|------|------|-----------------|
+| `src/local_openai2anthropic/config.py` | 15 | `model_config` | Class variable | vulture (60%) | **HIGH RISK** - Pydantic Settings configuration. Required by pydantic-settings. |
+| `src/local_openai2anthropic/config.py` | 52 | `openai_auth_headers` | Property | vulture (60%) | **HIGH RISK** - Public property that may be used by external code. Verify usage before removal. |
+
+#### 2. TypedDict Fields (May be Required for API Compatibility)
+
+| File | Line | Field | Parent Class | Tool | Risk Assessment |
+|------|------|-------|--------------|------|-----------------|
+| `src/local_openai2anthropic/openai_types.py` | 20 | `description` | `ChatCompletionToolFunction` | vulture (60%) | **HIGH RISK** - TypedDict field for OpenAI API compatibility. |
+| `src/local_openai2anthropic/openai_types.py` | 21 | `parameters` | `ChatCompletionToolFunction` | vulture (60%) | **HIGH RISK** - TypedDict field for OpenAI API compatibility. |
+| `src/local_openai2anthropic/openai_types.py` | 41 | `stop` | `CompletionCreateParams` | vulture (60%) | **HIGH RISK** - TypedDict field for OpenAI API compatibility. |
+| `src/local_openai2anthropic/openai_types.py` | 44 | `stream_options` | `CompletionCreateParams` | vulture (60%) | **HIGH RISK** - TypedDict field for OpenAI API compatibility. |
+| `src/local_openai2anthropic/openai_types.py` | 46 | `chat_template_kwargs` | `CompletionCreateParams` | vulture (60%) | **HIGH RISK** - TypedDict field for vLLM/SGLang compatibility. |
+
+#### 3. Pydantic Model Fields (Default Values)
+
+| File | Line | Field | Parent Class | Tool | Risk Assessment |
+|------|------|-------|--------------|------|-----------------|
+| `src/local_openai2anthropic/openai_types.py` | 80 | `index` | `Choice` | vulture (60%) | **MEDIUM RISK** - Model field with default value. May be used in serialization. |
+| `src/local_openai2anthropic/openai_types.py` | 95 | `index` | `ChatCompletionDeltaToolCall` | vulture (60%) | **MEDIUM RISK** - Model field with default value. |
+| `src/local_openai2anthropic/openai_types.py` | 114 | `index` | `StreamingChoice` | vulture (60%) | **MEDIUM RISK** - Model field with default value. |
+| `src/local_openai2anthropic/openai_types.py` | 135 | `created` | `ChatCompletion` | vulture (60%) | **MEDIUM RISK** - Model field. Required for OpenAI API compatibility. |
+| `src/local_openai2anthropic/openai_types.py` | 146 | `created` | `ChatCompletionChunk` | vulture (60%) | **MEDIUM RISK** - Model field. Required for OpenAI API compatibility. |
+
+#### 4. Unused Classes
+
+| File | Line | Class | Tool | Risk Assessment |
+|------|------|-------|------|-----------------|
+| `src/local_openai2anthropic/openai_types.py` | 141 | `ChatCompletionChunk` | vulture (60%) | **HIGH RISK** - Public Pydantic model. May be imported by external code. |
+
+#### 5. Protocol Model Fields (Web Search Tool Types)
+
+| File | Line | Field | Parent Class | Tool | Risk Assessment |
+|------|------|-------|--------------|------|-----------------|
+| `src/local_openai2anthropic/protocol.py` | 77 | `city` | `ApproximateLocation` | vulture (60%) | **MEDIUM RISK** - Optional field for web search location. |
+| `src/local_openai2anthropic/protocol.py` | 78 | `region` | `ApproximateLocation` | vulture (60%) | **MEDIUM RISK** - Optional field for web search location. |
+| `src/local_openai2anthropic/protocol.py` | 79 | `country` | `ApproximateLocation` | vulture (60%) | **MEDIUM RISK** - Optional field with default. |
+| `src/local_openai2anthropic/protocol.py` | 80 | `timezone` | `ApproximateLocation` | vulture (60%) | **MEDIUM RISK** - Optional field for web search location. |
+| `src/local_openai2anthropic/protocol.py` | 89 | `allowed_domains` | `WebSearchToolDefinition` | vulture (60%) | **MEDIUM RISK** - Optional field for domain filtering. |
+| `src/local_openai2anthropic/protocol.py` | 90 | `blocked_domains` | `WebSearchToolDefinition` | vulture (60%) | **MEDIUM RISK** - Optional field for domain filtering. |
+| `src/local_openai2anthropic/protocol.py` | 91 | `user_location` | `WebSearchToolDefinition` | vulture (60%) | **MEDIUM RISK** - Optional field for location context. |
+| `src/local_openai2anthropic/protocol.py` | 148 | `web_search_requests` | `ServerToolUseUsage` | vulture (60%) | **MEDIUM RISK** - Usage tracking field. |
+| `src/local_openai2anthropic/protocol.py` | 158 | `server_tool_use` | `UsageWithServerToolUse` | vulture (60%) | **MEDIUM RISK** - Usage tracking field. |
+
+---
+
+## Test File Analysis
+
+The test files contain many mock-related unused attributes. These are typically false positives because:
+
+1. **Mock return_value assignments** - Used to configure mock behavior even if not directly referenced
+2. **Mock side_effect assignments** - Configure mock behavior
+3. **Test fixtures** - May be used indirectly by pytest
+
+### Test File Recommendations
+
+| File | Issue Count | Recommendation |
+|------|-------------|----------------|
+| `test_daemon.py` | 4 | Review mock configurations - likely false positives |
+| `test_daemon_advanced.py` | 5 | Review mock configurations - likely false positives |
+| `test_daemon_runner.py` | 8 | Remove unused `datetime` import; review mocks |
+| `test_main.py` | 14 | Review mock configurations - likely false positives |
+| `test_router_comprehensive.py` | 7 | Remove unused imports; review helper functions |
+| `test_router_edge_cases.py` | 2 | Review mock configurations |
+| `test_router_streaming.py` | 6 | Remove unused imports; review helper functions |
+| `test_server_tools.py` | 13 | Review test fixtures - `cls` in `@classmethod` is required |
+| `test_tavily_client.py` | 14 | Review mock configurations - likely false positives |
+
+**Note:** The `cls` variable in `@classmethod` decorators is a false positive - it's required by Python even if not explicitly used.
+
+---
+
+## Recommendations
+
+### Immediate Actions (SAFE)
+
+1. **Remove unused imports:**
+   ```python
+   # In converter.py - Remove lines 18-22 and 26
+   from openai.types.chat import ChatCompletionChunk  # REMOVE
+   from local_openai2anthropic.server_tools import ServerToolRegistry  # REMOVE
+   ```
+
+2. **Remove unused variable assignments:**
+   ```python
+   # In converter.py line 297
+   is_error = getattr(block, "is_error", False)  # REMOVE or use
+   ```
+
+3. **Fix f-strings without placeholders:**
+   ```python
+   # Change f"..." to "..." where no placeholders exist
+   ```
+
+4. **Clean up test file imports** as listed above
+
+### Cautionary Actions (CAUTION)
+
+1. **Verify API endpoint functions** in `router.py`:
+   - Check if `create_message`, `list_models`, `count_tokens`, `health_check` are registered as routes
+   - If not registered, they can be removed
+
+2. **Review exception handlers** in `main.py`:
+   - Verify if they're registered with FastAPI's exception handlers
+   - If not registered, they can be removed
+
+### Do Not Remove (DANGER)
+
+1. **Pydantic model fields** - Required for API compatibility
+2. **TypedDict fields** - Define API contract
+3. **Config properties** - May be used externally
+4. **Web search tool type fields** - Part of feature implementation
+
+---
+
+## Verification Commands
+
+Before removing any code, verify with these commands:
+
+```bash
+# Check if a function is referenced anywhere
+grep -r "function_name" --include="*.py" .
+
+# Check for dynamic imports
+grep -r "importlib\|__import__\|import_module" --include="*.py" .
+
+# Check for string-based references (common in web frameworks)
+grep -r "\"create_message\"\|'create_message'" --include="*.py" .
+
+# Run tests after changes
+pytest tests/ -v
+
+# Check for import errors
+python -c "from local_openai2anthropic import *"
+```
+
+---
+
+## Appendix: Tool Configuration
+
+### Vulture
+```bash
+vulture src/ tests/ --min-confidence 60
+```
+
+### Pylint
+```bash
+pylint src/ --disable=all --enable=W0611,W0613,W0614,W1300,W1301,W1302,W1303,W1304,W1305,W1306,W1307,W1401,W1404,W1405,W1505
+```
+
+### Ruff
+```bash
+ruff check src/ tests/ --select F401,F541,F841
+```
+
+---
+
+## Changelog
+
+| Date | Version | Changes |
+|------|---------|---------|
+| 2026-02-03 | 1.0 | Initial analysis report |
+
+---
+
+**Report Generated By:** Claude Code Dead Code Analysis Agent
+**Disclaimer:** This report is generated by automated tools and requires human review before any code deletion. Always verify findings with grep searches and run the full test suite before removing code.

{local_openai2anthropic-0.2.4 → local_openai2anthropic-0.3.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: local-openai2anthropic
-Version: 0.2.4
+Version: 0.3.6
 Summary: A lightweight proxy server that converts Anthropic Messages API to OpenAI API
 Project-URL: Homepage, https://github.com/dongfangzan/local-openai2anthropic
 Project-URL: Repository, https://github.com/dongfangzan/local-openai2anthropic
@@ -24,6 +24,7 @@ Requires-Dist: httpx>=0.25.0
 Requires-Dist: openai>=1.30.0
 Requires-Dist: pydantic-settings>=2.0.0
 Requires-Dist: pydantic>=2.0.0
+Requires-Dist: tomli>=2.0.0; python_version < '3.11'
 Requires-Dist: uvicorn[standard]>=0.23.0
 Provides-Extra: dev
 Requires-Dist: black>=23.0.0; extra == 'dev'
@@ -55,6 +56,7 @@ This proxy translates Claude SDK calls to OpenAI API format in real-time, enabli
 - **Offline development** without cloud API costs
 - **Privacy-first AI** - data never leaves your machine
 - **Seamless model switching** between cloud and local
+- **Web Search tool** - built-in Tavily web search for local models
 
 ---
 
@@ -79,7 +81,11 @@ Other OpenAI-compatible backends may work but are not fully tested.
 pip install local-openai2anthropic
 ```
 
-### 2. Start Your Local LLM Server
+### 2. Configure Your LLM Backend (Optional)
+
+**Option A: Start a local LLM server**
+
+If you don't have an LLM server running, you can start one locally:
 
 Example with vLLM:
 ```bash
@@ -93,6 +99,16 @@ sglang launch --model-path meta-llama/Llama-2-7b-chat-hf --port 8000
 # SGLang starts at http://localhost:8000/v1
 ```
 
+**Option B: Use an existing OpenAI-compatible API**
+
+If you already have a deployed OpenAI-compatible API (local or remote), you can use it directly. Just note the base URL for the next step.
+
+Examples:
+- Local vLLM/SGLang: `http://localhost:8000/v1`
+- Remote API: `https://api.example.com/v1`
+
+> **Note:** If you're using [Ollama](https://ollama.com), it natively supports the Anthropic API format, so you don't need this proxy. Just point your Claude SDK directly to `http://localhost:11434/v1`.
+
 ### 3. Start the Proxy
 
 **Option A: Run in background (recommended)**
@@ -155,22 +171,31 @@ You can configure [Claude Code](https://github.com/anthropics/claude-code) to us
 
 ### Configuration Steps
 
-1. **Create or edit Claude Code config file** at `~/.claude/CLAUDE.md`:
-
-```markdown
-# Claude Code Configuration
-
-## API Settings
-
-- Claude API Base URL: http://localhost:8080
-- Claude API Key: dummy-key
-
-## Model Settings
-
-Use model: meta-llama/Llama-2-7b-chat-hf  # Your local model name
+1. **Edit Claude Code config file** at `~/.claude/settings.json`:
+
+```json
+{
+  "env": {
+    "ANTHROPIC_BASE_URL": "http://localhost:8080",
+    "ANTHROPIC_API_KEY": "dummy-key",
+    "ANTHROPIC_MODEL": "meta-llama/Llama-2-7b-chat-hf",
+    "ANTHROPIC_DEFAULT_SONNET_MODEL": "meta-llama/Llama-2-7b-chat-hf",
+    "ANTHROPIC_DEFAULT_OPUS_MODEL": "meta-llama/Llama-2-7b-chat-hf",
+    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "meta-llama/Llama-2-7b-chat-hf",
+    "ANTHROPIC_REASONING_MODEL": "meta-llama/Llama-2-7b-chat-hf"
+  }
+}
 ```
 
-2. **Alternatively, set environment variables** before running Claude Code:
+| Variable | Description |
+|----------|-------------|
+| `ANTHROPIC_MODEL` | General model setting |
+| `ANTHROPIC_DEFAULT_SONNET_MODEL` | Default model for Sonnet mode (Claude Code default) |
+| `ANTHROPIC_DEFAULT_OPUS_MODEL` | Default model for Opus mode |
+| `ANTHROPIC_DEFAULT_HAIKU_MODEL` | Default model for Haiku mode |
+| `ANTHROPIC_REASONING_MODEL` | Default model for reasoning tasks |
+
+2. **Or set environment variables** before running Claude Code:
 
 ```bash
 export ANTHROPIC_BASE_URL=http://localhost:8080
@@ -179,38 +204,36 @@ export ANTHROPIC_API_KEY=dummy-key
 claude
 ```
 
-3. **Or use the `--api-key` and `--base-url` flags**:
-
-```bash
-claude --api-key dummy-key --base-url http://localhost:8080
-```
-
 ### Complete Workflow Example
 
+Make sure `~/.claude/settings.json` is configured as described above.
+
 Terminal 1 - Start your local LLM:
 ```bash
 vllm serve meta-llama/Llama-2-7b-chat-hf
 ```
 
-Terminal 2 - Start the proxy:
+Terminal 2 - Start the proxy (background mode):
 ```bash
 export OA2A_OPENAI_BASE_URL=http://localhost:8000/v1
 export OA2A_OPENAI_API_KEY=dummy
 export OA2A_TAVILY_API_KEY="tvly-your-tavily-api-key"  # Optional: enable web search
 
-oa2a
+oa2a start
 ```
 
-Terminal 3 - Launch Claude Code with local LLM:
+Terminal 3 - Launch Claude Code:
 ```bash
-export ANTHROPIC_BASE_URL=http://localhost:8080
-export ANTHROPIC_API_KEY=dummy-key
-
 claude
 ```
 
 Now Claude Code will use your local LLM instead of the cloud API.
 
+To stop the proxy:
+```bash
+oa2a stop
+```
+
 ---
 
 ## Features