Empire LLM for Codex · v1.7.2

A second opinion.
Inside Codex.

Ask another AI to review your code, help with a plan, or create an image. Empire brings the result back to Codex to check. Your task stays in one place.

Get the plugin See how it works
  • Choose what to share
  • Set a spending limit
  • Codex checks the result

Check these changes for bugs and missing tests.

Another AI reviews

This change may leave an old request running.

Codex checks

Check the code and test the reported issue.

You see the result

Read the findings and decide what to use.

Kimi Qwen Claude

Example workflow. The model and result depend on your request.

Get started

Add Empire to Codex.

Find Empire LLM for Codex in the Plugin Store. Install it, start a new task, and ask for help. GitHub and the version 1.7.2 download are also available below.

  1. Open Codex → Plugins
  2. Search “Empire LLM for Codex”
  3. Select Install, then start a new task
Install from the terminal
git clone https://github.com/web5labs/empire-llm-codex.git cd empire-llm-codex codex plugin marketplace add "$PWD" codex plugin add empire-llm-codex@empire-local

Start with one of these

Review my recent code changes for mistakes and suggest improvements. Compare today's leading AI models and show me which ones best fit my task. Help me connect my AI services and set a spending limit.

Open Plugin Store Download v1.7.2 View source on GitHub
1 workspaceKeep working in Codex
1 outside viewOne model for a focused request
1 spending limitChoose what you can spend
0 direct editsOutside models cannot change files
Kimi Qwen Gemini Grok Claude OpenRouter

More help, close at hand

Review. Plan. Create.

Get another view on your code, turn an idea into a draft, or compare models for your next task. Codex checks what comes back.

01 / Review

Give your code a second look.

Ask another model to look for bugs, broken behavior, and missing tests in selected changes. Codex checks the findings against your project.

  • Selected code and context
  • Saved partial replies
  • See the model and cost

02 / Handoff

Start with something you can review.

Create a plan, a proposed code file, an image, or a short video. Preview the result before you bring it into your project.

  • Files checked on return
  • Preview before use
  • No automatic code changes

03 / Control

Choose the model. Know the cost.

Compare models by test results, price, speed, and task fit. Set a free-only rule, a limit for one request, or a project budget.

  • Keys stored securely
  • Clear spending limits
  • No silent paid retries

For the way you work

Keep your work moving.

Use an extra AI view when you want to check a decision, explore an idea, or find a better next step.

01

Building on your own

Get another view on a tricky change without copying your work between chats.

02

Working with a team

Bring extra findings to a review while Codex keeps control of project edits.

03

Trying a new idea

Turn a short brief into a plan or a draft you can inspect.

From question to next step

Ask. Check. Keep going.

  1. 01

    Choose

    Codex selects the code and context for your question.

  2. 02

    Match

    Empire checks suitable models and your spending limit.

  3. 03

    Ask

    One outside model reviews the request and replies.

  4. 04

    Check

    Codex compares the findings with your project.

  5. 05

    Use

    Review the result and choose the next step.

Explore your options

What matters for this task?

Try a preference to see how model selection can change. This demo uses example models and scores. It does not make a model request.

Worker route

Kimi K3

moonshotai/kimi-k3

Demo score9.4/10
Task fitExcellent
EfficiencyFocused
EvidenceBounded
Example synthesis< $0.10 ceiling

Kimi contributes a front-end proposal. Codex verifies accessibility, repository fit, and implementation details before writing any project file.

Demo only · models and scores are examples

You set the limits

Your project. Your call.

An outside model gets only the code and context chosen for the request. It cannot edit your project. Store service keys in your system’s secure key store and set a spending limit before paid work.

Read how protection works
  • Published in the OpenAI Plugin StoreVersion 1.7.2 is available now
  • Secure key storageKeep API keys out of chat
  • Small, focused requestsUp to 8 files and 80 KB by default
  • Another model’s viewCodex stays in charge
  • Review before useOutside models have no file access