如何在 GitHub Actions 中使用 Claude Code (2026)

Claude Code 是 Anthropic 推出的 CLI 工具，可将 Claude 直接引入您的终端和代码库。通过将其集成到 GitHub Actions 中，您可以实现 AI 驱动的代码审查自动化、生成 PR 摘要、运行自动重构，并为 CI/CD 流水线添加智能质量检测。本指南将逐步介绍如何进行配置。

Claude Code 在 CI/CD 中能做什么？

以下是 Claude Code 在 GitHub Actions 中最实用的场景：

使用场景	描述	触发条件
PR 代码审查	提供改进建议的自动化审查	Pull request 已创建/已更新
PR 摘要	根据差异（diff）生成描述	Pull request 已创建
缺陷检测	扫描修改后的文件以查找常见 bug	Pull request, push
文档编写	为新函数自动生成文档	Pull request
测试生成	为变更的代码创建测试用例	Pull request
迁移辅助	针对依赖项变更提供更新建议	定时任务, 手动触发

前提条件

一个 GitHub 仓库
Anthropic API key（在 console.anthropic.com 获取）
对 GitHub Actions YAML 语法有基本了解

第 1 步：将 API Key 存储为 Secret

切勿在工作流文件中硬编码 API Key。请将您的 Anthropic API key 添加为 GitHub 仓库的 secret：

1. 进入 GitHub 上的仓库
2. 点击 Settings > Secrets and variables > Actions
3. 点击 "New repository secret"
4. 名称 (Name): ANTHROPIC_API_KEY
5. 值 (Value): sk-ant-xxxxx (您的 key)
6. 点击 "Add secret"

第 2 步：基础 Claude Code 工作流

在 .github/workflows/claude-review.yml 处创建一个工作流文件：

name: Claude Code Review

on:
  pull_request:
    types: [opened, synchronize]

permissions:
  contents: read
  pull-requests: write

jobs:
  claude-review:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4
        with:
          fetch-depth: 0  # 获取完整历史记录以进行 diff

      - name: Setup Node.js
        uses: actions/setup-node@v4
        with:
          node-version: "20"

      - name: Install Claude Code
        run: npm install -g @anthropic-ai/claude-code

      - name: Get PR diff
        id: diff
        run: |
          git diff origin/${{ github.base_ref }}...HEAD > pr_diff.txt
          echo "diff_size=$(wc -c < pr_diff.txt)" >> $GITHUB_OUTPUT

      - name: Run Claude Code Review
        if: steps.diff.outputs.diff_size > 0
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
        run: |
          claude -p --output-format text "$(cat <<'PROMPT'
          Review the following pull request diff. Focus on:
          1. Potential bugs or logic errors
          2. Security concerns
          3. Performance issues
          4. Code style and readability

          Be concise. Only mention real issues, not style nitpicks.
          Format your response as a markdown list.

          Diff:
          $(cat pr_diff.txt)
          PROMPT
          )" > review_output.md

      - name: Post review comment
        if: steps.diff.outputs.diff_size > 0
        uses: actions/github-script@v7
        with:
          script: |
            const fs = require('fs');
            const review = fs.readFileSync('review_output.md', 'utf8');

            await github.rest.issues.createComment({
              owner: context.repo.owner,
              repo: context.repo.repo,
              issue_number: context.issue.number,
              body: `## Claude Code Review\n\n${review}\n\n---\n*Automated review by Claude Code*`
            });

第 3 步：PR 摘要生成器

根据代码变更自动生成 PR 描述：

name: Claude PR Summary

on:
  pull_request:
    types: [opened]

permissions:
  contents: read
  pull-requests: write

jobs:
  generate-summary:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4
        with:
          fetch-depth: 0

      - name: Setup Node.js
        uses: actions/setup-node@v4
        with:
          node-version: "20"

      - name: Install Claude Code
        run: npm install -g @anthropic-ai/claude-code

      - name: Generate PR summary
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
        run: |
          # 获取 diff
          DIFF=$(git diff origin/${{ github.base_ref }}...HEAD)

          # 获取提交信息
          COMMITS=$(git log origin/${{ github.base_ref }}...HEAD --oneline)

          claude -p --output-format text "$(cat <<PROMPT
          Generate a pull request summary based on these changes.

          Format:
          ## Summary
          (2-3 sentences describing what this PR does)

          ## Changes
          (Bulleted list of key changes)

          ## Testing
          (Suggested testing steps)

          Commits:
          $COMMITS

          Diff:
          $DIFF
          PROMPT
          )" > summary.md

      - name: Update PR description
        uses: actions/github-script@v7
        with:
          script: |
            const fs = require('fs');
            const summary = fs.readFileSync('summary.md', 'utf8');

            await github.rest.pulls.update({
              owner: context.repo.owner,
              repo: context.repo.repo,
              pull_number: context.issue.number,
              body: summary
            });

第 4 步：自动化测试生成

为新文件或修改后的文件生成测试：

name: Claude Test Generator

on:
  pull_request:
    types: [opened, synchronize]
    paths:
      - "src/**/*.ts"
      - "src/**/*.js"
      - "!src/**/*.test.*"
      - "!src/**/*.spec.*"

permissions:
  contents: write
  pull-requests: write

jobs:
  generate-tests:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4
        with:
          fetch-depth: 0
          ref: ${{ github.head_ref }}

      - name: Setup Node.js
        uses: actions/setup-node@v4
        with:
          node-version: "20"

      - name: Install dependencies
        run: |
          npm ci
          npm install -g @anthropic-ai/claude-code

      - name: Find changed files without tests
        id: changed
        run: |
          FILES=$(git diff --name-only origin/${{ github.base_ref }}...HEAD \
            | grep -E '^src/.*\.(ts|js)$' \
            | grep -v -E '\.(test|spec)\.' \
            | head -5)
          echo "files=$FILES" >> $GITHUB_OUTPUT

      - name: Generate tests with Claude
        if: steps.changed.outputs.files != ''
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
        run: |
          for FILE in ${{ steps.changed.outputs.files }}; do
            TEST_FILE="${FILE%.ts}.test.ts"
            if [ ! -f "$TEST_FILE" ]; then
              echo "Generating tests for $FILE..."
              claude -p --output-format text \
                "Read the file $FILE and generate comprehensive unit tests for it. \
                 Use the existing test framework in this project. \
                 Output only the test file content, no explanations." \
                > "$TEST_FILE"
            fi
          done

      - name: Commit generated tests
        run: |
          git config user.name "claude-code[bot]"
          git config user.email "claude-code[bot]@users.noreply.github.com"
          git add "src/**/*.test.ts"
          git diff --staged --quiet || git commit -m "test: add AI-generated tests for changed files"
          git push

第 5 步：安全扫描工作流

使用 Claude 识别代码变更中的安全问题：

name: Claude Security Scan

on:
  pull_request:
    types: [opened, synchronize]

permissions:
  contents: read
  pull-requests: write

jobs:
  security-scan:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4
        with:
          fetch-depth: 0

      - name: Setup Node.js
        uses: actions/setup-node@v4
        with:
          node-version: "20"

      - name: Install Claude Code
        run: npm install -g @anthropic-ai/claude-code

      - name: Run security analysis
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
        run: |
          DIFF=$(git diff origin/${{ github.base_ref }}...HEAD)

          claude -p --output-format text "$(cat <<PROMPT
          Analyze this code diff for security vulnerabilities.
          Check for:
          - SQL injection
          - XSS vulnerabilities
          - Hardcoded secrets or credentials
          - Insecure deserialization
          - Path traversal
          - Missing input validation
          - Insecure dependencies

          If no issues found, respond with "No security issues detected."
          If issues found, list each with severity (HIGH/MEDIUM/LOW),
          the file and line, and a suggested fix.

          Diff:
          $DIFF
          PROMPT
          )" > security_report.md

      - name: Post security report
        uses: actions/github-script@v7
        with:
          script: |
            const fs = require('fs');
            const report = fs.readFileSync('security_report.md', 'utf8');

            if (!report.includes('No security issues detected')) {
              await github.rest.issues.createComment({
                owner: context.repo.owner,
                repo: context.repo.repo,
                issue_number: context.issue.number,
                body: `## Security Scan Results\n\n${report}\n\n---\n*Scanned by Claude Code*`
              });
            }

成本管理

在 CI/CD 中使用 Claude Code 可能会快速消耗 API 额度。以下是控制成本的策略：

策略	实现方式
限制 Diff 大小	截断超过 10,000 行的 diff
跳过草稿 PR	添加 `if: github.event.pull_request.draft == false`
按文件类型过滤	在触发器中使用 `paths` 过滤器
使用缓存	缓存 Claude Code 的 npm 安装过程
速率限制	仅在最后一次 commit 时运行，而非每次 push
使用 `--max-turns 1`	防止 Agent 进入多轮对话循环

Diff 大小防护门限制

- name: Check diff size
  id: size_check
  run: |
    DIFF_LINES=$(git diff origin/${{ github.base_ref }}...HEAD | wc -l)
    if [ "$DIFF_LINES" -gt 5000 ]; then
      echo "skip=true" >> $GITHUB_OUTPUT
      echo "Diff too large ($DIFF_LINES lines), skipping AI review"
    else
      echo "skip=false" >> $GITHUB_OUTPUT
    fi

- name: Run Claude review
  if: steps.size_check.outputs.skip == 'false'
  # ... step 的其余部分

故障排除

问题	原因	解决方案
"API key not found"	Secret 未配置	将 `ANTHROPIC_API_KEY` 添加到仓库 secrets
工作流超时	Diff 过大或响应缓慢	在 job 中添加 `timeout-minutes: 10`
审查输出为空	Diff 超过上下文长度	截断 diff 或仅审查特定文件
PR 评论权限被拒绝	缺少权限配置	在 permissions 中添加 `pull-requests: write`
找不到 Claude Code	安装失败	固定 npm 版本：`npm install -g @anthropic-ai/claude-code@latest`

最佳实践

设置超时。 添加 timeout-minutes: 10 以防止无休止的工作流运行。
使用 --print 模式。 在 CI 中务必使用 claude -p 以避免交互式提示。
限制上下文。 仅发送 diff 和相关文件，不要发送整个代码库。
从容处理失败。 使用 continue-on-error: true，确保 AI 审查不会阻塞代码合并。
审视 AI 输出。 将 Claude 的评论视为建议，而非权威裁定。
固定版本。 固定 Claude Code 和 action 的版本以确保构建的可复现性。

结论

将 Claude Code 集成到 GitHub Actions 中可以为您的 CI/CD 流水线增添一个智能层。无论您将其用于代码审查、PR 摘要、测试生成还是安全扫描，其配置过程都非常直接：安装 Claude Code、传递 diff 并发布结果。关键在于通过过滤触发器和限制 diff 大小来管理成本。

对于正在构建 AI 生成媒体（视频、图像或数智人）应用程序的开发团队，Hypereal AI 提供了生产级的 API，可自然地集成到 CI/CD 流水线中，实现自动化的资源生成和测试。