Discover and explore top open-source AI tools and projects—updated daily.
mistydewDeepSeek-focused desktop AI client for advanced local workflows
New!
Top 87.4% on SourcePulse
This project provides a desktop GUI application, TOKENICODE DeepSeek Alpha, a fork of TOKENICODE, tailored for local Large Language Model (LLM) workflows, particularly with DeepSeek/CC Switch and Codex skills. It targets developers, researchers, and power users seeking a streamlined interface for interacting with local LLMs, managing AI-powered coding tools, and enhancing productivity through integrated features like file browsing, code editing, and web previews.
How It Works
TOKENICODE DeepSeek Alpha enhances the original TOKENICODE framework with specific modifications for local LLM usage. It adapts the UI and API handling for DeepSeek/CC Switch models (e.g., mapping DeepseekV4Pro to deepseek-v4-pro), introduces independent provider configurations supporting Anthropic/OpenAI formats with custom base URLs, API keys, model mappings, and proxies. The project integrates Codex/Agent skills management, allowing scanning, editing, and translation of skill descriptions and previews, all built upon the Tauri 2 + React 19 desktop application framework with local file and web preview capabilities.
Quick Start & Requirements
.dmg files).Highlighted Details
DeepseekV4Pro/DeepseekV4Flash, automatically mapping to actual API models for seamless integration.Maintenance & Community
This repository is a personal fork ("魔改版") of the original TOKENICODE project, focusing on specific local LLM workflows. Maintenance is handled within this mistydew/tokenicode-deepseek-alpha repository. Updates are provided via GitHub Releases. macOS builds currently lack automatic update signing due to missing Tauri signing keys. No explicit community channels (Discord/Slack) are mentioned.
Licensing & Compatibility
Licensed under Apache License 2.0, inheriting from the original TOKENICODE project. This license is generally permissive for commercial use and integration, allowing modification and distribution, provided attribution and license terms are met.
Limitations & Caveats
As a fork, it diverges from the original TOKENICODE project, focusing on specific enhancements. macOS builds currently lack automatic update signing. Users must manually configure API providers and potentially Ollama for local models. The project's primary focus is on DeepSeek/CC Switch and Codex skills workflows, which may not encompass all functionalities or use cases of the original TOKENICODE.
4 days ago
Inactive
SilasMarvin