@xmglab: https://x.com/xmglab/status/2101932146416075073
Summary
Jev is a new AI model from TypeSafe, focused on decision-making and analysis. It is now available to all users and can be integrated into coding agents like Claude Code and Codex via Skills, enhancing coding decision capabilities at a low cost.
View Cached Full Text
Cached at: 09/22/26, 07:51 AM
Jev Integration with Claude Code & Codex: A Step-by-Step Guide to Supercharge Your Coding Agent’s Decision-Making!
Hello everyone, I’m Xiao Ma Ge.
Yesterday, I shared how newcomers can quickly get started with Jev:
Xiao Ma Ge@xmglab·Sep 20 Article What exactly is the viral Jev? A beginner’s guide to getting started from scratch—step-by-step tutorial to quickly try it out! Hello everyone, I’m Xiao Ma Ge.
The Jev model has recently gained a lot of attention. Unlike traditional LLMs such as ChatGPT and Claude, Jev is quite different.
Jev was released on September 15, 2026, as the first System One model by TypeSafe… 631226413K
Jev is now open to everyone—no more waitlist, just sign up and start using it right away. The speed of this rollout is truly impressive.
Click here to start using it:
https://console.typesafe.ai
All new users currently receive an initial credit of 5 USD, which the official documentation says is enough for approximately 1.2 billion tokens. There’s no recharge option yet, but it may become available later.
Today, I’ll show you how to use Jev with Claude Code, Codex, and other Coding Agents. If you don’t have time to read now, feel free to bookmark this guide.
Obtain an API Key
Go to the API keys page to create an API Key:
https://console.typesafe.ai/keys
After obtaining the API Key, save it securely and set up a TYPESAFE_API_KEY environment variable locally. You can use it directly in AI tools later without worrying about leaks.
Jev Model Pricing
Jev is not free, but it’s incredibly affordable—costing 40 to 400 times less than other models. Output-side usage is currently free.
Test with curl
You can send a request directly:
curl -X POST https://api.typesafe.ai/v1/systemone \
-H "Authorization: Bearer $TYPESAFE_API_KEY" \
-H "Content-Type: application/json" \
-d @- <<'EOF'
{
"state": "My Claude account got banned and hasn’t been unbanned. I can’t use it. Could you help me check?",
"model": "jev-latest",
"questions": {
"urgency": {
"type": "noul",
"instructions": "Is this issue related to login?"
}
}
}
EOF
Test output result:
Use Jev in Coding Agents
Jev provides a plug-and-play Skill:
https://github.com/typesafe-ai/skills
It offers complete context for the TypeSafe API:
- Three types of questions
- Architectural patterns
- Best practices for organizational assessments
This Skill can be used with Claude Code, Codex, and other Coding Agent tools.
Using Jev in Claude Code
Install the Skill
Run the following two commands in your terminal to install:
claude plugin marketplace add typesafe-ai/skills
claude plugin install typesafe@typesafe-ai
Using Jev
Those familiar with Skills know you can simply say “Use TypeSafe skill…” to trigger it, or use the slash command “/typesafe:typesafe-ai” to invoke the TypeSafe Skill directly.
Here’s an example:
/typesafe:typesafe-ai Is it worth buying a Mac mini right now? Give me a direct judgment.
It returns a result in 12 seconds, advising against buying a Mac mini currently.
/typesafe:typesafe-ai Analyze this project and identify replaceable complex logic or weak code.
The output includes a TypeSafe section, indicating which code needs to be removed, retained, or extracted into shared components. TypeSafe provides the judgment here.
Update the Skill
This applies only to plugins installed via the Claude Code marketplace:
claude plugin marketplace update typesafe-ai
claude plugin update typesafe@typesafe-ai
Restart Claude Code or run /reload-plugins to load updates.
To enable auto-updates, open /plugin, then go to Marketplaces → typesafe-ai → enable auto-updates.
Using Jev in Codex and Other Tools
Install the Skill
Run the following installation command in your terminal:
npx skills add typesafe-ai/skills --skill typesafe-ai
By default, it includes mainstream Coding Agents such as Codex, Cursor, Gemini CLI, OpenCode. You can customize other selections below. Press Enter to complete the installation.
Using Jev
Similarly, in Codex, you use Jev just like any Skill. Type “/Typesafe” and it will auto-suggest:
Watch the video demo:
Update the Skill
If you installed via skills.sh, run:
npx skills update
If you manually copied the Skill, replace the entire skill directory with the latest version from GitHub.
Automatic Installation via Prompt
You can also use a prompt to install automatically:
Install the TypeSafe skill. If you are in Claude Code, run
claude plugin marketplace add typesafe-ai/skills, then runclaude plugin install typesafe@typesafe-ai. If you are in another Coding Agent, runnpx skills add typesafe-ai/skills --skill typesafe-aiand select your Coding Agent. Use one of these installation methods. You can read the skill documentation here: https://github.com/typesafe-ai/skills/blob/main/skills/typesafe-ai/SKILL.md (original: https://raw.githubusercontent.com/typesafe-ai/skills/main/skills/typesafe-ai/SKILL.md). After installation, you can use the TypeSafe skill while developing this project.
Paste this into any Skills-supported Coding Agent, and it will automatically install based on the current environment.
That’s all for today’s Jev sharing.
After testing it out, the most interesting thing about Jev is that it offers a completely different way of working. It doesn’t generate content—it focuses more on helping you with judgment, analysis, and decision-making.
So I believe the best way to use Jev is not alone, but in combination with Coding Agents like Claude Code or Codex—Claude Code/Codex handles the execution, while Jev handles the judgment.
Plus, since the official Skill is already provided, there’s no need to manually wrap APIs. Once installed, you can call it directly within your Coding Agent.
Currently, new users still have 5 USD credit, about 1.2 billion tokens, and Jev itself is very affordable—perfect for extensive testing.
Today I only demonstrated basic Jev applications, but it has several killer use cases. I’ll share more in the future and continue testing Jev’s performance in real-world projects, code refactoring, and complex technical decisions.
If you find this helpful, please follow—I’ll be sharing more hardcore insights.
I’m Xiao Ma Ge. See you next time.
Similar Articles
@Khazix0918: https://x.com/Khazix0918/status/2100850184129220829
Jev is a new AI model from TypeSafe AI, designed for high-speed decision-making. As a general-purpose classifier, it offers extremely fast response speeds and low cost, making it ideal for real-time tasks like content filtering, gaming, and trading.
@bozhou_ai: https://x.com/bozhou_ai/status/2100966488022864272
Jev is a closed-source AI model released by TypeSafe, designed for rapid judgment and selection tasks, featuring low latency and low cost. It is widely used in automated workflows and Agent systems.
@noisyb0y1: https://x.com/noisyb0y1/status/2102309802072272956
Jev is a new AI model by TypeSafe AI optimized for rapid, low-cost decision-making, offering significant speed and cost improvements over traditional AI models.
@sydneyrunkle: https://x.com/sydneyrunkle/status/2100754364545761643
TypeSafe AI releases Jev, a System One model for fast, structured decisions in agent loops, offering up to 200x faster inference and 400x lower cost for classification tasks compared to traditional LLMs.
@mvanhorn: https://x.com/mvanhorn/status/2100784142850097482
Jev is a new AI model focused on rapid decision-making, offering 20-200x faster and 40-400x cheaper performance than frontier LLMs, as released by Diogo Almeida.