{"id":85813,"date":"2026-07-27T22:42:49","date_gmt":"2026-07-27T14:42:49","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/85813.html"},"modified":"2026-07-27T22:42:49","modified_gmt":"2026-07-27T14:42:49","slug":"ai%e9%a1%b9%e7%9b%ae%e4%bb%8e%e5%85%a5%e9%97%a8%e5%88%b0%e4%b8%8a%e7%ba%bf21-%e4%bb%8e%e9%9c%80%e6%b1%82%e5%88%86%e6%9e%90%e5%88%b0%e9%83%a8%e7%bd%b2%e8%bf%90%e7%bb%b4%e5%85%a8%e8%ae%a9ai%e5%b9%b2","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/85813.html","title":{"rendered":"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8"},"content":{"rendered":"

<\/h2>\n

\u201c\u8001\u677f，\u6211\u4eec\u7684DevOps\u5e73\u53f0\u63a5\u5165\u4e86\u5927\u6a21\u578b\u2014\u2014\u73b0\u5728\u4e0d\u4ec5\u4ee3\u7801\u662fAI\u5199\u7684，\u8fde\u9700\u6c42\u6587\u6863\u3001\u6d4b\u8bd5\u7528\u4f8b\u3001\u90e8\u7f72\u811a\u672c\u3001\u8fd0\u7ef4\u544a\u8b66\u90fd\u662fAI\u5728\u641e，\u4f60\u8bf4\u6211\u4eec\u7a0b\u5e8f\u5458\u8fd8\u80fd\u5e72\u5565？\u201d \u201c\u2026\u2026\u6478\u9c7c\u3002\u201d<\/p>\n

\n

\u76ee\u5f55<\/h3>\n

\u4e00\u3001\u5f00\u7bc7\u66b4\u51fb：AI DevOps\u5230\u5e95\u662f\u4e0d\u662f\u753b\u997c？<\/p>\n

\u4e8c\u3001\u4e00\u5f20\u56fe\u770b\u7a7fAI DevOps\u5168\u666f<\/p>\n

\u4e09\u3001\u5404\u73af\u8282AI\u5207\u5165\u5b9e\u6218<\/p>\n

3.1 \u9700\u6c42\u2192\u7528\u6237\u6545\u4e8b：\u522b\u8ba9\u4ea7\u54c1\u7ecf\u7406\u51993\u5929PRD\u4e86<\/p>\n

3.2 \u4ee3\u7801\u751f\u6210：Copilot\u53ea\u662f\u524d\u83dc<\/p>\n

3.3 \u6d4b\u8bd5：AI\u5199\u6d4b\u8bd5\u7528\u4f8b\u6bd4\u4f60\u5199\u4ee3\u7801\u8fd8\u5feb<\/p>\n

3.4 \u90e8\u7f72\u4e0e\u8fd0\u7ef4：AI\u5e2e\u4f60"\u770b\u5bb6"<\/p>\n

\u56db\u3001LLM\u7f16\u6392\u5f15\u64ce\u2014\u2014\u5e73\u53f0\u7684"\u5927\u8111"<\/p>\n

4.1 Prompt\u6a21\u677f\u7ba1\u7406\u2014\u2014\u522b\u6bcf\u6b21\u90fd\u624b\u6413Prompt<\/p>\n

\u4e94\u3001\u77e5\u8bc6\u5e93\u4f53\u7cfb：\u8ba9AI\u8bb0\u4f4f\u4f60\u7684\u9879\u76ee\u7ecf\u9a8c<\/p>\n

\u516d\u3001\u5b89\u5168\u5408\u89c4：\u522b\u8ba9AI\u628a\u4f60\u7684\u4ee3\u7801\u6cc4\u9732\u5230\u516c\u7f51<\/p>\n

\u4e03\u3001MVP\u843d\u5730：48\u5c0f\u65f6\u8dd1\u901aFlask\u5168\u6d41\u7a0b<\/p>\n

MVP\u6280\u672f\u6808（\u6700\u5c0f\u53ef\u884c\u7248）<\/p>\n

\u516b\u3001\u67b6\u6784\u5168\u666f：L5\u5230\u5e95\u957f\u4ec0\u4e48\u6837<\/p>\n

\u4e5d\u3001\u26a0\ufe0f\u8e29\u5751\u5b9e\u5f55\u4e0e💡\u6548\u7387\u6280\u5de7<\/p>\n

\u26a0\ufe0f \u8e29\u5751\u5927\u5168（\u8840\u6cea\u6362\u6765\u7684\u7ecf\u9a8c）<\/p>\n

💡 \u6548\u7387\u6280\u5de7（\u8ba9\u4f60\u5c11\u52a0\u73ed）<\/p>\n

\n

\u4e00\u3001\u5f00\u7bc7\u66b4\u51fb：AI DevOps\u5230\u5e95\u662f\u4e0d\u662f\u753b\u997c？<\/h3>\n

2025\u5e74\u4e86，\u4f60\u8fd8\u5728\u624b\u5199\u9700\u6c42\u6587\u6863\u3001\u624b\u6413\u5355\u5143\u6d4b\u8bd5\u3001\u624b\u52a8\u6572git push\u7136\u540e\u7948\u7977CI\u522b\u6302？<\/p>\n

\u606d\u559c\u4f60，\u4f60\u7684DevOps\u8fd8\u505c\u7559\u5728L2\u6c34\u5e73\u2014\u2014\u76f8\u5f53\u4e8e2020\u5e74\u7684"\u6211\u4eec\u4f1a\u5199Dockerfile，\u6211\u4eec\u5f88\u5148\u8fdb"\u3002<\/p>\n

GitHub Copilot\u53ea\u662f\u5f00\u80c3\u83dc\u3002 \u771f\u6b63\u7684AI DevOps，\u662f\u628a\u5927\u6a21\u578b\u585e\u8fdb\u8f6f\u4ef6\u5de5\u7a0b\u7684\u6bcf\u4e00\u4e2a\u6bdb\u5b54\u91cc：\u9700\u6c42\u5206\u6790\u3001\u67b6\u6784\u8bbe\u8ba1\u3001\u4ee3\u7801\u751f\u6210\u3001\u6d4b\u8bd5\u7f16\u5199\u3001CI\/CD\u89e6\u53d1\u3001\u90e8\u7f72\u56de\u6eda\u3001\u8fd0\u7ef4\u76d1\u63a7\u2014\u2014\u5168\u94fe\u8defAI\u5316\u3002<\/p>\n

\u542c\u8d77\u6765\u50cf\u79d1\u5e7b？\u4e09\u5e74\u524d\u6709\u4eba\u8bf4"AI\u80fd\u5199\u4ee3\u7801"，\u4f60\u4e5f\u89c9\u5f97\u662f\u79d1\u5e7b\u3002<\/p>\n

\u4eca\u5929，\u6211\u4eec\u6765\u62c6\u4e00\u62c6\u8fd9\u4e2a"\u79d1\u5e7b"\u7684\u67b6\u6784\u3002<\/p>\n

\u5148\u770b\u4e00\u5f20\u56fe，\u8ba9\u4f60\u4e24\u5206\u949f\u5185\u7406\u89e3AI DevOps\u5e73\u53f0\u7684\u5168\u5c40：<\/p>\n

flowchart LR
\n subgraph \u9700\u6c42["📋 \u9700\u6c42\u9636\u6bb5"]
\n A1["\u7528\u6237\u63cf\u8ff0<br\/>\u6a21\u7cca\u9700\u6c42"] –> A2["🤖 AI\u7406\u89e3+\u62c6\u89e3"]
\n A2 –> A3["\u7528\u6237\u6545\u4e8b + \u9a8c\u6536\u6807\u51c6"]
\n end<\/p>\n

subgraph \u8bbe\u8ba1["🎨 \u8bbe\u8ba1\u9636\u6bb5"]
\n B1["\u7528\u6237\u6545\u4e8b"] –> B2["🤖 AI\u751f\u6210\u67b6\u6784\u65b9\u6848"]
\n B2 –> B3["\u63a5\u53e3\u5b9a\u4e49 + \u6570\u636e\u6a21\u578b"]
\n end<\/p>\n

subgraph \u5f00\u53d1["💻 \u5f00\u53d1\u9636\u6bb5"]
\n C1["\u8bbe\u8ba1\u65b9\u6848"] –> C2["🤖 AI\u4ee3\u7801\u751f\u6210<br\/>Copilot \/ FIM"]
\n C2 –> C3["\u4ee3\u7801\u8bc4\u5ba1"]
\n end<\/p>\n

subgraph \u6d4b\u8bd5["🧪 \u6d4b\u8bd5\u9636\u6bb5"]
\n D1["\u4ee3\u7801+\u9700\u6c42"] –> D2["🤖 AI\u751f\u6210\u6d4b\u8bd5\u7528\u4f8b"]
\n D2 –> D3["\u81ea\u52a8\u6267\u884c+\u8986\u76d6\u7387"]
\n end<\/p>\n

subgraph \u90e8\u7f72["🚀 \u90e8\u7f72\u9636\u6bb5"]
\n E1["\u6d4b\u8bd5\u901a\u8fc7"] –> E2["🤖 AI\u89e6\u53d1CI\/CD"]
\n E2 –> E3["\u7070\u5ea6\u53d1\u5e03 + \u81ea\u52a8\u56de\u6eda"]
\n end<\/p>\n

subgraph \u8fd0\u7ef4["🔧 \u8fd0\u7ef4\u9636\u6bb5"]
\n F1["\u7ebf\u4e0a\u8fd0\u884c"] –> F2["🤖 AI\u5f02\u5e38\u68c0\u6d4b"]
\n F2 –> F3["\u81ea\u52a8\u8bca\u65ad + \u4fee\u590d\u5efa\u8bae"]
\n end<\/p>\n

\u9700\u6c42 –> \u8bbe\u8ba1 –> \u5f00\u53d1 –> \u6d4b\u8bd5 –> \u90e8\u7f72 –> \u8fd0\u7ef4
\n \u8fd0\u7ef4 -.->|"🔄 \u53cd\u9988\u95ed\u73af"| \u9700\u6c42<\/p>\n

style A2 fill:#4a90d9,color:#fff
\n style B2 fill:#4a90d9,color:#fff
\n style C2 fill:#4a90d9,color:#fff
\n style D2 fill:#4a90d9,color:#fff
\n style E2 fill:#4a90d9,color:#fff
\n style F2 fill:#4a90d9,color:#fff<\/p>\n

\u770b\u5230\u4e86\u5417？\u516d\u4e2a\u9636\u6bb5，\u516d\u4e2aAI\u4ecb\u5165\u70b9，\u4e00\u4e2a\u53cd\u9988\u95ed\u73af\u3002 \u8fd9\u4e0d\u662f\u79d1\u5e7b\u7247\u5206\u955c，\u8fd9\u662f2025\u5e74\u53ef\u4ee5\u843d\u5730\u7684\u5de5\u7a0b\u65b9\u6848\u3002<\/p>\n

\n

\u4e8c\u3001\u4e00\u5f20\u56fe\u770b\u7a7fAI DevOps\u5168\u666f<\/h3>\n

\u4f20\u7edfDevOps\u7684\u75db\u70b9，\u5708\u5185\u4eba\u61c2\u5f97\u90fd\u61c2：<\/p>\n

\u201c\u9700\u6c42\u6587\u6863\u5199\u5b8c\u4e86，\u5f00\u53d1\u8bf4\u770b\u4e0d\u61c2\u3002\u5f00\u53d1\u5199\u5b8c\u4e86，\u6d4b\u8bd5\u8bf4\u6d4b\u4e0d\u52a8\u3002\u6d4b\u8bd5\u6d4b\u5b8c\u4e86，\u8fd0\u7ef4\u8bf4\u4e0d\u6562\u4e0a\u3002\u8fd0\u7ef4\u4e0a\u4e86，\u7528\u6237\u8bf4\u4e0d\u662f\u6211\u60f3\u8981\u7684\u3002\u201d<\/p>\n

\u8fd9\u4e2a\u6b7b\u5faa\u73af\u7684\u672c\u8d28\u662f\u4ec0\u4e48？\u4fe1\u606f\u5728\u4e0d\u540c\u9636\u6bb5\u4f20\u9012\u65f6\u88ab\u524a\u5e73\u4e86\u3002<\/p>\n

AI DevOps\u7684\u6838\u5fc3\u903b\u8f91，\u5c31\u662f\u7528LLM（\u5927\u8bed\u8a00\u6a21\u578b）\u5145\u5f53"\u5168\u6d41\u7a0b\u7ffb\u8bd1\u5b98"\u2014\u2014\u5b83\u628a\u6a21\u7cca\u9700\u6c42\u7ffb\u8bd1\u6210\u7ed3\u6784\u5316\u7528\u6237\u6545\u4e8b，\u628a\u7528\u6237\u6545\u4e8b\u7ffb\u8bd1\u6210\u4ee3\u7801，\u628a\u4ee3\u7801\u7ffb\u8bd1\u6210\u6d4b\u8bd5\u7528\u4f8b，\u628a\u5f02\u5e38\u65e5\u5fd7\u7ffb\u8bd1\u6210\u4fee\u590d\u5efa\u8bae\u3002<\/p>\n

\u522b\u518d\u8bf4\u4ec0\u4e48"\u4eba\u4eba\u90fd\u662f\u4ea7\u54c1\u7ecf\u7406"\u4e86，AI DevOps\u7684\u7406\u60f3\u662f"\u4eba\u4eba\u90fd\u662f\u5168\u6808\u5de5\u7a0b\u5e08"\u3002<\/p>\n

\u4e0b\u9762\u8fd9\u5f20\u56fe\u5c55\u793aAI\u4ecb\u5165\u7684\u5177\u4f53\u65b9\u5f0f：<\/p>\n

flowchart TB
\n subgraph \u8f93\u5165\u5c42["📥 \u8f93\u5165\u5c42"]
\n I1["\u81ea\u7136\u8bed\u8a00\u9700\u6c42"]
\n I2["\u5386\u53f2\u4ee3\u7801\u4ed3\u5e93"]
\n I3["\u8fd0\u7ef4\u65e5\u5fd7\/\u76d1\u63a7\u6570\u636e"]
\n I4["\u56e2\u961f\u77e5\u8bc6\u6587\u6863"]
\n end<\/p>\n

subgraph AI\u5f15\u64ce["🧠 AI\u7f16\u6392\u5f15\u64ce（\u6838\u5fc3）"]
\n E1["\u591a\u6a21\u578b\u8def\u7531<br\/>GPT-4o \/ Claude \/ \u56fd\u4ea7\u6a21\u578b"]
\n E2["Prompt\u6a21\u677f\u7ba1\u7406<br\/>\u9700\u6c42\u2192\u8bbe\u8ba1\u2192\u4ee3\u7801\u2192\u6d4b\u8bd5\u2192\u8fd0\u7ef4"]
\n E3["\u4e0a\u4e0b\u6587\u7a97\u53e3\u7ba1\u7406<br\/>\u957f\u6587\u672c\u5206\u7247+\u6458\u8981"]
\n end<\/p>\n

subgraph \u77e5\u8bc6\u5e93["📚 \u4f01\u4e1a\u77e5\u8bc6\u5e93"]
\n K1["\u5411\u91cf\u6570\u636e\u5e93<br\/>Chromadb \/ Milvus"]
\n K2["\u56fe\u6570\u636e\u5e93<br\/>Neo4j \u5b58\u9879\u76ee\u4f9d\u8d56\u5173\u7cfb"]
\n K3["\u89c4\u5219\u5f15\u64ce<br\/>\u5b89\u5168\u5408\u89c4\u9ed1\u767d\u540d\u5355"]
\n end<\/p>\n

subgraph \u8f93\u51fa\u5c42["📤 \u8f93\u51fa\u5c42"]
\n O1["\u7528\u6237\u6545\u4e8b\u6587\u6863"]
\n O2["\u4ee3\u7801 + Code Review"]
\n O3["\u6d4b\u8bd5\u7528\u4f8b + \u62a5\u544a"]
\n O4["CI\/CD Pipeline YAML"]
\n O5["\u544a\u8b66\u8bca\u65ad + \u4fee\u590dPR"]
\n end<\/p>\n

subgraph \u53cd\u9988\u95ed\u73af["🔄 \u53cd\u9988\u95ed\u73af"]
\n FB["\u4eba\u5de5\u8bc4\u5ba1\/\u7ebf\u4e0a\u6307\u6807<br\/>\u2192 \u56de\u6d41\u5230\u77e5\u8bc6\u5e93<br\/>\u2192 \u4f18\u5316Prompt\u6a21\u677f"]
\n end<\/p>\n

\u8f93\u5165\u5c42 –> AI\u5f15\u64ce
\n AI\u5f15\u64ce <–> \u77e5\u8bc6\u5e93
\n AI\u5f15\u64ce –> \u8f93\u51fa\u5c42
\n \u8f93\u51fa\u5c42 –> \u53cd\u9988\u95ed\u73af
\n \u53cd\u9988\u95ed\u73af -.-> \u77e5\u8bc6\u5e93<\/p>\n

style AI\u5f15\u64ce fill:#e74c3c,color:#fff,stroke:#c0392b,stroke-width:3px
\n style \u77e5\u8bc6\u5e93 fill:#27ae60,color:#fff
\n style \u53cd\u9988\u95ed\u73af fill:#f39c12,color:#fff<\/p>\n

\u4e09\u4e2a\u5173\u952e\u8bcd：AI\u7f16\u6392\u5f15\u64ce\u662f\u5927\u8111，\u77e5\u8bc6\u5e93\u662f\u8bb0\u5fc6，\u53cd\u9988\u95ed\u73af\u662f\u8fdb\u5316\u673a\u5236\u3002\u540e\u9762\u9010\u4e2a\u62c6\u89e3\u3002<\/p>\n

\n

\u4e09\u3001\u5404\u73af\u8282AI\u5207\u5165\u5b9e\u6218<\/h3>\n

3.1 \u9700\u6c42\u2192\u7528\u6237\u6545\u4e8b：\u522b\u8ba9\u4ea7\u54c1\u7ecf\u7406\u51993\u5929PRD\u4e86<\/h4>\n

\u4f20\u7edf\u65b9\u5f0f：\u4ea7\u54c1\u7ecf\u7406\u5199PRD \u2192 \u5f00\u8bc4\u5ba1\u4f1a \u2192 \u5f00\u53d1\u8bf4"\u8fd9\u91cc\u4e0d\u6e05\u6670" \u2192 \u6539PRD \u2192 \u518d\u5ba1 \u2192 3\u5929\u8fc7\u53bb\u4e86\u3002<\/p>\n

AI\u7684\u65b9\u5f0f：<\/p>\n

# AI\u9700\u6c42\u5206\u6790\u670d\u52a1 \u26a0\ufe0f \u6838\u5fc3\u4f2a\u4ee3\u7801，\u5c55\u793a\u8bbe\u8ba1\u601d\u8def
\nfrom openai import OpenAI<\/p>\n

def analyze_requirement(raw_text: str) -> dict:
\n """\u5c06\u6a21\u7cca\u9700\u6c42\u8f6c\u5316\u4e3a\u7ed3\u6784\u5316\u7528\u6237\u6545\u4e8b"""
\n client = OpenAI()<\/p>\n

system_prompt = """\u4f60\u662f\u4e00\u4e2a\u8d44\u6df1\u9700\u6c42\u5206\u6790\u5e08\u3002\u5c06\u7528\u6237\u8f93\u5165\u8f6c\u5316\u4e3a\u6807\u51c6\u7528\u6237\u6545\u4e8b：
\n \u683c\u5f0f：\u4f5c\u4e3a<\u89d2\u8272>，\u6211\u60f3\u8981<\u529f\u80fd>，\u4ee5\u4fbf<\u4ef7\u503c>
\n \u5e76\u7ed9\u51fa\u9a8c\u6536\u6807\u51c6（\u81f3\u5c113\u6761Given-When-Then）
\n \u8f93\u51faJSON\u683c\u5f0f\u3002"""<\/p>\n

response = client.chat.completions.create(
\n model="gpt-4o",
\n messages=[
\n {"role": "system", "content": system_prompt},
\n {"role": "user", "content": f"\u9700\u6c42\u63cf\u8ff0：{raw_text}"}
\n ],
\n response_format={"type": "json_object"}
\n )
\n return response.choices[0].message.content<\/p>\n

# \u793a\u4f8b\u8f93\u5165：\u300c\u7528\u6237\u767b\u5f55\u540e\u80fd\u770b\u5230\u81ea\u5df1\u7684\u8ba2\u5355\u5217\u8868，\u652f\u6301\u6309\u65f6\u95f4\u7b5b\u9009\u300d
\n# AI\u8f93\u51fa\u2192
\n# {
\n# "user_story": "\u4f5c\u4e3a\u6ce8\u518c\u7528\u6237，\u6211\u60f3\u8981\u67e5\u770b\u6211\u7684\u8ba2\u5355\u5217\u8868，\u4ee5\u4fbf\u8ffd\u8e2a\u8d2d\u4e70\u5386\u53f2",
\n# "acceptance_criteria": [
\n# "Given \u7528\u6237\u5df2\u767b\u5f55 When \u8fdb\u5165\u8ba2\u5355\u9875\u9762 Then \u663e\u793a\u8fd130\u5929\u8ba2\u5355",
\n# "Given \u8ba2\u5355\u5217\u8868\u9875 When \u9009\u62e9\u65e5\u671f\u8303\u56f4 Then \u6309\u7b5b\u9009\u6761\u4ef6\u5c55\u793a",
\n# "Given \u65e0\u8ba2\u5355\u7528\u6237 When \u8fdb\u5165\u8ba2\u5355\u9875 Then \u663e\u793a\u7a7a\u72b6\u6001\u63d0\u793a"
\n# ]
\n# }<\/p>\n

\u4e00\u53e5\u8bdd\u603b\u7ed3：\u4ee5\u524dPM\u51993\u5929，\u73b0\u5728AI\u51fa\u521d\u7a3f3\u5206\u949f，PM\u53ea\u9700\u82b130\u5206\u949f\u5fae\u8c03\u3002\u4e0d\u662f\u53d6\u4ee3PM，\u662f\u8ba9PM\u4ece"\u5199\u6587\u6863"\u53d8\u6210"\u5ba1\u6587\u6863"\u3002<\/p>\n

🤣 \u5e7d\u9ed8\u65f6\u523b：\u8001\u677f\u542c\u8bf4AI\u80fd\u5199\u9700\u6c42\u6587\u6863，\u5174\u594b\u5730\u8bf4"\u90a3\u6211\u4eec\u88c1\u6389\u4ea7\u54c1\u7ecf\u7406\u5427"，\u7136\u540e\u53d1\u73b0\u2014\u2014AI\u5199\u7684\u9700\u6c42\u66f4\u79bb\u8c31，\u56e0\u4e3a\u6ca1\u4eba\u544a\u8bc9AI"\u8fd9\u4e2a\u529f\u80fd\u8001\u677f\u4e0d\u559c\u6b22"\u3002\u4ea7\u54c1\u7ecf\u7406\u7684\u4ef7\u503c\u4e0d\u5728"\u5199"，\u5728"\u7ffb\u8bd1"\u2014\u2014\u628a\u8001\u677f\u7684"\u6211\u60f3\u8981\u4e00\u4e2a\u50cf\u6dd8\u5b9d\u90a3\u6837\u7684\u7cfb\u7edf"\u7ffb\u8bd1\u6210AI\u80fd\u7406\u89e3\u7684"\u4f60\u8981\u7684\u662f\u4e00\u4e2a\u7535\u5546\u5e73\u53f0，\u540e\u7aef\u7528\u5fae\u670d\u52a1，\u524d\u7aef\u7528React"\u3002<\/p>\n

3.2 \u4ee3\u7801\u751f\u6210：Copilot\u53ea\u662f\u524d\u83dc<\/h4>\n

GitHub Copilot\u505a\u7684\u662f\u884c\u7ea7\u8865\u5168\u2014\u2014\u4f60\u5199def calculate_，\u5b83\u5e2e\u4f60\u8865tax(price)\u3002<\/p>\n

\u771f\u6b63\u7684AI DevOps\u8981\u505a\u529f\u80fd\u7ea7\u751f\u6210\u2014\u2014\u4f60\u7ed9\u4e00\u4e2a\u7528\u6237\u6545\u4e8b，\u5b83\u8f93\u51fa\u5b8c\u6574\u4ee3\u7801+\u5355\u6d4b+Dockerfile\u3002<\/p>\n

# \u26a0\ufe0f AI\u751f\u6210\u7684CI\/CD\u914d\u7f6e（GitLab CI\u793a\u4f8b）
\n# \u7531LLM\u6839\u636e\u9879\u76ee\u7ed3\u6784\u81ea\u52a8\u751f\u6210
\nstages:
\n – lint
\n – test
\n – build
\n – deploy<\/p>\n

variables:
\n DOCKER_IMAGE: $CI_REGISTRY_IMAGE:$CI_COMMIT_SHORT_SHA<\/p>\n

lint:
\n stage: lint
\n image: python:3.11
\n script:
\n – pip install ruff
\n – ruff check .
\n only:
\n – merge_requests<\/p>\n

unit-test:
\n stage: test
\n image: python:3.11
\n script:
\n – pip install pytest pytest-cov
\n – pytest –cov=.\/ –cov-report=xml
\n artifacts:
\n reports:
\n coverage_report:
\n coverage_format: cobertura
\n path: coverage.xml<\/p>\n

build:
\n stage: build
\n image: docker:24
\n services:
\n – docker:24-dind
\n script:
\n – docker build -t $DOCKER_IMAGE .
\n – docker push $DOCKER_IMAGE<\/p>\n

deploy-staging:
\n stage: deploy
\n script:
\n – kubectl set image deployment\/app app=$DOCKER_IMAGE -n staging
\n environment:
\n name: staging
\n only:
\n – main<\/p>\n

🤣 \u5e7d\u9ed8\u65f6\u523b：AI\u751f\u6210\u7684CI\u914d\u7f6e\u7b2c\u4e00\u6b21\u8dd1\u5c31\u901a\u8fc7\u4e86\u2014\u2014\u5168\u7ec4\u6c89\u9ed8\u4e8630\u79d2\u3002\u4e0d\u662f\u56e0\u4e3a\u6fc0\u52a8，\u662f\u56e0\u4e3a\u4e0d\u6562\u76f8\u4fe1\u3002\u7136\u540eleader\u8bf4"\u91cd\u65b0\u5199\u4e00\u904d\u5427，\u4e07\u4e00\u6709bug\u5462"\u3002\u4eba\u7c7b\u5bf9AI\u6700\u5927\u7684\u4e0d\u4fe1\u4efb，\u4e0d\u662f\u5b83\u4e0d\u884c，\u662f\u5b83\u592a\u5feb\u4e86\u3002<\/p>\n

3.3 \u6d4b\u8bd5：AI\u5199\u6d4b\u8bd5\u7528\u4f8b\u6bd4\u4f60\u5199\u4ee3\u7801\u8fd8\u5feb<\/h4>\n

\u8fd9\u662fAI DevOps\u6700"\u9999"\u7684\u73af\u8282\u2014\u2014\u8ba9AI\u8bfb\u4f60\u7684\u4ee3\u7801，\u7136\u540e\u81ea\u52a8\u751f\u6210\u6d4b\u8bd5\u7528\u4f8b\u3002<\/p>\n

# \u26a0\ufe0f AI\u81ea\u52a8\u751f\u6210\u7684\u6d4b\u8bd5\u7528\u4f8b（\u57fa\u4e8e\u4e0a\u9762\u7684\u8ba2\u5355\u670d\u52a1）
\nimport pytest
\nfrom app.services.order_service import OrderService<\/p>\n

class TestOrderService:
\n """AI\u6839\u636e\u63a5\u53e3\u7b7e\u540d+\u6587\u6863\u81ea\u52a8\u751f\u6210"""<\/p>\n

def test_list_orders_authenticated_user(self, auth_client):
\n """Given \u5df2\u767b\u5f55\u7528\u6237 When \u8bf7\u6c42\u8ba2\u5355\u5217\u8868 Then \u8fd4\u56de200+\u8ba2\u5355\u6570\u636e"""
\n response = auth_client.get("\/api\/orders")
\n assert response.status_code == 200
\n assert "orders" in response.json()<\/p>\n

def test_list_orders_unauthenticated_returns_401(self, client):
\n """Given \u672a\u767b\u5f55\u7528\u6237 When \u8bf7\u6c42\u8ba2\u5355\u5217\u8868 Then \u8fd4\u56de401"""
\n response = client.get("\/api\/orders")
\n assert response.status_code == 401<\/p>\n

def test_list_orders_with_date_filter(self, auth_client):
\n """Given \u5df2\u767b\u5f55 When \u4f20start_date\/end_date Then \u6309\u65f6\u95f4\u8303\u56f4\u8fd4\u56de"""
\n response = auth_client.get(
\n "\/api\/orders?start_date=2025-01-01&end_date=2025-06-30"
\n )
\n assert response.status_code == 200
\n orders = response.json()["orders"]
\n for order in orders:
\n assert "2025-01-01" <= order["created_at"] <= "2025-06-30"<\/p>\n

@pytest.mark.parametrize("invalid_date", [
\n "not-a-date", "2025-13-01", "2025\/01\/01", ""
\n ])
\n def test_date_filter_rejects_invalid_format(self, auth_client, invalid_date):
\n """Given \u5df2\u767b\u5f55 When \u4f20\u975e\u6cd5\u65e5\u671f Then \u8fd4\u56de400"""
\n response = auth_client.get(f"\/api\/orders?start_date={invalid_date}")
\n assert response.status_code == 400<\/p>\n

AI\u751f\u6210\u7684\u6d4b\u8bd5\u6709\u4e09\u4e2a\u597d\u5904：<\/p>\n

  • \u8fb9\u754c\u6761\u4ef6\u5168\u8986\u76d6：\u6b63\u5e38\u4eba\u53ea\u4f1a\u5199"\u6b63\u5e38\u60c5\u51b5"，AI\u4f1a\u5199"\u4f2013\u6708\u4f1a\u600e\u6837"<\/li>\n
  • \u8986\u76d6\u7387\u4e0d\u9020\u5047：AI\u751f\u6210\u7684\u6d4b\u8bd5\u662f\u771f\u6b63\u5728\u6d4b\u903b\u8f91，\u4e0d\u662fassert True<\/li>\n
  • \u4f60\u6562\u91cd\u6784\u4e86：\u6709\u4e86\u5168\u91cf\u6d4b\u8bd5，\u6539\u4ee3\u7801\u4e0d\u7528\u63d0\u5fc3\u540a\u80c6<\/li>\n

    🤣 \u5e7d\u9ed8\u65f6\u523b：\u6709\u4e00\u6b21AI\u7ed9\u4e00\u4e2a\u51fd\u6570\u5199\u4e8620\u4e2a\u6d4b\u8bd5\u7528\u4f8b，\u5f00\u53d1\u770b\u4e86\u8bf4"\u8fd9\u51fd\u6570\u603b\u5171\u624d8\u884c，\u4f60\u5199\u4e8620\u4e2a\u6d4b\u8bd5\u662f\u4e0d\u662f\u8fc7\u4e86？" AI\u8bf4"\u6211\u770b\u4f60\u540c\u4e8b\u6539\u8fd9\u51fd\u6570\u6539\u4e866\u6b21，\u6bcf\u6b21\u90fd\u5f15\u5165\u65b0bug"\u3002\u6709\u65f6\u5019，AI\u6bd4\u4f60\u66f4\u4e86\u89e3\u4f60\u7684\u56e2\u961f\u3002<\/p>\n

    3.4 \u90e8\u7f72\u4e0e\u8fd0\u7ef4：AI\u5e2e\u4f60"\u770b\u5bb6"<\/h4>\n

    \u90e8\u7f72\u73af\u8282，AI\u7684\u5207\u5165\u70b9\u662f\u667a\u80fd\u7070\u5ea6\u7b56\u7565\u548c\u81ea\u52a8\u56de\u6eda\u51b3\u7b56\u3002<\/p>\n

    \u8fd0\u7ef4\u73af\u8282，AI\u7684\u6838\u5fc3\u4ef7\u503c\u662f\u5f02\u5e38\u68c0\u6d4b+\u6839\u56e0\u5206\u6790+\u4fee\u590d\u5efa\u8bae\u3002<\/p>\n

    # \u26a0\ufe0f AI\u8fd0\u7ef4Agent\u6838\u5fc3\u903b\u8f91（\u4f2a\u4ee3\u7801）
    \ndef ai_ops_agent(alert: dict) -> dict:
    \n """\u63a5\u6536\u544a\u8b66，\u8fd4\u56de\u5206\u6790\u7ed3\u679c"""<\/p>\n

    # 1. \u4ece\u77e5\u8bc6\u5e93\u68c0\u7d22\u76f8\u4f3c\u5386\u53f2\u544a\u8b66
    \n similar_cases = vector_db.search(
    \n query=alert["message"],
    \n collection="historical_incidents",
    \n top_k=3
    \n )<\/p>\n

    # 2. \u62c9\u53d6\u76f8\u5173\u65e5\u5fd7
    \n logs = log_fetcher.get_logs(
    \n service=alert["service"],
    \n time_range=(-5, +1), # \u544a\u8b66\u524d5\u5206\u949f\u5230\u540e1\u5206\u949f
    \n )<\/p>\n

    # 3. \u8ba9LLM\u5206\u6790
    \n prompt = f"""
    \n \u544a\u8b66：{alert}
    \n \u76f8\u4f3c\u5386\u53f2\u6848\u4f8b：{similar_cases}
    \n \u76f8\u5173\u65e5\u5fd7：{logs[:5000]}<\/p>\n

    \u8bf7\u5206\u6790：
    \n 1. \u6839\u56e0\u662f\u4ec0\u4e48？
    \n 2. \u5f71\u54cd\u8303\u56f4\u591a\u5927？
    \n 3. \u5efa\u8bae\u7684\u4fee\u590d\u6b65\u9aa4？
    \n 4. \u662f\u5426\u9700\u8981\u81ea\u52a8\u6267\u884c？（\u662f\/\u5426）
    \n """<\/p>\n

    analysis = llm.call(prompt)<\/p>\n

    # 4. \u5982\u679c\u7f6e\u4fe1\u5ea6\u9ad8，\u81ea\u52a8\u5efa\u4fee\u590dPR
    \n if analysis["auto_fix_confidence"] > 0.85:
    \n auto_create_fix_pr(analysis["fix_suggestion"])<\/p>\n

    return analysis<\/p>\n

    \u91cd\u70b9\u6765\u4e86：AI\u8fd0\u7ef4\u4e0d\u662f\u66ff\u4ee3on-call，\u662f\u8ba9on-call\u4ece"\u51cc\u66683\u70b9\u88ab\u53eb\u9192\u67e5\u65e5\u5fd7"\u53d8\u6210"\u65e9\u4e0a9\u70b9\u770bAI\u7684\u62a5\u544a\u51b3\u5b9a\u8981\u4e0d\u8981\u70b9Merge"\u3002<\/p>\n

    \n

    \u56db\u3001LLM\u7f16\u6392\u5f15\u64ce\u2014\u2014\u5e73\u53f0\u7684"\u5927\u8111"<\/h3>\n

    \u5982\u679c\u4f60\u4ee5\u4e3aAI DevOps\u5c31\u662f"\u628aGPT\u7684API\u63a5\u5230Jenkins\u4e0a"，\u90a3\u4f60\u4f4e\u4f30\u4e86\u8fd9\u4e2a\u7cfb\u7edf90%\u7684\u590d\u6742\u5ea6\u3002<\/p>\n

    \u771f\u6b63\u7684\u96be\u70b9\u5728\u7f16\u6392\u5c42\u3002 \u7b54\u6848\u4e0d\u662f\u4e00\u4e2a\u6a21\u578b\u3001\u4e00\u4e2aPrompt\u80fd\u641e\u5b9a\u7684\u3002\u9700\u6c42\u5206\u6790\u7528Claude\u6548\u679c\u597d，\u4ee3\u7801\u751f\u6210\u7528GPT-4o\u5feb，\u4ee3\u7801\u5ba1\u67e5\u7528DeepSeek\u4fbf\u5b9c\u2014\u2014\u8fd9\u5c31\u9700\u8981\u4e00\u4e2a\u591a\u6a21\u578b\u8def\u7531\u5c42\u3002<\/p>\n

    # \u26a0\ufe0f LLM\u7f16\u6392\u5f15\u64ce\u6838\u5fc3\u2014\u2014\u591a\u6a21\u578b\u8def\u7531\u5668
    \nfrom enum import Enum
    \nfrom dataclasses import dataclass<\/p>\n

    class TaskType(Enum):
    \n REQUIREMENT_ANALYSIS = "requirement" # \u9700\u6c42\u5206\u6790 \u2192 Claude 3.5 Sonnet
    \n CODE_GENERATION = "code_gen" # \u4ee3\u7801\u751f\u6210 \u2192 GPT-4o
    \n CODE_REVIEW = "code_review" # \u4ee3\u7801\u5ba1\u67e5 \u2192 DeepSeek-V3
    \n TEST_GENERATION = "test_gen" # \u6d4b\u8bd5\u751f\u6210 \u2192 GPT-4o-mini
    \n DEPLOY_SCRIPT = "deploy" # \u90e8\u7f72\u811a\u672c \u2192 GPT-4o
    \n OPS_ANALYSIS = "ops" # \u8fd0\u7ef4\u5206\u6790 \u2192 Claude 3.5 Sonnet<\/p>\n

    @dataclass
    \nclass ModelConfig:
    \n model: str
    \n cost_per_1k: float # \u6bcf\u5343token\u6210\u672c（\u7f8e\u5143）
    \n context_window: int
    \n strengths: list[str]<\/p>\n

    # \u591a\u6a21\u578b\u914d\u7f6e（\u6210\u672c\u654f\u611f\u578b\u7b56\u7565）
    \nMODEL_REGISTRY = {
    \n TaskType.REQUIREMENT_ANALYSIS: ModelConfig(
    \n model="claude-3-5-sonnet",
    \n cost_per_1k=0.003,
    \n context_window=200000,
    \n strengths=["\u7ed3\u6784\u5316\u8f93\u51fa", "\u9700\u6c42\u7406\u89e3"]
    \n ),
    \n TaskType.CODE_GENERATION: ModelConfig(
    \n model="gpt-4o",
    \n cost_per_1k=0.0025,
    \n context_window=128000,
    \n strengths=["\u4ee3\u7801\u751f\u6210", "\u51fd\u6570\u8865\u5168"]
    \n ),
    \n TaskType.CODE_REVIEW: ModelConfig(
    \n model="deepseek-chat",
    \n cost_per_1k=0.00014, # 👈 \u4fbf\u5b9c18\u500d！
    \n context_window=128000,
    \n strengths=["\u4ee3\u7801\u5206\u6790", "\u9ad8\u6027\u4ef7\u6bd4"]
    \n ),
    \n TaskType.TEST_GENERATION: ModelConfig(
    \n model="gpt-4o-mini",
    \n cost_per_1k=0.00015,
    \n context_window=128000,
    \n strengths=["\u6279\u91cf\u751f\u6210", "\u683c\u5f0f\u89c4\u8303"]
    \n ),
    \n}<\/p>\n

    class LLMRouter:
    \n """\u591a\u6a21\u578b\u8def\u7531\u5668\u2014\u2014\u6309\u4efb\u52a1\u7c7b\u578b+\u6210\u672c+\u8d28\u91cf\u81ea\u52a8\u8def\u7531"""<\/p>\n

    def route(self, task_type: TaskType) -> ModelConfig:
    \n return MODEL_REGISTRY.get(task_type, MODEL_REGISTRY[TaskType.CODE_GENERATION])<\/p>\n

    def call(self, task_type: TaskType, prompt: str, **kwargs) -> str:
    \n config = self.route(task_type)
    \n # \u5b9e\u9645\u8c03\u7528\u5bf9\u5e94\u6a21\u578b\u7684API（\u7b80\u5316）
    \n return self._invoke_model(config.model, prompt, **kwargs)<\/p>\n

    🤣 \u5e7d\u9ed8\u65f6\u523b：\u6709\u4eba\u95ee"\u4e3a\u4ec0\u4e48\u4e0d\u6240\u6709\u4efb\u52a1\u90fd\u7528GPT-4o？" \u7b54：\u201c\u56e0\u4e3a\u8d35\u554a\u5144\u5f1f\u3002\u4ee3\u7801\u5ba1\u67e5\u8fd9\u79cd\u5929\u5929\u8dd1\u51e0\u5341\u6b21\u7684\u6d3b，\u7528DeepSeek\u4e00\u5929\u82b13\u6bdb，\u7528GPT-4o\u4e00\u5929\u82b130\u5757\u2014\u2014\u4e00\u4e2a\u6708\u5dee900\u5757，\u591f\u4e70\u4e00\u5e74\u7684GitHub Copilot\u4e86\u3002\u505a\u4eba\u8981\u4f1a\u8fc7\u65e5\u5b50\u3002\u201d<\/p>\n

    4.1 Prompt\u6a21\u677f\u7ba1\u7406\u2014\u2014\u522b\u6bcf\u6b21\u90fd\u624b\u6413Prompt<\/h4>\n

    \u7f16\u6392\u5f15\u64ce\u7684\u7b2c\u4e8c\u4e2a\u6838\u5fc3\u662fPrompt\u6a21\u677f\u4f53\u7cfb\u3002\u4f60\u4e0d\u80fd\u6bcf\u6b21\u751f\u6210\u6d4b\u8bd5\u7528\u4f8b\u90fd\u91cd\u65b0\u5199\u4e00\u904dPrompt\u2014\u2014\u90a3\u548c\u624b\u5199\u6d4b\u8bd5\u7528\u4f8b\u6709\u4ec0\u4e48\u533a\u522b？<\/p>\n

    # \u26a0\ufe0f Prompt\u6a21\u677f\u7ba1\u7406（YAML\u914d\u7f6e）
    \ntemplates:
    \n code_review:
    \n system: |
    \n \u4f60\u662f\u4e00\u4e2a\u8d44\u6df1\u4ee3\u7801\u5ba1\u67e5\u4e13\u5bb6\u3002\u5ba1\u67e5\u4ee5\u4e0b\u4ee3\u7801，\u5173\u6ce8：
    \n 1. \u5b89\u5168\u6f0f\u6d1e（SQL\u6ce8\u5165\u3001XSS\u3001\u786c\u7f16\u7801\u5bc6\u94a5）
    \n 2. \u6027\u80fd\u95ee\u9898（N+1\u67e5\u8be2\u3001\u4e0d\u5fc5\u8981\u5faa\u73af）
    \n 3. \u53ef\u7ef4\u62a4\u6027（\u547d\u540d\u3001\u6ce8\u91ca\u3001\u51fd\u6570\u957f\u5ea6）
    \n 4. \u26a0\ufe0f \u82e5\u6d89\u53ca\u654f\u611f\u4fe1\u606f（API Key\u3001\u5bc6\u7801），\u6807\u8bb0\u4e3a\u3010\u5b89\u5168\u98ce\u9669\u3011
    \n \u8f93\u51faMarkdown\u683c\u5f0f，\u6309\u4e25\u91cd\u7a0b\u5ea6\u6392\u5e8f\u3002
    \n user_template: |
    \n \u9879\u76ee\u8bed\u8a00：{language}
    \n \u6539\u52a8\u6587\u4ef6：{files}
    \n diff\u5185\u5bb9：
    \n {diff_content}
    \n temperature: 0.3 # \u4ee3\u7801\u5ba1\u67e5\u8981\u4e25\u8c28，\u4e0d\u8981\u521b\u610f<\/p>\n

    test_generation:
    \n system: |
    \n \u4f60\u662f\u4e00\u4e2a\u6d4b\u8bd5\u5de5\u7a0b\u5e08\u3002\u4e3a\u4ee5\u4e0b\u4ee3\u7801\u751f\u6210pytest\u6d4b\u8bd5：
    \n – \u6bcf\u4e2apublic\u51fd\u6570\u81f3\u5c113\u4e2a\u6d4b\u8bd5\u7528\u4f8b
    \n – \u5305\u542b\u6b63\u5e38\u8def\u5f84\u3001\u8fb9\u754c\u6761\u4ef6\u3001\u5f02\u5e38\u60c5\u51b5
    \n – \u4f7f\u7528@pytest.mark.parametrize\u51cf\u5c11\u91cd\u590d
    \n – \u4e0d\u8981mock\u6570\u636e\u5e93，\u4f7f\u7528test fixtures
    \n \u53ea\u8f93\u51fa\u4ee3\u7801，\u4e0d\u8981\u89e3\u91ca\u3002
    \n user_template: |
    \n \u6e90\u4ee3\u7801：
    \n {source_code}<\/p>\n

    \u5df2\u6709\u6d4b\u8bd5（\u907f\u514d\u91cd\u590d）：
    \n {existing_tests}
    \n temperature: 0.5 # \u6d4b\u8bd5\u751f\u6210\u9700\u8981\u4e00\u70b9\u521b\u9020\u529b<\/p>\n

    💡 \u6548\u7387\u6280\u5de7 \u2460：\u628a\u6240\u6709Prompt\u6a21\u677f\u5b58\u6210YAML\u6587\u4ef6，\u7528Jinja2\u6e32\u67d3\u3002\u8fd9\u6837\u6539Prompt\u4e0d\u7528\u6539\u4ee3\u7801，\u4ea7\u54c1\u7ecf\u7406\u90fd\u80fd\u8c03\u2014\u2014\u867d\u7136\u4ed6\u4eec\u8c03\u5b8c\u4e4b\u540eAI\u53d8\u5f97\u66f4\u4e0d\u4f1a\u5199\u4ee3\u7801\u4e86\u3002<\/p>\n

    \n

    \u4e94\u3001\u77e5\u8bc6\u5e93\u4f53\u7cfb：\u8ba9AI\u8bb0\u4f4f\u4f60\u7684\u9879\u76ee\u7ecf\u9a8c<\/h3>\n

    \u77e5\u8bc6\u5e93\u662fAI DevOps\u5e73\u53f0\u7684"\u957f\u671f\u8bb0\u5fc6"\u3002\u6ca1\u6709\u77e5\u8bc6\u5e93\u7684AI，\u6bcf\u6b21\u90fd\u662f\u4ece\u96f6\u5f00\u59cb\u3002<\/p>\n

    \u4e24\u5c42\u77e5\u8bc6\u5e93\u67b6\u6784：<\/p>\n

    flowchart LR
    \n subgraph \u5199\u5165["📥 \u77e5\u8bc6\u6c89\u6dc0"]
    \n W1["\u4ee3\u7801\u4ed3\u5e93<br\/>Git\u5386\u53f2"]
    \n W2["\u6545\u969c\u590d\u76d8<br\/>Postmortem"]
    \n W3["\u8bbe\u8ba1\u6587\u6863<br\/>ADR\/PRD"]
    \n W4["\u7ebf\u4e0a\u6307\u6807<br\/>APM\u6570\u636e"]
    \n end<\/p>\n

    subgraph \u5b58\u50a8["💾 \u53cc\u5c42\u5b58\u50a8"]
    \n S1["🔷 \u5411\u91cf\u6570\u636e\u5e93<br\/>\u8bed\u4e49\u641c\u7d22<br\/>Chromadb\/PGVector"]
    \n S2["🔶 \u56fe\u6570\u636e\u5e93<br\/>\u5173\u7cfb\u63a8\u7406<br\/>Neo4j \u5b58\u670d\u52a1\u4f9d\u8d56"]
    \n end<\/p>\n

    subgraph \u67e5\u8be2["🔍 \u68c0\u7d22\u589e\u5f3a"]
    \n Q1["\u9700\u6c42\u5206\u6790\u65f6<br\/>\u2192 \u641c\u76f8\u4f3c\u7528\u6237\u6545\u4e8b"]
    \n Q2["\u6545\u969c\u6392\u67e5\u65f6<br\/>\u2192 \u641c\u76f8\u4f3c\u5386\u53f2\u544a\u8b66"]
    \n Q3["\u4ee3\u7801\u751f\u6210\u65f6<br\/>\u2192 \u641c\u9879\u76ee\u73b0\u6709\u4ee3\u7801\u98ce\u683c"]
    \n Q4["\u67b6\u6784\u8bc4\u5ba1\u65f6<br\/>\u2192 \u641c\u670d\u52a1\u4f9d\u8d56\u5173\u7cfb"]
    \n end<\/p>\n

    \u5199\u5165 –> \u5b58\u50a8
    \n \u5b58\u50a8 –> \u67e5\u8be2<\/p>\n

    style S1 fill:#4a90d9,color:#fff
    \n style S2 fill:#27ae60,color:#fff<\/p>\n

    \u5411\u91cf\u5e93\u8d1f\u8d23"\u6a21\u7cca\u5339\u914d"：\u641c"\u7528\u6237\u767b\u5f55\u5931\u8d25"，\u80fd\u627e\u5230\u6240\u6709\u548c"\u8ba4\u8bc1"\u3001\u201csession\u201d\u3001"token\u8fc7\u671f"\u76f8\u5173\u7684\u5386\u53f2\u6545\u969c\u3002<\/p>\n

    \u56fe\u5e93\u8d1f\u8d23"\u5173\u7cfb\u63a8\u7406"：\u95ee"\u6539\u4e86\u8fd9\u4e2a\u5fae\u670d\u52a1\u7684API，\u4f1a\u5f71\u54cd\u54ea\u4e9b\u4e0b\u6e38\u670d\u52a1？"\u2014\u2014\u56fe\u6570\u636e\u5e93\u79d2\u51fa\u7b54\u6848\u3002<\/p>\n

    # \u26a0\ufe0f \u77e5\u8bc6\u5e93\u68c0\u7d22\u589e\u5f3a\u751f\u6210（RAG）\u6838\u5fc3\u4ee3\u7801
    \nfrom chromadb import PersistentClient
    \nfrom sentence_transformers import SentenceTransformer<\/p>\n

    class KnowledgeRetriever:
    \n """\u77e5\u8bc6\u5e93\u68c0\u7d22\u2014\u2014\u8ba9AI\u6709"\u8bb0\u5fc6""""<\/p>\n

    def __init__(self):
    \n self.embedder = SentenceTransformer("BAAI\/bge-large-zh-v1.5")
    \n self.chroma = PersistentClient(path=".\/knowledge_db")
    \n self.collection = self.chroma.get_or_create_collection("project_knowledge")<\/p>\n

    def index_incident(self, incident: dict):
    \n """\u5c06\u6545\u969c\u590d\u76d8\u6587\u6863\u5199\u5165\u5411\u91cf\u5e93"""
    \n text = f"\u6545\u969c：{incident['title']}\\\\n\u6839\u56e0：{incident['root_cause']}\\\\n\u4fee\u590d：{incident['fix']}"
    \n embedding = self.embedder.encode(text).tolist()
    \n self.collection.add(
    \n documents=[text],
    \n embeddings=[embedding],
    \n metadatas=[{"date": incident["date"], "severity": incident["severity"]}],
    \n ids=[f"incident_{incident['id']}"]
    \n )<\/p>\n

    def search_similar_issues(self, query: str, top_k: int = 5) -> list:
    \n """\u641c\u7d22\u76f8\u4f3c\u5386\u53f2\u95ee\u9898"""
    \n embedding = self.embedder.encode(query).tolist()
    \n results = self.collection.query(
    \n query_embeddings=[embedding],
    \n n_results=top_k
    \n )
    \n return results["documents"][0] if results["documents"] else []<\/p>\n

    💡 \u6548\u7387\u6280\u5de7 \u2461：\u522b\u7528OpenAI\u7684Embedding API\u505a\u4e2d\u6587\u68c0\u7d22，BAAI\/bge-large-zh-v1.5\u5728\u4e2d\u6587\u8bed\u4e49\u76f8\u4f3c\u5ea6\u4e0a\u540a\u6253\u6240\u6709\u56fd\u5916\u6a21\u578b\u2014\u2014\u800c\u4e14\u514d\u8d39，\u672c\u5730\u8dd1，\u6bcf\u79d2\u51e0\u767e\u6761\u3002\u7701\u4e0b\u6765\u7684Embedding API\u8d39\u7528\u591f\u8bf7\u5168\u7ec4\u559d\u4e00\u4e2a\u6708\u5976\u8336\u3002<\/p>\n

    💡 \u6548\u7387\u6280\u5de7 \u2462：\u77e5\u8bc6\u5e93\u4e0d\u662f"\u642d\u597d\u5c31\u5b8c\u4e8b"\u3002\u5b9a\u4e00\u4e2aSOP\u2014\u2014\u6bcf\u6b21\u7ebf\u4e0a\u6545\u969c\u89e3\u51b3\u540e，\u5fc5\u987b\u572830\u5206\u949f\u5185\u628a\u590d\u76d8\u6587\u6863\u5582\u7ed9\u5411\u91cf\u5e93\u3002\u5426\u5219\u4f60\u7684AI\u8fd0\u7ef4Agent\u5c31\u662f\u4e2a"\u5931\u5fc6\u75c7\u60a3\u8005"\u3002<\/p>\n

    \n

    \u516d\u3001\u5b89\u5168\u5408\u89c4：\u522b\u8ba9AI\u628a\u4f60\u7684\u4ee3\u7801\u6cc4\u9732\u5230\u516c\u7f51<\/h3>\n

    \u8fd9\u4e00\u8282\u5f88\u77ed，\u4f46\u53ef\u80fd\u662f\u4f60\u8001\u677f\u552f\u4e00\u8ba4\u771f\u770b\u7684\u4e00\u8282\u3002<\/p>\n

    AI DevOps\u6700\u5927\u7684\u98ce\u9669\u4e0d\u662f\u6280\u672f，\u662f\u5b89\u5168\u3002 \u4f60\u8ba9AI\u8bfb\u4f60\u7684\u6240\u6709\u4ee3\u7801\u3001\u6240\u6709\u65e5\u5fd7\u3001\u6240\u6709\u9700\u6c42\u6587\u6863\u2014\u2014\u7136\u540e\u628a\u8fd9\u4e9b\u6570\u636e\u53d1\u7ed9OpenAI\u7684API？<\/p>\n

    \u4e09\u4e2a\u94c1\u5f8b：<\/p>\n

  • \u654f\u611f\u4fe1\u606f\u8131\u654f：\u53d1\u7ed9LLM\u4e4b\u524d，\u8131\u654f\u6240\u6709API Key\u3001\u5bc6\u7801\u3001\u5185\u7f51IP<\/li>\n

    # \u26a0\ufe0f \u654f\u611f\u4fe1\u606f\u8fc7\u6ee4\u5668（\u53d1\u9001LLM\u524d\u5fc5\u8c03）
    \nimport re<\/p>\n

    SENSITIVE_PATTERNS = [
    \n (r'sk-[A-Za-z0-9]{32,}', '[OPENAI_API_KEY]'), # OpenAI Key
    \n (r'AKIA[0-9A-Z]{16}', '[AWS_ACCESS_KEY]'), # AWS Key
    \n (r'(?<=password=)[^\\\\s&]+', '[PASSWORD]'), # \u5bc6\u7801\u53c2\u6570
    \n (r'\\\\b\\\\d{1,3}\\\\.\\\\d{1,3}\\\\.\\\\d{1,3}\\\\.\\\\d{1,3}\\\\b', '[INTERNAL_IP]'), # \u5185\u7f51IP
    \n (r'Bearer\\\\s+[A-Za-z0-9\\\\-._~+\/]+=*', '[AUTH_TOKEN]'), # JWT\/Token
    \n]<\/p>\n

    def sanitize_before_llm(text: str) -> str:
    \n """\u53d1\u9001LLM\u524d\u8131\u654f"""
    \n for pattern, replacement in SENSITIVE_PATTERNS:
    \n text = re.sub(pattern, replacement, text)
    \n return text<\/p>\n

  • \n

    \u79c1\u6709\u5316\u90e8\u7f72：\u6838\u5fc3\u4ee3\u7801\u76f8\u5173\u7684Prompt\u8d70\u672c\u5730\u90e8\u7f72\u7684\u5f00\u6e90\u6a21\u578b（\u5982Qwen2.5-Coder\u3001DeepSeek-Coder），\u53ea\u6709\u975e\u654f\u611f\u4efb\u52a1\u624d\u8d70\u4e91\u7aefAPI\u3002<\/p>\n<\/li>\n

  • \n

    \u5ba1\u8ba1\u65e5\u5fd7：\u6240\u6709LLM\u8c03\u7528\u5fc5\u987b\u7559\u75d5\u2014\u2014\u8c01\u3001\u4ec0\u4e48\u65f6\u5019\u3001\u4f20\u4e86\u4ec0\u4e48Prompt\u3001\u8fd4\u56de\u4e86\u4ec0\u4e48\u3002\u4e07\u4e00\u51fa\u95ee\u9898，\u80fd\u6eaf\u6e90\u3002<\/p>\n<\/li>\n

    # \u26a0\ufe0f LLM\u8c03\u7528\u5ba1\u8ba1\u65e5\u5fd7（\u6bcf\u6761\u8c03\u7528\u90fd\u8bb0\u5f55）
    \nimport json, time, hashlib<\/p>\n

    def audit_llm_call(user: str, task: str, prompt: str, response: str, model: str):
    \n """\u6240\u6709LLM\u8c03\u7528\u5fc5\u987b\u7ecf\u8fc7\u6b64\u51fd\u6570"""
    \n record = {
    \n "timestamp": time.time(),
    \n "user": user,
    \n "task_type": task,
    \n "model": model,
    \n "prompt_hash": hashlib.sha256(prompt.encode()).hexdigest()[:16],
    \n "prompt_length": len(prompt),
    \n "response_length": len(response),
    \n "tokens_used": len(prompt.split()) + len(response.split()) # \u7c97\u7565\u4f30\u7b97
    \n }
    \n # \u5199\u5165\u5ba1\u8ba1\u65e5\u5fd7（\u4e0d\u5b58prompt\u539f\u6587，\u53ea\u5b58hash）
    \n with open("\/var\/log\/llm_audit.jsonl", "a") as f:
    \n f.write(json.dumps(record) + "\\\\n")<\/p>\n

    \u26a0\ufe0f \u907f\u5751 \u2460：\u4e0d\u8981\u628a\u5b8c\u6574\u4ee3\u7801\u76f4\u63a5\u8d34\u7ed9\u516c\u7f51LLM API\u5904\u7406\u3002\u5148\u8131\u654f\u3002\u5148\u8131\u654f\u3002\u5148\u8131\u654f\u3002\u91cd\u8981\u7684\u4e8b\u8bf4\u4e09\u904d\u3002\u5df2\u7ecf\u6709\u591a\u5bb6\u516c\u53f8\u56e0\u4e3a\u628a\u5305\u542b\u5185\u90e8API endpoint\u7684\u4ee3\u7801\u53d1\u7ed9ChatGPT\u5bfc\u81f4\u4fe1\u606f\u6cc4\u9732\u2014\u2014Sam Altman\u4e0d\u4f1a\u4e3b\u52a8\u770b\u4f60\u4ee3\u7801，\u4f46\u4e07\u4e00OpenAI\u88ab\u9ed1\u4e86\u5462？<\/p>\n

    \u26a0\ufe0f \u907f\u5751 \u2461：\u6a21\u578b\u8f93\u51fa\u4e0d\u8981\u76f4\u63a5\u6267\u884c\u3002AI\u751f\u6210\u7684\u90e8\u7f72\u811a\u672c\u3001\u6570\u636e\u5e93\u8fc1\u79fbSQL\u3001kubectl\u547d\u4ee4\u2014\u2014\u5fc5\u987b\u7ecf\u8fc7\u4eba\u7c7b\u786e\u8ba4\u3002\u642d\u5efa\u4e00\u4e2a"\u4eba\u5de5\u786e\u8ba4\u95f8\u95e8"，\u9ad8\u98ce\u9669\u7684\u547d\u4ee4（\u5982DROP TABLE\u3001kubectl delete）\u89e6\u53d1\u4e8c\u6b21\u786e\u8ba4\u3002<\/p>\n

    \u26a0\ufe0f \u907f\u5751 \u2462：\u522b\u628a\u5411\u91cf\u6570\u636e\u5e93\u5f53\u6570\u636e\u5e93\u7528\u3002\u5411\u91cf\u68c0\u7d22\u662f"\u8fd1\u4f3c\u641c\u7d22"，\u4e0d\u662f"\u7cbe\u786e\u641c\u7d22"\u3002\u5982\u679c\u4f60\u9700\u8981\u5728\u77e5\u8bc6\u5e93\u91cc\u7cbe\u786e\u67e5"\u8ba2\u5355\u53f7ABC123\u7684\u6240\u6709\u64cd\u4f5c\u8bb0\u5f55"，\u90a3\u5e94\u8be5\u7528SQL\/ES，\u4e0d\u662f\u5411\u91cf\u5e93\u3002<\/p>\n

    \n

    \u4e03\u3001MVP\u843d\u5730：48\u5c0f\u65f6\u8dd1\u901aFlask\u5168\u6d41\u7a0b<\/h3>\n

    \u804a\u4e86\u8fd9\u4e48\u591a\u67b6\u6784，\u6765\u70b9\u5b9e\u5728\u7684\u3002\u4e24\u5929\u4e4b\u5185，\u4f60\u80fd\u642d\u51fa\u4e00\u4e2a\u4ec0\u4e48？<\/p>\n

    \u76ee\u6807：\u4e00\u4e2aFlask\u5e94\u7528，\u4ece\u9700\u6c42\u2192\u4ee3\u7801\u2192\u6d4b\u8bd5\u2192\u90e8\u7f72\u2192\u76d1\u63a7，AI\u5168\u94fe\u8def\u53c2\u4e0e\u3002<\/p>\n

    flowchart TD
    \n Start["🚀 \u5f00\u59cbMVP\u642d\u5efa"] –> Step1["Day1 AM<br\/>\u642d\u5efa\u7f16\u6392\u5f15\u64ce<br\/>+ \u591a\u6a21\u578b\u8def\u7531"]
    \n Step1 –> Step2["Day1 PM<br\/>\u63a5\u5165\u77e5\u8bc6\u5e93<br\/>+ 5\u4e2aPrompt\u6a21\u677f"]
    \n Step2 –> Step3["Day1 Night<br\/>\u8dd1\u901a\u9700\u6c42\u5206\u6790<br\/>+ \u4ee3\u7801\u751f\u6210 Pipeline"]
    \n Step3 –> Step4["Day2 AM<br\/>\u63a5\u5165CI\/CD<br\/>+ \u81ea\u52a8\u90e8\u7f72\u811a\u672c"]
    \n Step4 –> Step5["Day2 PM<br\/>\u63a5\u5165\u65e5\u5fd7\u76d1\u63a7<br\/>+ AI\u544a\u8b66\u5206\u6790"]
    \n Step5 –> Done["\u2705 MVP Ready<br\/>\u7aef\u5230\u7aefDemo\u53ef\u6f14\u793a"]<\/p>\n

    style Start fill:#27ae60,color:#fff
    \n style Done fill:#4a90d9,color:#fff<\/p>\n

    MVP\u6280\u672f\u6808（\u6700\u5c0f\u53ef\u884c\u7248）<\/h4>\n

    # \u26a0\ufe0f MVP\u6280\u672f\u6808（docker-compose.yml）
    \nversion: "3.8"
    \nservices:
    \n # \u7f16\u6392\u5f15\u64ce
    \n llm-orchestrator:
    \n image: python:3.11-slim
    \n volumes:
    \n – .\/orchestrator:\/app
    \n command: python \/app\/main.py
    \n environment:
    \n – OPENAI_API_KEY=${OPENAI_API_KEY}
    \n – DEEPSEEK_API_KEY=${DEEPSEEK_API_KEY}
    \n – ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY}
    \n ports:
    \n – "8000:8000"<\/p>\n

    # \u5411\u91cf\u6570\u636e\u5e93
    \n chromadb:
    \n image: chromadb\/chroma:latest
    \n volumes:
    \n – .\/chroma_data:\/chroma\/chroma
    \n ports:
    \n – "8001:8000"<\/p>\n

    # \u56fe\u6570\u636e\u5e93
    \n neo4j:
    \n image: neo4j:5
    \n environment:
    \n – NEO4J_AUTH=neo4j\/password123
    \n ports:
    \n – "7474:7474"
    \n – "7687:7687"<\/p>\n

    # \u76ee\u6807\u5e94\u7528（\u90a3\u4e2a\u88abAI"\u7167\u987e"\u7684Flask）
    \n target-app:
    \n build: .\/flask-app
    \n ports:
    \n – "5000:5000"
    \n depends_on:
    \n – chromadb<\/p>\n

    # \u26a0\ufe0f \u7f16\u6392\u5f15\u64ce\u6838\u5fc3\u5165\u53e3（flask-app\/orchestrator\/main.py）
    \nfrom flask import Flask, request, jsonify
    \nfrom llm_router import LLMRouter
    \nfrom knowledge_retriever import KnowledgeRetriever
    \nfrom prompt_templates import load_template
    \nfrom sanitizer import sanitize_before_llm<\/p>\n

    app = Flask(__name__)
    \nrouter = LLMRouter()
    \nretriever = KnowledgeRetriever()<\/p>\n

    @app.route("\/api\/devops\/analyze-requirement", methods=["POST"])
    \ndef analyze_requirement():
    \n """AI\u5206\u6790\u9700\u6c42 \u2192 \u7528\u6237\u6545\u4e8b"""
    \n data = request.json
    \n raw_req = data.get("requirement", "")<\/p>\n

    # Step 1: \u4ece\u77e5\u8bc6\u5e93\u68c0\u7d22\u76f8\u4f3c\u9700\u6c42（\u907f\u514d\u91cd\u590d\u9020\u8f6e\u5b50）
    \n similar = retriever.search_similar_issues(raw_req, top_k=3)<\/p>\n

    # Step 2: \u52a0\u8f7dPrompt\u6a21\u677f
    \n prompt = load_template("requirement_analysis").format(
    \n requirement=raw_req,
    \n similar_cases="\\\\n".join(similar)
    \n )<\/p>\n

    # Step 3: \u8131\u654f\u540e\u8c03\u7528LLM
    \n safe_prompt = sanitize_before_llm(prompt)
    \n result = router.call(TaskType.REQUIREMENT_ANALYSIS, safe_prompt)<\/p>\n

    return jsonify({"user_stories": result})<\/p>\n

    @app.route("\/api\/devops\/generate-tests", methods=["POST"])
    \ndef generate_tests():
    \n """AI\u751f\u6210\u6d4b\u8bd5\u7528\u4f8b"""
    \n data = request.json
    \n source_code = data.get("source_code", "")
    \n existing_tests = data.get("existing_tests", "")<\/p>\n

    prompt = load_template("test_generation").format(
    \n source_code=sanitize_before_llm(source_code),
    \n existing_tests=existing_tests
    \n )<\/p>\n

    result = router.call(TaskType.TEST_GENERATION, prompt)
    \n return jsonify({"test_code": result})<\/p>\n

    if __name__ == "__main__":
    \n app.run(host="0.0.0.0", port=8000)<\/p>\n

    \u4e24\u5929\u4e4b\u540e\u7684Demo\u6f14\u793a\u6548\u679c：\u4f60\u6253\u5f00\u6d4f\u89c8\u5668，\u8f93\u5165\u4e00\u53e5"\u6211\u9700\u8981\u4e00\u4e2a\u7528\u6237\u6ce8\u518c\u767b\u5f55\u7684API"，\u7cfb\u7edf\u81ea\u52a8\u751f\u6210：<\/p>\n

      \n
    • \u2705 4\u4e2a\u7528\u6237\u6545\u4e8b + \u9a8c\u6536\u6807\u51c6<\/li>\n
    • \u2705 Flask\u4ee3\u7801（\u542bJWT\u9274\u6743\u3001\u5bc6\u7801\u54c8\u5e0c\u3001\u90ae\u7bb1\u9a8c\u8bc1）<\/li>\n
    • \u2705 15\u4e2apytest\u6d4b\u8bd5\u7528\u4f8b<\/li>\n
    • \u2705 Dockerfile + docker-compose\u914d\u7f6e<\/li>\n
    • \u2705 GitLab CI Pipeline<\/li>\n<\/ul>\n

      \u5168\u7a0b\u4f60\u53ea\u5199\u4e86\u4e00\u53e5\u9700\u6c42\u3002<\/p>\n

      🤣 \u5e7d\u9ed8\u65f6\u523b：MVP\u6f14\u793a\u5b8c\u540e，\u8001\u677f\u6c89\u9ed8\u4e86\u5f88\u4e45，\u95ee\u4e86\u4e00\u53e5：\u201c\u90a3\u6211\u4eec\u4ee5\u540e\u8fd8\u9700\u8981\u62db\u4eba\u5417？\u201d \u6211\u8bf4"\u9700\u8981，\u9700\u8981\u6709\u4eba\u7ed9AI\u5199Prompt"\u2014\u2014\u5f53\u7136\u8fd9\u662f\u73a9\u7b11\u3002\u771f\u6b63\u9700\u8981\u7684\u662f：\u6709\u4eba\u5b9a\u4e49"AI\u5e94\u8be5\u505a\u4ec0\u4e48"，\u6709\u4eba\u5ba1\u67e5"AI\u505a\u5f97\u5bf9\u4e0d\u5bf9"\u3002\u4eba\u7c7b\u7684\u89d2\u8272\u4ece"\u6267\u884c\u8005"\u53d8\u6210\u4e86"\u51b3\u7b56\u8005"\u3002<\/p>\n

      \n

      \u516b\u3001\u67b6\u6784\u5168\u666f：L5\u5230\u5e95\u957f\u4ec0\u4e48\u6837<\/h3>\n

      \u5982\u679c\u4f60\u770b\u5230\u8fd9\u91cc\u8fd8\u6ca1\u5173\u9875\u9762，\u8bf4\u660e\u4f60\u662f\u771f\u60f3\u641e\u660e\u767d\u3002\u6765，\u4e0a\u786c\u83dc\u2014\u2014L5\u5168\u666f\u67b6\u6784\u56fe\u3002<\/p>\n

      flowchart TB
      \n subgraph \u7528\u6237\u5c42["👤 \u7528\u6237\u4ea4\u4e92\u5c42"]
      \n U1["Web IDE<br\/>\u9700\u6c42\u8f93\u5165"]
      \n U2["CLI\u5de5\u5177<br\/>ai-devops deploy"]
      \n U3["\u4f01\u4e1a\u5fae\u4fe1\/Slack<br\/>Bot\u4ea4\u4e92"]
      \n end<\/p>\n

      subgraph \u7f51\u5173\u5c42["🔐 API Gateway"]
      \n GW["\u8ba4\u8bc1\u9274\u6743 + \u9650\u6d41<br\/>+ \u654f\u611f\u4fe1\u606f\u8fc7\u6ee4"]
      \n end<\/p>\n

      subgraph \u7f16\u6392\u5c42["🧠 AI\u7f16\u6392\u5f15\u64ce"]
      \n OR["\u4efb\u52a1\u8c03\u5ea6\u5668"]
      \n MR["\u591a\u6a21\u578b\u8def\u7531\u5668"]
      \n PM["Prompt\u6a21\u677f\u5f15\u64ce"]
      \n CT["\u4e0a\u4e0b\u6587\u7ba1\u7406\u5668"]
      \n end<\/p>\n

      subgraph \u77e5\u8bc6\u5c42["📚 \u77e5\u8bc6\u5e93"]
      \n VD["\u5411\u91cf\u6570\u636e\u5e93<br\/>Chromadb\/Milvus"]
      \n GD["\u56fe\u6570\u636e\u5e93<br\/>Neo4j"]
      \n RE["\u89c4\u5219\u5f15\u64ce<br\/>\u5b89\u5168\u767d\u540d\u5355\/\u67b6\u6784\u7ea6\u675f"]
      \n end<\/p>\n

      subgraph \u80fd\u529b\u5c42["🔧 AI\u80fd\u529b\u77e9\u9635"]
      \n C1["\u9700\u6c42\u5206\u6790<br\/>Agent"]
      \n C2["\u4ee3\u7801\u751f\u6210<br\/>Agent"]
      \n C3["\u6d4b\u8bd5\u751f\u6210<br\/>Agent"]
      \n C4["\u90e8\u7f72\u7f16\u6392<br\/>Agent"]
      \n C5["\u8fd0\u7ef4\u8bca\u65ad<br\/>Agent"]
      \n end<\/p>\n

      subgraph \u6267\u884c\u5c42["\u26a1 \u6267\u884c\u5f15\u64ce"]
      \n E1["Git\u96c6\u6210<br\/>\u81ea\u52a8\u521b\u5efaPR"]
      \n E2["CI\/CD\u89e6\u53d1<br\/>Jenkins\/GitLab"]
      \n E3["K8s\u64cd\u4f5c<br\/>\u7070\u5ea6\/\u56de\u6eda"]
      \n E4["\u76d1\u63a7\u544a\u8b66<br\/>Prometheus\/Grafana"]
      \n end<\/p>\n

      subgraph \u57fa\u7840\u8bbe\u65bd["🖥\ufe0f \u57fa\u7840\u8bbe\u65bd"]
      \n INF["Kubernetes + Docker<br\/>+ \u79c1\u6709\u6a21\u578b\u90e8\u7f72<br\/>+ GPU\u96c6\u7fa4"]
      \n end<\/p>\n

      \u7528\u6237\u5c42 –> \u7f51\u5173\u5c42
      \n \u7f51\u5173\u5c42 –> \u7f16\u6392\u5c42
      \n \u7f16\u6392\u5c42 <–> \u77e5\u8bc6\u5c42
      \n \u7f16\u6392\u5c42 –> \u80fd\u529b\u5c42
      \n \u80fd\u529b\u5c42 –> \u6267\u884c\u5c42
      \n \u6267\u884c\u5c42 –> \u57fa\u7840\u8bbe\u65bd
      \n \u6267\u884c\u5c42 -.->|"📊 \u6307\u6807\u56de\u6d41"| \u77e5\u8bc6\u5c42<\/p>\n

      style \u7f16\u6392\u5c42 fill:#e74c3c,color:#fff,stroke:#c0392b,stroke-width:3px
      \n style \u77e5\u8bc6\u5c42 fill:#27ae60,color:#fff
      \n style \u80fd\u529b\u5c42 fill:#4a90d9,color:#fff<\/p>\n

      \u67b6\u6784\u89e3\u8bfb（\u7ed9\u8001\u677f\u770b\u7248）：<\/p>\n\n\u5c42\u7ea7\u804c\u8d23\u4e00\u53e5\u8bdd\u89e3\u91ca<\/tr>\n\n\n\n\n\n\n\n\n
      \u7528\u6237\u5c42<\/td>\n\u8f93\u5165\u9700\u6c42\u3001\u67e5\u770b\u7ed3\u679c<\/td>\n\u4f60\u4e0d\u7528\u61c2\u4ee3\u7801，\u8bf4\u4eba\u8bdd\u5c31\u884c<\/td>\n<\/tr>\n
      \u7f51\u5173\u5c42<\/td>\n\u5b89\u5168\u3001\u9274\u6743\u3001\u9650\u6d41<\/td>\n\u4f60\u7684\u4ee3\u7801\u4e0d\u4f1a\u6cc4\u9732\u5230\u516c\u7f51<\/td>\n<\/tr>\n
      \u7f16\u6392\u5c42<\/td>\n\u5927\u8111，\u8c03\u5ea6\u4e00\u5207<\/td>\n\u591a\u4e2aAI\u6a21\u578b\u534f\u540c\u5de5\u4f5c<\/td>\n<\/tr>\n
      \u77e5\u8bc6\u5c42<\/td>\n\u8bb0\u5fc6，\u5b58\u7ecf\u9a8c<\/td>\n\u4e0a\u6b21\u8e29\u8fc7\u7684\u5751，\u4e0b\u6b21\u4e0d\u4f1a\u518d\u8e29<\/td>\n<\/tr>\n
      \u80fd\u529b\u5c42<\/td>\n\u5177\u4f53AI\u4efb\u52a1<\/td>\n\u9700\u6c42\u5206\u6790\u3001\u5199\u4ee3\u7801\u3001\u5199\u6d4b\u8bd5\u2026\u2026<\/td>\n<\/tr>\n
      \u6267\u884c\u5c42<\/td>\n\u5b9e\u9645\u64cd\u4f5c<\/td>\n\u63d0\u4ea4PR\u3001\u89e6\u53d1CI\u3001\u90e8\u7f72\u4e0a\u7ebf<\/td>\n<\/tr>\n
      \u57fa\u7840\u8bbe\u65bd<\/td>\n\u5e95\u5c42\u8d44\u6e90<\/td>\nK8s + GPU，\u8dd1AI\u548c\u4f60\u7684\u5e94\u7528<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
      \n

      \u4e5d\u3001\u26a0\ufe0f\u8e29\u5751\u5b9e\u5f55\u4e0e💡\u6548\u7387\u6280\u5de7<\/h3>\n

      \u26a0\ufe0f \u8e29\u5751\u5927\u5168（\u8840\u6cea\u6362\u6765\u7684\u7ecf\u9a8c）<\/h4>\n

      \u26a0\ufe0f \u907f\u5751 \u2460：LLM\u8f93\u51fa\u4e0d\u7a33\u5b9a，\u522b\u628aAuto-Pilot\u5f53Autopilot<\/p>\n

      AI\u751f\u6210\u7684\u4ee3\u7801，\u8fd9\u6b21\u80fd\u8dd1，\u4e0b\u6b21\u53ef\u80fd\u5c31\u8dd1\u4e0d\u4e86\u2014\u2014\u56e0\u4e3aLLM\u6709\u968f\u673a\u6027（temperature\u53c2\u6570）\u3002\u89e3\u51b3\u65b9\u6848：\u6240\u6709AI\u8f93\u51fa\u5fc5\u987b\u7ecf\u8fc7\u81ea\u52a8\u5316\u9a8c\u8bc1\u624d\u80fd\u8fdb\u5165\u4e0b\u4e00\u73af\u8282\u3002<\/p>\n

      # \u26a0\ufe0f \u9a8c\u8bc1\u95e8\u7981\u2014\u2014AI\u751f\u6210\u7684\u4ee3\u7801\u5148\u8dd1\u6d4b\u8bd5\u518d\u5408\u5165
      \ndef validate_ai_output(code: str, language: str) -> bool:
      \n checks = [
      \n syntax_check(code, language), # \u8bed\u6cd5\u68c0\u67e5
      \n lint_check(code, language), # \u4ee3\u7801\u89c4\u8303
      \n security_scan(code), # \u5b89\u5168\u626b\u63cf（\u786c\u7f16\u7801\u5bc6\u94a5\u3001SQL\u6ce8\u5165）
      \n run_ai_tests(code), # \u8dd1AI\u81ea\u5df1\u751f\u6210\u7684\u6d4b\u8bd5\u7528\u4f8b
      \n ]
      \n return all(checks)<\/p>\n

      \u26a0\ufe0f \u907f\u5751 \u2461：\u4e0a\u4e0b\u6587\u7a97\u53e3\u4e0d\u591f\u7528，\u5927\u9879\u76ee\u4ee3\u7801\u585e\u4e0d\u8fdb<\/p>\n

      GPT-4o\u7684128K\u7a97\u53e3\u770b\u4f3c\u5f88\u5927，\u4f46\u4e00\u4e2a\u4e2d\u578b\u9879\u76ee\u7684\u4ee3\u7801\u8fdc\u8d85\u8fd9\u4e2a\u91cf\u3002\u89e3\u51b3\u65b9\u6848：\u4ee3\u7801\u5206\u7247+\u6458\u8981+\u6309\u9700\u68c0\u7d22\u3002<\/p>\n

      # \u26a0\ufe0f \u4ee3\u7801\u5206\u7247\u7b56\u7565
      \ndef chunk_repository(repo_path: str, chunk_size: int = 5000) -> list:
      \n """\u5c06\u5927\u578b\u4ee3\u7801\u5e93\u5206\u7247，\u6bcf\u72475000\u5b57\u7b26"""
      \n chunks = []
      \n for file in walk_repo(repo_path):
      \n content = read_file(file)
      \n if len(content) > chunk_size:
      \n # \u5927\u6587\u4ef6：\u6309\u51fd\u6570\/\u7c7b\u8fb9\u754c\u5207\u5206
      \n chunks.extend(split_by_functions(content, chunk_size))
      \n else:
      \n chunks.append({"file": file, "content": content})
      \n return chunks<\/p>\n

      # \u68c0\u7d22\u65f6\u7528\u5411\u91cf\u641c\u7d22\u627e\u5230\u76f8\u5173\u4ee3\u7801\u7247，\u800c\u4e0d\u662f\u628a\u6574\u4e2a\u9879\u76ee\u585e\u7ed9LLM<\/p>\n

      \u26a0\ufe0f \u907f\u5751 \u2462：AI\u4e0d\u77e5\u9053"\u8fd9\u4e2a\u4e0d\u80fd\u6539"<\/p>\n

      \u4f60\u628a\u4e00\u4e2a\u8001\u9879\u76ee\u4ea4\u7ed9AI，\u201c\u4f18\u5316\u4e00\u4e0b\u4ee3\u7801\u7ed3\u6784\u201d\u2014\u2014\u5b83\u53ef\u80fd\u628a\u4f60\u7684\u6838\u5fc3\u4e1a\u52a1\u903b\u8f91\u91cd\u5199\u4e86，\u800c\u4e14\u770b\u8d77\u6765\u8fd8"\u66f4\u4f18\u96c5"\u3002\u7136\u540e\u7ebf\u4e0a\u70b8\u4e86\u3002\u89e3\u51b3\u65b9\u6848：\u67b6\u6784\u7ea6\u675f\u89c4\u5219\u5f15\u64ce\u3002<\/p>\n

      # \u26a0\ufe0f \u67b6\u6784\u7ea6\u675f\u914d\u7f6e（AI\u4e0d\u80fd\u8fdd\u53cd\u7684\u89c4\u5219）
      \nconstraints:
      \n no_modify:
      \n – "payment_service\/*" # \u652f\u4ed8\u6a21\u5757，AI\u7981\u6b62\u4fee\u6539
      \n – "core\/business_rules\/*" # \u6838\u5fc3\u4e1a\u52a1\u89c4\u5219
      \n must_not_use:
      \n – "eval()" # \u7981\u6b62eval
      \n – "os.system()" # \u7981\u6b62\u6267\u884c\u7cfb\u7edf\u547d\u4ee4
      \n – "pickle.loads(.*untrusted" # \u7981\u6b62\u53cd\u5e8f\u5217\u5316\u4e0d\u53ef\u4fe1\u6570\u636e
      \n architecture_rules:
      \n – "controller\u5c42\u4e0d\u5141\u8bb8\u76f4\u63a5\u8bbf\u95ee\u6570\u636e\u5e93"
      \n – "\u6240\u6709API\u5fc5\u987b\u5e26\u8ba4\u8bc1\u4e2d\u95f4\u4ef6"<\/p>\n

      💡 \u6548\u7387\u6280\u5de7（\u8ba9\u4f60\u5c11\u52a0\u73ed）<\/h4>\n

      💡 \u6548\u7387\u6280\u5de7 \u2460：Prompt\u6a21\u677f\u7528YAML\u7ba1\u7406，Jinja2\u6e32\u67d3<\/p>\n

      \u524d\u9762\u63d0\u8fc7，\u518d\u5f3a\u8c03\u4e00\u904d：\u6539Prompt\u4e0d\u8981\u6539\u4ee3\u7801\u3002\u628aPrompt\u62bd\u6210YAML\u6587\u4ef6，\u4ea7\u54c1\u7ecf\u7406\u90fd\u80fd\u8c03Prompt（\u867d\u7136\u7ed3\u679c\u662f\u53e6\u4e00\u56de\u4e8b）\u3002<\/p>\n

      💡 \u6548\u7387\u6280\u5de7 \u2461：\u56fd\u4ea7Embedding\u6a21\u578b\u505a\u4e2d\u6587\u8bed\u4e49\u68c0\u7d22，\u514d\u8d39\u4e14\u66f4\u51c6<\/p>\n

      BAAI\/bge-large-zh-v1.5 + Chromadb，\u672c\u5730\u8dd1，\u4e2d\u6587\u8bed\u4e49\u641c\u7d22\u6548\u679c\u78be\u538bOpenAI\u7684text-embedding-ada-002\u3002\u4e0d\u7528\u7ed9OpenAI\u4ea4Embedding\u7a0e\u3002<\/p>\n

      💡 \u6548\u7387\u6280\u5de7 \u2462：AI\u751f\u6210\u7684\u4ee3\u7801\u5fc5\u987b\u5148\u8dd1Lint+Security Scan<\/p>\n

      \u4e00\u6761Pipeline\u5c31\u80fd\u7701\u53bb\u65e0\u6570\u4e2aCode Review\u91cc\u7684"\u8fd9\u91cc\u7a7a\u884c\u4e0d\u5bf9""\u8fd9\u91cc\u8be5\u7528const"\u7684\u8bc4\u8bba\u3002\u8ba9\u4eba\u7c7breviewer\u628a\u65f6\u95f4\u82b1\u5728\u67b6\u6784\u8bbe\u8ba1\u4e0a，\u800c\u4e0d\u662f\u6570\u7a7a\u683c\u3002<\/p>\n

      # \u26a0\ufe0f AI\u4ee3\u7801\u8d28\u91cf\u95e8\u7981（GitLab CI）
      \nai-code-quality-gate:
      \n stage: validate
      \n script:
      \n – ruff check ai_generated_code\/ # Python Lint
      \n – bandit -r ai_generated_code\/ # \u5b89\u5168\u626b\u63cf
      \n – pytest ai_generated_tests\/ # \u8dd1AI\u751f\u6210\u7684\u6d4b\u8bd5
      \n rules:
      \n – if: $CI_PIPELINE_SOURCE == "ai_generated" # \u4ec5AI\u751f\u6210\u7684\u4ee3\u7801\u89e6\u53d1<\/p>\n

      \n

      \u5341\u3001\u603b\u7ed3\u4e0e\u7cfb\u5217\u9884\u544a<\/h3>\n

      \u6838\u5fc3\u8981\u70b9\u56de\u987e<\/h4>\n
    • AI DevOps\u4e0d\u662f"AI\u66ff\u4ee3DevOps"，\u662fAI\u5d4c\u5165\u8f6f\u4ef6\u5168\u751f\u547d\u5468\u671f\u7684\u6bcf\u4e00\u4e2a\u73af\u8282<\/li>\n
    • LLM\u7f16\u6392\u5f15\u64ce\u662f\u5927\u8111\u2014\u2014\u591a\u6a21\u578b\u8def\u7531 + Prompt\u6a21\u677f + \u4e0a\u4e0b\u6587\u7ba1\u7406，\u4e09\u4e2a\u7f3a\u4e00\u4e0d\u53ef<\/li>\n
    • \u77e5\u8bc6\u5e93\u662f\u8bb0\u5fc6\u2014\u2014\u5411\u91cf\u5e93（\u6a21\u7cca\u8bed\u4e49\u5339\u914d）+ \u56fe\u5e93（\u5173\u7cfb\u63a8\u7406），\u8ba9AI\u62e5\u6709"\u9879\u76ee\u7ecf\u9a8c"<\/li>\n
    • \u5b89\u5168\u662f\u7ea2\u7ebf\u2014\u2014\u8131\u654f\u8fc7\u6ee4 + \u4eba\u5de5\u786e\u8ba4\u95f8\u95e8 + \u5ba1\u8ba1\u65e5\u5fd7，\u4e09\u6761\u5e95\u7ebf<\/li>\n
    • MVP\u4e24\u5468\u53ef\u843d\u5730\u2014\u2014\u5148\u8dd1\u901aFlask\u5168\u6d41\u7a0b，\u518d\u6269\u5c55<\/li>\n

      \u6587\u672b\u4e09\u4ef6\u5957<\/h4>\n\n\u9879\u76ee\u5185\u5bb9<\/tr>\n\n\n\n\n
      📂 \u6e90\u7801（\u6982\u5ff5\u53c2\u8003）<\/td>\nLangChain Agent Protocol<\/td>\n<\/tr>\n
      📺 \u63a8\u8350\u89c6\u9891<\/td>\nAndrej Karpathy – Intro to Large Language Models<\/td>\n<\/tr>\n
      📚 \u63a8\u8350\u9605\u8bfb<\/td>\nGoogle – DORA DevOps Capabilities \u3001Building LLM Applications for Production<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
      \n

      \u7cfb\u5217\u6587\u7ae0\u5bfc\u822a<\/h4>\n\n\u7f16\u53f7\u6807\u9898\u72b6\u6001<\/tr>\n\n\n\n\n\n\n\n\n
      18<\/td>\nL5-1：MLflow\u5b9e\u9a8c\u8ffd\u8e2a\u4e0e\u6a21\u578b\u7ba1\u7406<\/td>\n\u2705 \u5df2\u53d1\u5e03<\/td>\n<\/tr>\n
      19<\/td>\nL5-2：\u6a21\u578b\u8bc4\u4f30\u4e0eA\/B\u6d4b\u8bd5<\/td>\n\u2705 \u5df2\u53d1\u5e03<\/td>\n<\/tr>\n
      20<\/td>\nL5-3：\u6a21\u578b\u670d\u52a1\u5316\u4e0e\u76d1\u63a7\u544a\u8b66<\/td>\n\u2705 \u5df2\u53d1\u5e03<\/td>\n<\/tr>\n
      21<\/td>\nL5\u5168\u666f：AI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8<\/td>\n👈 \u4f60\u5728\u8fd9\u91cc<\/td>\n<\/tr>\n
      22<\/td>\nL5\u5b9e\u6218：\u667a\u80fd\u9700\u6c42\u5206\u6790\u4e0e\u7528\u6237\u6545\u4e8b\u751f\u6210<\/td>\n🔜 \u5373\u5c06\u53d1\u5e03<\/td>\n<\/tr>\n
      23<\/td>\nL5\u5b9e\u6218：AI\u4ee3\u7801\u5ba1\u67e5Pipeline\u5168\u89e3\u6790<\/td>\n📅 \u8ba1\u5212\u4e2d<\/td>\n<\/tr>\n
      24<\/td>\nL5\u5b9e\u6218：\u667a\u80fd\u8fd0\u7ef4\u2014\u2014\u4ece\u544a\u8b66\u5230\u81ea\u52a8\u4fee\u590d<\/td>\n📅 \u8ba1\u5212\u4e2d<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
      \n

      🔜 \u4e0b\u4e00\u7bc7\u9884\u544a：\u300aL5\u5b9e\u6218\u2014\u2014\u667a\u80fd\u9700\u6c42\u5206\u6790\u4e0e\u7528\u6237\u6545\u4e8b\u751f\u6210\u300b<\/h4>\n

      \u4e0b\u4e00\u7bc7\u6211\u4eec\u5c06\u4ece\u67b6\u6784\u843d\u5230\u4ee3\u7801，\u624b\u628a\u624b\u6559\u4f60\u642d\u5efa\u4e00\u4e2a\u667a\u80fd\u9700\u6c42\u5206\u6790Agent：<\/p>\n

        \n
      • \u2728 Nature Language \u2192 \u7ed3\u6784\u5316\u7528\u6237\u6545\u4e8b（\u542b\u9a8c\u6536\u6807\u51c6）<\/li>\n
      • 📊 \u9700\u6c42\u6a21\u7cca\u5ea6\u8bc4\u4f30（\u81ea\u52a8\u8ffd\u95ee\u4e0d\u591f\u6e05\u6670\u7684\u9700\u6c42）<\/li>\n
      • 🔍 \u5386\u53f2\u9700\u6c42\u67e5\u91cd（\u201c\u8fd9\u4e2a\u9700\u6c42\u53bb\u5e74\u505a\u8fc7，\u522b\u91cd\u590d\u9020\u8f6e\u5b50\u201d）<\/li>\n
      • 🎯 \u81ea\u52a8\u4f30\u7b97\u5de5\u4f5c\u91cf（\u57fa\u4e8e\u5386\u53f2Story Point\u6570\u636e）<\/li>\n<\/ul>\n

        \u5173\u6ce8\u6211，\u4e0d\u8ff7\u8def\u3002<\/p>\n

        \n

        \u5982\u679c\u8fd9\u7bc7\u6587\u7ae0\u8ba9\u4f60\u5bf9AI DevOps\u6709\u4e86\u65b0\u7684\u8ba4\u8bc6，\u8bf7 👍\u70b9\u8d5e\u3001\u2b50\u6536\u85cf\u3001💬\u8bc4\u8bba\u4e09\u8fde\u3002\u4f60\u7684\u652f\u6301\u662f\u4f5c\u8005\u6301\u7eed\u8f93\u51fa\u7684\u6700\u5927\u52a8\u529b\u3002<\/p>\n

        \u6807\u7b7e：AI DevOps\u3001MLOps\u3001CI\/CD\u3001\u5168\u6d41\u7a0b\u81ea\u52a8\u5316\u3001\u667a\u80fd\u8fd0\u7ef4\u3001LLMOps\u3001\u8f6f\u4ef6\u5de5\u7a0b<\/p>\n","protected":false},"excerpt":{"rendered":"

        \u201c\u8001\u677f，\u6211\u4eec\u7684DevOps\u5e73\u53f0\u63a5\u5165\u4e86\u5927\u6a21\u578b\u2014\u2014\u73b0\u5728\u4e0d\u4ec5\u4ee3\u7801\u662fAI\u5199\u7684，\u8fde\u9700\u6c42\u6587\u6863\u3001\u6d4b\u8bd5\u7528\u4f8b\u3001\u90e8\u7f72\u811a\u672c\u3001\u8fd0\u7ef4\u544a\u8b66\u90fd\u662fAI\u5728\u641e，\u4f60\u8bf4\u6211\u4eec\u7a0b\u5e8f\u5458\u8fd8\u80fd\u5e72\u5565？\u201d \u201c\u2026\u2026\u6478\u9c7c\u3002\u201d \u76ee\u5f55
        \n\u4e00\u3001\u5f00\u7bc7\u66b4\u51fb：AI DevOps\u5230\u5e95\u662f\u4e0d\u662f\u753b\u997c？
        \n\u4e8c\u3001\u4e00\u5f20\u56fe\u770b\u7a7fAI DevOps\u5168\u666f
        \n\u4e09\u3001\u5404\u73af\u8282AI\u5207\u5165\u5b9e\u6218
        \n3.1 \u9700\u6c42\u2192\u7528\u6237\u6545\u4e8b：\u522b\u8ba9\u4ea7\u54c1\u7ecf\u7406\u51993\u5929PRD\u4e86
        \n3.2 \u4ee3\u7801\u751f\u6210：C<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[9645,9646,9647,9648,597],"topic":[],"class_list":["post-85813","post","type-post","status-publish","format-standard","hentry","category-server","tag-ai-devops","tag-mlops","tag-9647","tag-9648","tag-ci-cd"],"yoast_head":"\nAI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/85813.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u201c\u8001\u677f，\u6211\u4eec\u7684DevOps\u5e73\u53f0\u63a5\u5165\u4e86\u5927\u6a21\u578b\u2014\u2014\u73b0\u5728\u4e0d\u4ec5\u4ee3\u7801\u662fAI\u5199\u7684，\u8fde\u9700\u6c42\u6587\u6863\u3001\u6d4b\u8bd5\u7528\u4f8b\u3001\u90e8\u7f72\u811a\u672c\u3001\u8fd0\u7ef4\u544a\u8b66\u90fd\u662fAI\u5728\u641e，\u4f60\u8bf4\u6211\u4eec\u7a0b\u5e8f\u5458\u8fd8\u80fd\u5e72\u5565？\u201d \u201c\u2026\u2026\u6478\u9c7c\u3002\u201d \u76ee\u5f55 \u4e00\u3001\u5f00\u7bc7\u66b4\u51fb：AI DevOps\u5230\u5e95\u662f\u4e0d\u662f\u753b\u997c？ \u4e8c\u3001\u4e00\u5f20\u56fe\u770b\u7a7fAI DevOps\u5168\u666f \u4e09\u3001\u5404\u73af\u8282AI\u5207\u5165\u5b9e\u6218 3.1 \u9700\u6c42\u2192\u7528\u6237\u6545\u4e8b：\u522b\u8ba9\u4ea7\u54c1\u7ecf\u7406\u51993\u5929PRD\u4e86 3.2 \u4ee3\u7801\u751f\u6210：C\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/85813.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-07-27T14:42:49+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"14 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/85813.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/85813.html\",\"name\":\"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-07-27T14:42:49+00:00\",\"dateModified\":\"2026-07-27T14:42:49+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/85813.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/85813.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/85813.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/85813.html","og_locale":"zh_CN","og_type":"article","og_title":"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u201c\u8001\u677f，\u6211\u4eec\u7684DevOps\u5e73\u53f0\u63a5\u5165\u4e86\u5927\u6a21\u578b\u2014\u2014\u73b0\u5728\u4e0d\u4ec5\u4ee3\u7801\u662fAI\u5199\u7684，\u8fde\u9700\u6c42\u6587\u6863\u3001\u6d4b\u8bd5\u7528\u4f8b\u3001\u90e8\u7f72\u811a\u672c\u3001\u8fd0\u7ef4\u544a\u8b66\u90fd\u662fAI\u5728\u641e，\u4f60\u8bf4\u6211\u4eec\u7a0b\u5e8f\u5458\u8fd8\u80fd\u5e72\u5565？\u201d \u201c\u2026\u2026\u6478\u9c7c\u3002\u201d \u76ee\u5f55 \u4e00\u3001\u5f00\u7bc7\u66b4\u51fb：AI DevOps\u5230\u5e95\u662f\u4e0d\u662f\u753b\u997c？ \u4e8c\u3001\u4e00\u5f20\u56fe\u770b\u7a7fAI DevOps\u5168\u666f \u4e09\u3001\u5404\u73af\u8282AI\u5207\u5165\u5b9e\u6218 3.1 \u9700\u6c42\u2192\u7528\u6237\u6545\u4e8b：\u522b\u8ba9\u4ea7\u54c1\u7ecf\u7406\u51993\u5929PRD\u4e86 3.2 \u4ee3\u7801\u751f\u6210：C","og_url":"https:\/\/www.wsisp.com\/helps\/85813.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-07-27T14:42:49+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"14 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/85813.html","url":"https:\/\/www.wsisp.com\/helps\/85813.html","name":"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-07-27T14:42:49+00:00","dateModified":"2026-07-27T14:42:49+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/85813.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/85813.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/85813.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"AI\u9879\u76ee\u4ece\u5165\u95e8\u5230\u4e0a\u7ebf21-\u4ece\u9700\u6c42\u5206\u6790\u5230\u90e8\u7f72\u8fd0\u7ef4\u5168\u8ba9AI\u5e72\uff1fAI\u9a71\u52a8\u7684DevOps\u5e73\u53f0\u67b6\u6784\u63ed\u79d8"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/85813","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=85813"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/85813\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=85813"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=85813"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=85813"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=85813"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}