メニュー
Use this skill when working with Agent conversations, memory, or streaming output. Covers conversationId lifecycle, MemoryManager dual-track memory (Redis hot + DB cold), and executeStream context loading.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use this skill when designing REST APIs, creating controllers, or defining DTOs. This skill enforces RESTful API design standards, request/response formats, error handling, and parameter validation.
Use this skill when designing module structure, creating new modules, or organizing code layers. This skill enforces Lumina's simplified layered architecture pattern with clear separation of API, Service, Domain, and Infrastructure layers.
Use this skill when writing Java code, creating new classes, or reviewing code style. This skill enforces Lumina framework coding standards including naming conventions, package structure, class organization, and code quality practices.
Use this skill when designing domain entities, creating business logic, or implementing domain methods. This skill enforces domain-driven design principles with simplified implementation, ensuring business logic is encapsulated in domain entities.
Use this skill when designing, building, or modifying frontend UI — Vue 3 SFC, Element Plus components, CSS/SCSS, layouts, themes, or animations. Enforces Luminous design system consistency and accumulates design learnings across sessions.
Use this skill when generating git commit messages for staged changes. This skill enforces consistent commit message formatting, categorization of changes, and best practices for commit documentation.
| name | lumina_conversation |
| description | Use this skill when working with Agent conversations, memory, or streaming output. Covers conversationId lifecycle, MemoryManager dual-track memory (Redis hot + DB cold), and executeStream context loading. |
本技能包用于规范 Lumina 框架的 Agent 多轮对话、会话记忆管理和流式输出，涵盖 conversationId 生命周期、MemoryManager 双轨记忆（Redis 热记忆 + DB 冷存储）、executeStream 上下文加载机制。
| 表 | 说明 | 迁移版本 |
|---|---|---|
lumina_conversation | 会话主表（UUID、标题、Agent 关联） | Flyway V3 |
lumina_message | 消息记录（角色、内容、会话关联） | Flyway V3 |
POST /api/v1/conversations，生成 UUID 作为 conversationIdlumina_message 表@Autowired
private MemoryManager memoryManager;
// 加载上下文（从 Redis 热记忆）
List<Msg> context = memoryManager.loadMessages(conversationId);
// 保存消息（写入 Redis 热记忆）
memoryManager.saveMessage(conversationId, Msg.user(task));
memoryManager.saveMessage(conversationId, Msg.assistant(response));
| 属性 | 值 |
|---|---|
| 存储 | Redis |
| 上下文窗口 | CONTEXT_WINDOW = 20 条最近消息 |
| 特性 | 低延迟读取，自动滑动窗口裁剪 |
| 用途 | Agent 执行时快速加载对话上下文 |
| 属性 | 值 |
|---|---|
| 存储 | MySQL lumina_message 表 |
| 特性 | 永久持久化，全量消息记录 |
| 用途 | 历史回放、审计、分析 |
// 异步执行
Result<AgentResponse> execute(AgentExecutionRequest request);
// 同步执行
Result<AgentResponse> executeSync(AgentExecutionRequest request);
// 流式执行
SseEmitter executeStream(AgentExecutionRequest request);
AgentExecutionRequest 包含 conversationId 字段，用于关联会话上下文。
1. 接收 conversationId
2. buildContextMessages(conversationId)
→ 从 MemoryManager 加载历史消息（Redis 热记忆，最多 20 条）
→ 映射为 Msg 列表（MsgRole 映射：USER → user, ASSISTANT → assistant, SYSTEM → system）
3. 将当前用户输入追加到 Msg 列表
4. 执行 Agent（传入完整 Msg 列表）
5. 执行完成后，保存 user 消息和 assistant 响应到双轨记忆
| 数据库角色 | AgentScope MsgRole | 说明 |
|---|---|---|
USER | MsgRole.USER | 用户消息 |
ASSISTANT | MsgRole.ASSISTANT | Agent 响应 |
SYSTEM | MsgRole.SYSTEM | 系统提示词 |
| 方法 | 路径 | 说明 |
|---|---|---|
| POST | /api/v1/conversations | 创建会话 |
| GET | /api/v1/conversations | 会话列表 |
| GET | /api/v1/conversations/{uuid}/messages | 获取历史消息 |
| POST | /api/v1/agents/{id}/execute/stream?task=&conversationId= | 多轮流式执行 |
POST /api/v1/agents/{agentId}/execute/stream?task=你好&conversationId=abc-123
Accept: text/event-stream
→ data: {"chunk": "你"}
→ data: {"chunk": "好"}
→ data: {"done": true, "messageId": 456}
GET /api/v1/conversations/{uuid}/messages 回显历史