codex-auto-review API
Codex Auto Review is the automatic approval-review model in OpenAI’s Codex catalog. It assesses proposed actions against user authorization and environment constraints. Calling it through XiuAI does not install Codex sandboxing, approval policies or execution permissions.[1][2][6]
codex-auto-reviewModel overview
Published provider specifications. Available features depend on the current API route.
- API routes
- OpenAI-compatible API / openai-response / openai-response-compact / Anthropic Messages / Gemini GenerateContent / openai-alpha-search
What the model reviews
Its documented role is to review whether a proposed agent action, such as a command, network access or file operation, fits user authorization and applicable policy. That differs from reviewing a pull request, scanning vulnerabilities or providing general coding assistance; the word review does not make it a replacement for those products.[1][2]
If your application already defines authorization boundaries, evaluate review results with examples containing the user goal, proposed action and constraints. Your application supplies that context and integrates the result with its approval flow. A model call alone is not a complete permission system.[2][3]
Separate assessment from enforcement
Codex automatic review uses policy, conversation context and environment state, while the execution environment controls sandbox and tool permissions. A response obtained through XiuAI does not change file, network or account permissions, and should not act as a credential for bypassing existing authorization checks.[2][3]
The model identifier is separate from the client’s automatic-review feature. Authentication methods or providers may select different reviewer models. Choosing codex-auto-review does not configure the same policy, context or tools in Codex. Context settings in a public client catalog are not measured limits of an XiuAI route.[4][1]
API access
Access through XiuAI
Use the exact ID codex-auto-review with an existing XiuAI API key and a group that permits access. Follow the current API instructions on this page: Responses uses input, while a compatible chat endpoint uses messages. Client-specific configuration is not automatically an API parameter.[6][5]
Reconcile charges using the applicable XiuAI input, output and cache rates and actual usage. Automatic-review allowances or benefits under a particular Codex sign-in plan do not establish free XiuAI API calls or identical billing. Check this page’s prices and your request records.[6][2]
Send a request using your existing API key.
POST /v1/chat/completionscurl 'https://router-api.xiu.ai/v1/chat/completions' \
-H "Authorization: Bearer $XIUAI_API_KEY" \
-H 'Content-Type: application/json' \
--data-raw '{
"model": "codex-auto-review",
"messages": [
{
"role": "user",
"content": "Hello"
}
]
}'Read the API documentationFrequently asked questions
Is this a pull-request review model?
OpenAI’s catalog describes automatic approval review for Codex: evaluating proposed actions against authorization. Pull-request review and Codex Security serve different purposes; their product descriptions are not capability statements for this ID.[1][2]
Does a successful call authorize or execute an action?
No. The API returns a model result. Your application still decides what happens under user authorization, applicable policy and tool permissions. Model output, approval and execution are separate states.[2][3]
What makes a useful review input?
Describe the authorized goal, concrete proposed action, target resources and applicable limits. Evaluate known allowed and denied cases against your approval flow instead of asking only “is this safe?” without action and authorization context. Do not include unnecessary credentials or account information.[3][2]
Why not copy Codex client settings directly into the request?
The client catalog describes model selection, tool behavior and session policy, rather than a general model API schema. Check the fields accepted by your protocol and configure the model ID, client review settings and account permissions separately.[1][5]
Related models
gpt-6.1-solgpt-6-lunaSources and updates
Specifications reference provider documentation; prices and availability come from the current XiuAI catalog.
Content reviewed: 2026-11-16
[1] OpenAI Codex model catalog[2] OpenAI automatic approval review[3] OpenAI reviewer context configuration[4] OpenAI provider reviewer selection[5] OpenAI Responses API guide[6] Current XiuAI model and pricing data