Qoder Cloud Agents CN is a fully managed AI agent runtime platform. You do not need to build your own agent loop, manage a tool execution sandbox, or handle long-lived connections. Simply define an agent and start a session through the API to run complex tasks and receive real-time results in the cloud.
The new Qoder CN documentation site is now available. Content on this site may not reflect the latest product features and changes. Visit docs.qoder.cn for the latest product documentation.
Key concepts
|
Concept |
Description |
Analogy |
|
agent |
A reusable configuration template that defines the model, system prompt, and toolset. |
An employee's job description |
|
environment |
The container environment where a session runs, including network and dependency configurations. |
A desk and a toolbox |
|
session |
A specific instance of a conversation or task execution. |
A specific work session |
|
event |
A real-time stream of events generated within a session. |
A real-time broadcast of work progress |
Workflow
-
Define an agent — Specify the model, system prompt, and available tools.
-
Configure an environment — Select the container type, network policy, and pre-installed dependencies. New accounts must first use
POST /environmentsto create an environment, as none is provided by default. -
Start a session — Bind an agent to an environment to create a runtime instance.
-
Send messages and receive events — Send a
user.messageto the session, and then receive a real-time stream of events—such as agent thoughts, messages, and status changes—through Server-Sent Events (SSE) or polling.
Verify connectivity
# Verify PAT and list agents
curl -s https://api.qoder.com.cn/api/v1/cloud/agents \
-H "Authorization: Bearer $QODER_PAT"
A successful response looks like this:
{
"data": [ ],
"first_id": null,
"last_id": null,
"has_more": false
}
Use cases
-
Long-running asynchronous tasks — Code reviews, large-scale refactoring, and automated test generation.
-
API integration — Embed agent capabilities into your backend services without maintaining a runtime.
-
Batch processing — Start multiple sessions in parallel to process batch requests.
-
Scheduled tasks — Integrate with a scheduling system to periodically run agents for inspections or reporting.
Authentication
All API requests must include the following header:
|
Header |
Value |
Description |
|
|
|
Personal access token |
You can create a personal access token (PAT) in the Qoder console by navigating to Settings > Personal Access Tokens. Store your PAT securely and do not commit it to a code repository.
Pagination
List APIs use cursor-based pagination. The response structure is as follows:
{
"data": [...],
"first_id": "agent_019e451902fe7a2ca42c2dfc62d9320e",
"last_id": "agent_019e45369b3379e18bfaf59b3aad2fc9",
"has_more": true
}
Use the after_id and before_id query parameters to paginate.
FAQ
Q: Can I use Qoder Cloud Agents CN and the Qoder CLI at the same time?
A: Yes. The Qoder CLI is ideal for local, interactive development, while Qoder Cloud Agents CN is designed for automation and integration. The two are complementary.
Q: How many sessions can a single agent run concurrently?
A: There is no hard limit. An agent configuration can be associated with multiple active sessions simultaneously.
Q: How is data security ensured?
A: Each session runs in an isolated container sandbox, preventing access between sessions. All data is cleared when the environment is destroyed.
Next steps
-
Quickstart: Run your first cloud agent in five steps.
-
Define an agent: Learn more about agent configuration.
-
Cloud environments: Configure a runtime environment.
-
Start a session: Manage the session lifecycle.