@funnycode/myclaude 0.1.54 → 0.1.56

package/README.md

@@ -197,6 +197,69 @@ Use `MYCLAUDE_*` or `CLAUDE_CODE_*` (the latter takes priority).
 
 ---
 
+## 🔧 Custom Models
+
+myclaude supports third-party models (any provider compatible with the Anthropic API format) and local models.
+
+### Basic Configuration
+
+| Env Variable | What it does | Example |
+|--------------|-------------|---------|
+| `ANTHROPIC_BASE_URL` | Custom API endpoint | `https://your-api.example.com` |
+| `MYCLAUDE_MODEL` / `CLAUDE_CODE_MODEL` | Specify model name | `claude-sonnet-4-20250514`, `gemini-2.5-pro` |
+
+> Supports both `MYCLAUDE_*` and `CLAUDE_CODE_*` prefixes — the latter takes priority, fully compatible with Claude Code environment variable naming.
+
+### Common Scenarios
+
+**Using a proxy / relay API:**
+```bash
+export ANTHROPIC_BASE_URL=https://your-proxy.com/v1
+export ANTHROPIC_API_KEY=your-proxy-key
+export MYCLAUDE_MODEL=claude-sonnet-4-20250514
+npx @funnycode/myclaude
+```
+
+**Using Ollama local models (via compatibility layer):**
+```bash
+# First, start Ollama and pull a model
+# ollama pull llama3.2
+
+# Use tools like llm-cluster or litellm to provide an Anthropic-compatible endpoint
+export ANTHROPIC_BASE_URL=http://localhost:8000/v1
+export ANTHROPIC_API_KEY=sk-local
+export MYCLAUDE_MODEL=llama3.2
+npx @funnycode/myclaude
+```
+
+**Using Google Vertex AI:**
+```bash
+export MYCLAUDE_USE_VERTEX=true
+export CLOUD_ML_REGION=us-central1
+gcloud auth application-default login
+npx @funnycode/myclaude
+```
+
+**Using AWS Bedrock:**
+```bash
+export MYCLAUDE_USE_BEDROCK=true
+export AWS_REGION=us-east-1
+# AWS credentials via env vars or ~/.aws/credentials
+npx @funnycode/myclaude
+```
+
+**Using Microsoft Foundry:**
+```bash
+export MYCLAUDE_USE_FOUNDRY=true
+export AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com
+export AZURE_OPENAI_API_KEY=your-azure-key
+npx @funnycode/myclaude
+```
+
+> 💘 At runtime, you can also use the `/model` command to quickly switch between supported models.
+
+---
+
 ## 🛠 Dev
 
 ```bash

package/README.zh-CN.md

@@ -197,6 +197,69 @@ codegraph init
 
 ---
 
+## 🔧 自定义模型
+
+myclaude 支持接入第三方模型（兼容 Anthropic API 格式的任意提供商）或本地模型。
+
+### 基本配置
+
+| 环境变量 | 说明 | 示例 |
+|---------|------|------|
+| `ANTHROPIC_BASE_URL` | 自定义 API 端点地址 | `https://your-api.example.com` |
+| `MYCLAUDE_MODEL` / `CLAUDE_CODE_MODEL` | 指定模型名称 | `claude-sonnet-4-20250514`、`gemini-2.5-pro` |
+
+> 支持 `MYCLAUDE_*` 或 `CLAUDE_CODE_*` 两种前缀，后者优先级更高，完全兼容 Claude Code 的环境变量命名。
+
+### 常见场景示例
+
+**使用代理/中转 API：**
+```bash
+export ANTHROPIC_BASE_URL=https://your-proxy.com/v1
+export ANTHROPIC_API_KEY=your-proxy-key
+export MYCLAUDE_MODEL=claude-sonnet-4-20250514
+npx @funnycode/myclaude
+```
+
+**使用 Ollama 本地模型（通过兼容层）：**
+```bash
+# 先启动 Ollama 并拉取模型
+# ollama pull llama3.2
+
+# 使用 llm-cluster 或 litellm 等工具提供 Anthropic 兼容接口
+export ANTHROPIC_BASE_URL=http://localhost:8000/v1
+export ANTHROPIC_API_KEY=sk-local
+export MYCLAUDE_MODEL=llama3.2
+npx @funnycode/myclaude
+```
+
+**使用 Google Vertex AI：**
+```bash
+export MYCLAUDE_USE_VERTEX=true
+export CLOUD_ML_REGION=us-central1
+gcloud auth application-default login
+npx @funnycode/myclaude
+```
+
+**使用 AWS Bedrock：**
+```bash
+export MYCLAUDE_USE_BEDROCK=true
+export AWS_REGION=us-east-1
+# AWS 凭证通过环境变量或 ~/.aws/credentials 配置
+npx @funnycode/myclaude
+```
+
+**使用 Microsoft Foundry：**
+```bash
+export MYCLAUDE_USE_FOUNDRY=true
+export AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com
+export AZURE_OPENAI_API_KEY=your-azure-key
+npx @funnycode/myclaude
+```
+
+> 💡 运行时也可通过 `/model` 命令在支持的模型之间快速切换。
+
+---
+
 ## 🛠 开发
 
 ```bash

package/dist/myclaude.js

@@ -4,8 +4,8 @@
 // MACRO - build-time constants (injected by build.ts)
 // MACRO injected by build script
 globalThis.MACRO = {
-  VERSION: "0.1.54",
-  BUILD_TIME: "2026-06-29T13:05:24.446Z",
+  VERSION: "0.1.56",
+  BUILD_TIME: "2026-07-05T14:01:38.459Z",
   PACKAGE_URL: "@funnycode/myclaude",
   NATIVE_PACKAGE_URL: "@funnycode/myclaude",
   VERSION_CHANGELOG: '',
@@ -347778,7 +347778,8 @@ async function findSuitableShell() {
 }
 async function getShellConfigImpl() {
   const binShell = await findSuitableShell();
-  const provider = await createBashShellProvider(binShell);
+  const isWindows2 = getPlatform() === "windows";
+  const provider = await createBashShellProvider(binShell, { skipSnapshot: isWindows2 });
   return { provider };
 }
 async function exec3(command, abortSignal, shellType, options) {

package/dist/myclaude.mjs

@@ -4,8 +4,8 @@
 // MACRO - build-time constants (injected by build.ts)
 // MACRO injected by build script
 globalThis.MACRO = {
-  VERSION: "0.1.54",
-  BUILD_TIME: "2026-06-29T13:05:24.446Z",
+  VERSION: "0.1.56",
+  BUILD_TIME: "2026-07-05T14:01:38.459Z",
   PACKAGE_URL: "@funnycode/myclaude",
   NATIVE_PACKAGE_URL: "@funnycode/myclaude",
   VERSION_CHANGELOG: '',
@@ -347778,7 +347778,8 @@ async function findSuitableShell() {
 }
 async function getShellConfigImpl() {
   const binShell = await findSuitableShell();
-  const provider = await createBashShellProvider(binShell);
+  const isWindows2 = getPlatform() === "windows";
+  const provider = await createBashShellProvider(binShell, { skipSnapshot: isWindows2 });
   return { provider };
 }
 async function exec3(command, abortSignal, shellType, options) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@funnycode/myclaude",
-  "version": "0.1.54",
+  "version": "0.1.56",
   "private": false,
   "description": "An open-source AI coding assistant in your terminal - powered by Claude",
   "license": "MIT",