# OpenAI-compatible /v1/chat/completions for Cloud API

**URL:** <https://forum.cursor.com/t/openai-compatible-v1-chat-completions-for-cloud-api/164522>\
**Category:** Feature Requests\
**Tags:** cursor-sdk, cloud-agents\
**Created:** [July 1, 2026, 8:29am UTC](https://forum.cursor.com/t/openai-compatible-v1-chat-completions-for-cloud-api/164522 "2026-07-01T08:29:59Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![breakonly](https://avatars.discourse-cdn.com/v4/letter/b/cab0a1/32.png) [@breakonly](https://forum.cursor.com/u/breakonly)\
**Post date:** [July 1, 2026, 8:29am UTC](https://forum.cursor.com/t/openai-compatible-v1-chat-completions-for-cloud-api/164522/1 "2026-07-01T08:29:59Z")

</div>

### Describe the request

## Summary

Please provide an official OpenAI-compatible API (especially `POST /v1/chat/completions`, with optional streaming) for Cursor Cloud API, so third-party agent frameworks can integrate using the standard OpenAI SDK without custom adapters.

## Problem

We are building multi-agent workflows and internal automation that need to call Cursor programmatically from other systems (orchestrators, LangChain-style pipelines, enterprise agent platforms, etc.).  
Today, the Cursor API Key (`crsr_...`) exposes:

- Cloud Agents API (`POST /v1/agents`, runs, polling)
- Python / TypeScript SDK wrappers  
There is **no official** OpenAI-compatible chat endpoint such as:
- `POST /v1/chat/completions`
- Standard request/response shape: `messages` → `choices[0].message.content`  
As a result, other agents cannot call Cursor using the de-facto standard OpenAI client libraries.

## Use case

We need lightweight LLM-style calls for:

- Q&A and planning inside a multi-agent pipeline
- Synchronous request/response integration
- Standard `messages` input and text output
- Optional SSE streaming  
The current Cloud Agents API is powerful for repo-changing tasks, but it is a different model:
- Async runs with polling (often 30–60+ seconds)
- Agent/workspace lifecycle instead of simple chat
- Harder to plug into existing OpenAI-based orchestration tools

## What we are asking for

Please provide **one** of the following (preferably #1):

1. **Official OpenAI-compatible endpoint** on `api.cursor.com`:
  - `POST /v1/chat/completions`
  - `GET /v1/models`
  - Optional streaming
  - Bearer auth with existing API keys

2. **Official adapter / gateway** documented and supported by Cursor, mapping Agent API to OpenAI chat semantics
3. **Official integration guidance** for third-party agent frameworks if OpenAI compatibility is not planned

## Current workaround (not ideal)

- Use internal OpenAI-compatible LLM gateways for chat
- Use Cloud Agents API only for code/repo automation
- Build custom adapters around `/v1/agents`
- Community proxies wrapping Cursor CLI are unofficial and not suitable for enterprise production

## Why this matters

OpenAI-compatible APIs are the common integration layer across agent ecosystems. Without an official Cursor endpoint, every team must build and maintain its own adapter, which increases cost, fragility, and slows adoption of Cursor in automated / multi-agent environments.  
Thank you for considering this.