TasteSteak

TasteSteak / CodeSeeX

Public

CodeSeeX is a local Codex-compatible bridge for DeepSeek, with a desktop manager for proxy status, usage tracking, logs, tools, and configuration.

17
2
69% credibility
Found May 19, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

CodeSeeX is an unofficial bridge application that lets you use DeepSeek AI models with Codex, OpenAI's coding assistant. It works by running a local service on your computer that Codex connects to, which then forwards your requests to DeepSeek and brings back the answers. The project includes a desktop manager window where you can monitor your usage, adjust settings like thinking display and language preferences, and view logs. It's designed for developers who want to combine Codex's interface with DeepSeek's capabilities, and clearly states it's not affiliated with OpenAI or DeepSeek.

How It Works

1
💡 You want to use DeepSeek with Codex

You've heard about DeepSeek AI models and want to use them inside Codex, but Codex doesn't natively support DeepSeek yet.

2
📥 You download CodeSeeX

You find CodeSeeX on GitHub and download the release for your computer. It comes as a simple desktop app.

3
🔌 CodeSeeX creates a bridge

CodeSeeX acts as a middleman between Codex and DeepSeek, translating requests so both can understand each other.

4
⚙️ You connect Codex to CodeSeeX

You copy a simple configuration into Codex that tells it to send its requests to CodeSeeX running on your computer instead of OpenAI.

5
You manage everything your way
👁️
You watch your AI usage

See how many requests you've made, how much it cost, and how long each one took.

🔧
You tweak settings

Turn thinking display on or off, change language, adjust how long logs are kept.

6
🤖 You code with DeepSeek inside Codex

Now when you use Codex, your questions go to DeepSeek and the answers come back through CodeSeeX, all working seamlessly.

✨ You get the best of both worlds

You enjoy Codex's polished coding interface while DeepSeek powers the AI behind it, with full visibility into everything.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is CodeSeeX?

CodeSeeX is a local proxy that lets you run Codex, OpenAI's coding agent, with DeepSeek models instead of default options. It sits between Codex and DeepSeek's API, exposing a Codex-compatible endpoint while forwarding requests upstream. The package includes an Electron desktop app with a web-based manager UI for monitoring status, viewing logs, tracking token usage, and adjusting settings. Built-in tools handle file patching, web searches, and workspace operations through the proxy. Configuration happens through a simple TOML file in Codex pointing to the local proxy running on a single configurable port.

Why is it gaining traction?

The hook is straightforward: use DeepSeek V4 inside Codex without being locked to OpenAI's models. Developers who want DeepSeek's pricing or reasoning capabilities but prefer Codex's interface get a working bridge. The streaming support, reasoning display controls, and token usage tracking give visibility into what's actually happening. The model catalog adapter means DeepSeek models appear as native Codex options. For teams already using Codex, this avoids the friction of switching workflows entirely.

Who should use this?

Developers running Codex Desktop who want DeepSeek model access will find the setup documented and functional. Teams evaluating DeepSeek V4 for coding tasks without leaving their existing Codex workflow. Users comfortable with local proxies who need visibility into API usage patterns. Not suitable for production infrastructure at this stage given the early version and limited community validation.

Verdict

This fills a real gap for Codex users wanting DeepSeek integration, and the implementation covers the essential features. However, with 17 stars and a credibility score of 0.699999988079071%, the project is extremely early and lacks the battle-testing most developers expect for tool-wide infrastructure. The AGPL license and TOML configuration approach work well for personal use. Treat this as an experimental setup until the community and documentation mature.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.