ai-claude-start

ai-claude-start

English | 简体中文

Multi-profile Claude Code launcher with secure credential management.

Features

- Simplified Configuration: Only 3-4 fields needed - name, base URL, optional model, and token
- Multi-Profile Support: Manage multiple API configurations (Anthropic, Moonshot, BigModel, or custom)
- Secure Credential Storage: Uses keytar for OS-level secure storage, with automatic fallback to local file storage
- Environment Sanitization: Cleans all ANTHROPIC_* variables before injection to prevent conflicts
- Unified Authentication: Always uses ANTHROPIC_AUTH_TOKEN for credentials
- Interactive Setup: Guided wizard with 3 built-in presets
- Interactive Selection: Shows a selection menu when multiple profiles are configured
- Model Configuration: Optional model name configuration, automatically passed to Claude CLI
- Testing Support: Built-in --cmd flag and CLAUDE_CMD environment variable for testing without Claude CLI

Installation

Or use directly with npx:

Quick Start

1. First-time Setup

Run the setup wizard to create your first profile:

You'll be guided through selecting a preset (Anthropic, Moonshot, IMDS) or creating a custom profile.

2. Launch Claude

With interactive selection (multiple profiles):

Auto-select (single profile):

Specify profile directly:

Quick profile syntax (shorthand):

Pass arguments to Claude:

Commands

setup
Interactive wizard to create or update a profile.

list
Display all configured profiles with their settings and credential status.

default
Set the default profile to use when no profile is specified.

delete
Delete a profile and its stored credentials.

doctor
Check system health: keytar availability, profiles, credentials, and Claude CLI presence.

Profile Configuration

A profile consists of 3-4 fields:

Built-in…