Skip to content

qca-runtime.md

Latest commit

 

History

History
29 lines (19 loc) · 1.62 KB

qca-runtime.md

File metadata and controls

29 lines (19 loc) · 1.62 KB
Raw

QCA Runtime / QCA 运行时

Operra uses Qoder Cloud Agents as its only execution engine.

Operra 只使用 Qoder Cloud Agents 作为执行引擎。

Authentication / 认证

Create a run token at Qoder integrations and enter it in Operra Settings. Do not put the token in a project repository or task prompt.

在 Qoder integrations 创建 run token，并在 Operra 设置中填写。不要把 token 写进项目仓库或任务提示。

Execution lifecycle / 执行生命周期

  1. Operra creates or resumes a QCA session for a Task.

  2. QCA works against the connected GitHub repository.

  3. Streaming QCA events are normalized into plans, activity, review requests, failures, and completion.

  4. Follow-ups continue the same task context; interruption requests stop active work.

  5. Operra reconciles cloud state after network loss or application restart.

  6. Operra 为 Task 创建或恢复 QCA session。

  7. QCA 在连接的 GitHub 仓库中执行工作。

  8. QCA 流式事件被归一化为计划、活动、审核请求、失败与完成。

  9. 追问沿用同一任务上下文；中断请求停止正在进行的工作。

  10. 网络断开或应用重启后，Operra 会与云端状态对账。

Deployment-backed tasks use QCA deployments and share the same normalized Task model. The adapter is deliberately narrow so the product domain does not depend on transport payload details.

基于 deployment 的任务使用 QCA deployments，并共享同一套 Task 模型。适配器边界刻意保持狭窄，避免产品领域依赖传输层 payload 细节。