完善文档

This commit is contained in:
chenxin committed 2026-09-22 10:58:39 +08:00
1 parent 192ea96eb4
commit efc08dab5e
82 files changed
+1455 -109

No files matched your search

+3
View File
@@ -0,0 +1,3 @@
# Testing
测试与评估文档目录。
+13
View File
@@ -0,0 +1,13 @@
# Agent Evaluation
评估维度:
- Task Success
- Tool Selection
- Plan Quality
- Recovery
- Hallucination
- Cost
- Latency
评估数据集需要版本化。
+13
View File
@@ -0,0 +1,13 @@
# Integration Test
覆盖:
- PostgreSQL
- Redis
- VectorStore
- GraphStore
- FileStorage
- MCP
- Tool Runtime
外部服务优先使用隔离测试环境或 Mock。
+13
View File
@@ -0,0 +1,13 @@
# RAG Evaluation
评估:
- Retrieval Recall
- MRR
- NDCG
- Context Relevance
- Faithfulness
- Answer Accuracy
- Source Accuracy
模型、Embedding、Chunking、Retriever 变化时应重新运行基准集。
+18
View File
@@ -0,0 +1,18 @@
# Testing Strategy
测试层次:
```mermaid
flowchart TB
Unit --> Integration
Integration --> API
API --> Workflow
Workflow --> Evaluation
Evaluation --> E2E
```
Agent 系统除了传统测试,还必须建立:
- Agent Evaluation
- RAG Evaluation
- Tool Reliability Evaluation
+12
View File
@@ -0,0 +1,12 @@
# Unit Test
单元测试重点:
- Domain Logic
- Planner
- Memory Manager
- Tool Schema
- Permission
- Data Transformation
避免在单元测试中真实调用外部 LLM 或第三方 API。