New VS Code Extension Lets You Switch Claude Code Providers Instantly

A developer has released a free, open-source VS Code extension that solves a common frustration for AI-assisted coding: switching between multiple LLM providers in the Claude Code plugin without manually editing configuration files. The extension, called Claude Code Profile Manager, lets users store multiple API configurations as independent profiles and swap between them with a single click.

The tool arrives at a time when developers increasingly juggle multiple AI model providers — from Anthropic's Claude to DeepSeek, Kimi, and other alternatives — to leverage different strengths across various coding tasks. Until now, switching providers in Claude Code required manually editing the settings.json file to update the base URL and API token each time.

Key Takeaways

• One-click provider switching eliminates manual editing of VS Code's settings.json
• Supports unlimited profiles, each with independent base URL, API token, and model configuration
• Can import configurations directly from bash export statements
• Displays the currently active profile in the VS Code status bar
• Fully open-source and available on GitHub
• Built specifically for developers using Claude Code with third-party API providers

The Multi-Provider Problem Every Developer Faces

The modern AI development workflow rarely relies on a single model provider. Developers frequently switch between providers like OpenAI, Anthropic, DeepSeek, and Moonshot AI (the company behind Kimi) depending on the task at hand. Some models excel at code generation, others at debugging, and still others at documentation or complex reasoning.

For users of the Claude Code extension in VS Code, this multi-provider reality creates a persistent pain point. Every time a developer wants to switch to a different API backend, they must open their settings.json file, locate the relevant configuration keys, update the base URL to point to the new provider's endpoint, swap out the API token, and potentially change the model identifier. It is a tedious, error-prone process that breaks flow and wastes time.

The creator of Claude Code Profile Manager described the motivation succinctly: they were regularly using several different model providers including DeepSeek and Kimi, and the constant manual configuration changes were 'not very convenient.' A weekend of coding later, the extension was born.

How Claude Code Profile Manager Works

The extension introduces a simple but effective concept: named profiles. Each profile stores a complete set of API configuration parameters as an independent unit. Here is what each profile contains:

• Base URL: The API endpoint for the specific provider (e.g., DeepSeek's API URL vs. Anthropic's)
• API Token: The authentication key for that particular service
• Model identifier: The specific model to use (e.g., deepseek-coder, claude-3.5-sonnet, kimi-chat)

When a user activates a profile, the extension automatically writes the corresponding values into VS Code's settings.json file — exactly as if the developer had made the changes by hand, but instantaneously and without risk of typos or formatting errors.

The status bar integration provides constant visibility into which profile is currently active. A quick glance at the bottom of the VS Code window tells developers which provider and model they are working with at any given moment. This is particularly valuable during long coding sessions where context switching between projects and providers is common.

Import From Bash: A Developer-Friendly Touch

One feature that stands out is the ability to import configurations directly from bash export statements. Many developers store their API credentials as environment variables in shell configuration files, using the standard export syntax. Rather than requiring users to manually re-enter these values, the extension can parse statements like export ANTHROPIC_API_KEY=sk-ant-... and automatically populate a new profile.

This design choice reflects a deep understanding of developer workflows. API keys and configuration values often live in .bashrc, .zshrc, or .env files. Being able to copy-paste these directly into the extension removes yet another layer of friction from the setup process.

Compared to other configuration management approaches — such as maintaining multiple settings.json files or using VS Code workspace-specific settings — the profile-based approach is significantly more streamlined. It keeps all configurations in one place and makes switching a 1-click operation rather than a multi-step file editing exercise.

Why This Matters in the Current AI Landscape

The release of this extension reflects a broader trend in the AI development ecosystem: provider diversification. The days when developers committed to a single AI provider are fading. Several factors drive this shift:

• Cost optimization: DeepSeek's API pricing can be as low as $0.14 per million input tokens, compared to $3.00 for Claude 3.5 Sonnet — a 20x difference
• Capability specialization: Different models outperform others on specific tasks like math, coding, or creative writing
• Reliability concerns: Having multiple providers configured means instant failover when one service experiences downtime
• Regional considerations: Some providers offer lower latency or better compliance in specific geographic regions
• Experimentation: Developers want to quickly test how different models handle the same coding task

The AI coding assistant market has exploded in 2024 and 2025, with tools like GitHub Copilot, Cursor, Cody by Sourcegraph, and Claude Code competing for developer mindshare. While Cursor has built multi-model support directly into its editor, VS Code users relying on the Claude Code extension have lacked this flexibility — until now.

The Rise of 'Vibe Coding' and Weekend Projects

The creator described the development process using the term 'vibe coding' — a concept popularized by Andrej Karpathy in early 2025 that refers to using AI assistants to rapidly build software by describing what you want rather than writing every line manually. The irony is not lost: an AI-assisted development tool was itself built using AI-assisted development.

This trend of developers quickly 'vibing' out useful tools and extensions has accelerated dramatically. The barrier to creating developer tools has dropped significantly as AI coding assistants handle boilerplate code, API integration patterns, and VS Code extension scaffolding. What once required days of reading documentation and debugging can now be accomplished in a weekend.

The open-source nature of the project also means the community can contribute improvements. Potential enhancements could include encrypted token storage, team-shared profile configurations, automatic provider health checking, and integration with secret management tools like 1Password CLI or HashiCorp Vault.

Getting Started With the Extension

Developers interested in trying Claude Code Profile Manager can find it on GitHub. The setup process is straightforward:

• Clone the repository or install from the VS Code marketplace
• Create profiles for each API provider you use
• Configure the base URL, token, and model for each profile
• Use the status bar button or command palette to switch between profiles
• Optionally import existing configurations from bash export statements

The extension is lightweight and focused — it does one thing and does it well. There is no telemetry, no cloud sync, and no unnecessary complexity. For developers who work with 2 or more API providers through Claude Code, it eliminates a genuine daily annoyance.

Looking Ahead: Multi-Provider Support as Standard

This extension likely represents a stopgap solution. As the AI coding assistant ecosystem matures, native multi-provider support will almost certainly become a standard feature in major tools and extensions. Anthropic itself may eventually build provider-switching capabilities directly into the Claude Code extension, similar to how Cursor already handles multiple model backends.

However, until that happens, community-built tools like Claude Code Profile Manager fill an important gap. They demonstrate what developers actually need in their daily workflows and often serve as inspiration — or even blueprints — for features that eventually get incorporated into official products.

The extension also highlights a growing reality: the AI tooling ecosystem is still young enough that significant quality-of-life improvements can come from a single developer spending a weekend on a focused problem. As the landscape continues to evolve at breakneck speed throughout 2025, expect more of these targeted, community-driven solutions to emerge and shape how developers interact with AI coding assistants.