菜单
Monitors, diagnoses and troubleshoots RAG system health. Verifies Azure AI Search connectivity, index status, configuration, and provides real-time monitoring with actionable error reports.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | rag-diagnostics |
| description | Monitors, diagnoses and troubleshoots RAG system health. Verifies Azure AI Search connectivity, index status, configuration, and provides real-time monitoring with actionable error reports. |
Monitor, diagnose and troubleshoot your RAG system.
Collection of diagnostic and monitoring tools to verify Azure AI Search health, index status, and system configuration.
Check the general health of the RAG and status of components.
python .github/skills/rag-diagnostics/system-status.py
Output:
========================================================================
RAG SYSTEM STATUS REPORT
========================================================================
PHASE 1: Keyword + Semantic Search
Status: Running
Items processed: 113
Items failed: 0
Duration: 245000 ms
Index: rag-documents
PHASE 2: Vector Search
Status: Running
Items processed: 86
Items failed: 0
Duration: 123000 ms
Index: rag-documents-vectors
INDEX STATISTICS
rag-documents: 113 documents
rag-documents-vectors: 86 documents
Deep analysis of Azure AI Search configuration and issues.
python .github/skills/rag-diagnostics/diagnose.py
Output:
1 INDEXES
rag-documents
- Fields: 7
- Vectors: No
2 DATA SOURCES
blob-storage
- Type: AzureBlobStorage
3 SKILLSETS
ocr-skillset
- Skills: 4
- Types: OcrSkill, SplitSkill, MergeSkill
4 INDEXERS
blob-indexer
- Status: Running
- Schedule: Every hour
Continuous monitoring of indexer activity.
python .github/skills/rag-diagnostics/monitor.py
Output:
Monitoring indexer: blob-indexer
Press Ctrl+C to stop
[14:23:45] Status: Running | Processed: 45 | Failed: 0
[14:24:10] Status: Running | Procesados: 89 | Fallidos: 1
[14:24:35] Status: Completed | Procesados: 113 | Fallidos: 0
pip install -r .github/requirements.txt
.env with Azure AI Search credentials:
AZURE_SEARCH_ENDPOINTAZURE_SEARCH_KEYpython .github/skills/rag-diagnostics/system-status.py
python .github/skills/rag-diagnostics/diagnose.py
# View indexing in real time
python .github/skills/rag-diagnostics/monitor.py
| Issue | Diagnosis | Solution |
|---|---|---|
| Empty index | system-status.py shows 0 docs | Run rag-indexer skill |
| Indexer failed | diagnose.py shows status: Failed | Verify credentials in .env |
| Semantic search not working | Index mysing semantic config | Recreate index with semantic enabled |
| Indexing slow | monitor.py shows low throughput | Increase Search tier or batch size |
from system_status import check_status
status = check_status()
if status['index_count'] == 0:
print("No documents indexed yet")
else:
print(f"{status['index_count']} documents ready")
# Health check before deployment
python .github/skills/rag-diagnostics/diagnose.py || exit 1
Scale up or scale down Azure RAG configurations (Search, Log Analytics, Insights) and manage budgets/alerts automatically. Reversible changes with cost calculation before applying.
PowerShell helper for obtaining Azure Blob Storage credentials via Azure CLI. Provides connection strings used by RAG indexers and document upload pipelines to access Blob Storage.
Design document ingestion and indexing workflows for Azure AI Search, including chunking, metadata strategy, and incremental reindexing guidance.
Exposes RAG functionality as a REST API for external applications. Provides HTTP endpoints for document search and query with JSON request/response, async processing, CORS support, and observability metrics.
Professional executive report generation using Claude Opus 4.7. Generates high-quality DOCX reports with professional formatting, compelling narratives, and quantified impact metrics. Perfect for client presentations and stakeholder communication.
Hybrid-professional SharePoint integration for RAG. Two modes: Professional (Azure Search indexer, real-time sync, no duplication) or Local (download to knowledge/, coexists with traditional docs)