Integrations / Codex
Launch and agentsDocumented
Run Codex tasks in an isolated environment
Send Codex a task, close your laptop, and pull the patch back when it finishes.
The steps follow the tool's documented settings. No tested run is recorded for this page yet.
Why use Badgr for Codex
- Badgr interface
- Launch
- Where Codex fits
- Agent sandbox
- Why Badgr
- Code tasks and test verification
- The agent runs on an isolated machine, not on your laptop or your production environment.
- You get a patch to review and test reports, and nothing is applied to your repository automatically.
- A default cost cap applies, and you can lower it with --max-cost.
What you can do
- Write tests for a module while you do something else
- Fix a bug in a repository and review the patch
- Run a long refactor with a cost cap
Quickstart: Send a task
badgr launch codex "Write tests for the checkout module"
How it works
- 1.You send a task
- 2.Badgr starts an isolated machine with the agent
- 3.The agent works and logs stream back
- 4.You pull the patch and check it before you apply it
What you get
- Logs while the task runs
- A patch you pull with badgr pull <id>
- Test reports and other files with badgr artifacts <id>
- A receipt
Requirements
- The Badgr CLI installed and logged in (npm install -g badgr-cli)
- A funded Badgr balance
- Your own OpenAI credential, which the CLI asks for the first time and remembers
- A repository or folder for the task
What it costs
- Agent tasks run on a CPU machine and bill for the time they run. A default cost cap applies, and --max-cost sets your own.
- The model provider bills you separately for tokens where your own key is used. Badgr does not quote a price for that.
See pricing for how billing works.
Proof
No tested run is recorded for this integration yet. A receipt and example output will be added here after a real run, not before.
Troubleshooting
Task cannot see my repository
Run the command from the project folder, or give the repository explicitly. Private repositories need access you grant.
Credential prompt
The first run asks for your key and remembers it. Run badgr connect openai to set it up ahead of time.
Patch fails tests
Run your tests on the patch before merging. Use a check command with badgr job to have Badgr run them.
Developer reference
Ready to start with Codex?
Every run or service has a spending limit you set before it starts.