Back to Details

conduktor

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Conduktor Platform

Conduktor平台

Conduktor is three products that work together around Apache Kafka:
  • Console (
    conduktor/conduktor-console
    , port 8080) observes and manages Kafka clusters. It provides RBAC, a topic catalog, monitoring, a self-service framework, and data quality policies. Requires PostgreSQL.
  • Gateway (
    conduktor/conduktor-gateway
    , port 6969) is a transparent Kafka proxy that intercepts and modifies requests and responses. It provides interceptors, virtual clusters, field-level encryption, data masking, and traffic control. Clients connect to Gateway instead of Kafka directly.
  • CLI (
    conduktor
    binary) is a kubectl-style tool that manages both Console and Gateway resources declaratively via YAML apply/get/delete.
Console and Gateway are separate services. Console can manage Gateway, but they deploy independently.
Conduktor是围绕Apache Kafka协同工作的三款产品：
  • Console (
    conduktor/conduktor-console
    , 端口 8080) 用于观测和管理Kafka集群。它提供RBAC、主题目录、监控、自助服务框架和数据质量策略。需要PostgreSQL支持。
  • Gateway (
    conduktor/conduktor-gateway
    , 端口 6969) 是一个透明的Kafka代理，可拦截和修改请求与响应。它提供拦截器、虚拟集群、字段级加密、数据脱敏和流量控制能力。客户端直接连接到Gateway而非Kafka。
  • CLI (
    conduktor
    二进制文件) 是类kubectl的工具，可通过YAML的apply/get/delete命令声明式管理Console和Gateway资源。
Console和Gateway是独立的服务。Console可以管理Gateway，但二者独立部署。

Always read first

必读内容

For any Conduktor question, load references/mental-model.md to understand product boundaries, the Gateway resource model, virtual clusters, and the self-service framework.
For things AI assistants commonly get wrong, see references/anti-patterns.md.
For Conduktor-to-Kafka terminology mapping, see references/terminology.md.
如有任何Conduktor相关问题，请先查阅 references/mental-model.md 以了解产品边界、Gateway资源模型、虚拟集群和自助服务框架。
如需了解AI助手常出错的内容，请查看 references/anti-patterns.md。
如需查看Conduktor与Kafka的术语映射关系，请参阅 references/terminology.md。

Use cases

用例

Platform engineer

平台工程师

I want to...Read
Deploy Console and Gateway (Docker, Helm, Kubernetes)use-cases/platform/deploy-conduktor.md
Encrypt or mask Kafka data (field-level, payload)use-cases/platform/encrypt-kafka-data.md
Enforce data quality rules (CEL, JSON Schema)use-cases/platform/enforce-data-quality.md
Set up multi-tenancy (virtual clusters, ACLs, service accounts)use-cases/platform/multi-tenancy.md
Apply traffic control and safeguards (rate limits, topic policies)use-cases/platform/traffic-control.md
Automate with CLI and GitOps (apply, CI/CD, state management)use-cases/platform/gitops-automation.md
Manage infrastructure as code with Terraformuse-cases/platform/terraform.md
我想要...参考文档
部署Console和Gateway（Docker、Helm、Kubernetes）use-cases/platform/deploy-conduktor.md
加密或脱敏Kafka数据（字段级、payload）use-cases/platform/encrypt-kafka-data.md
执行数据质量规则（CEL、JSON Schema）use-cases/platform/enforce-data-quality.md
设置多租户（虚拟集群、ACL、服务账号）use-cases/platform/multi-tenancy.md
应用流量控制和安全防护（速率限制、主题策略）use-cases/platform/traffic-control.md
通过CLI和GitOps实现自动化（apply、CI/CD、状态管理）use-cases/platform/gitops-automation.md
通过Terraform实现基础设施即代码管理use-cases/platform/terraform.md

Application developer

应用开发者

I want to...Read
Get started with Kafka through Conduktor (connect, discover topics)use-cases/app-developer/onboard-to-kafka.md
Create a topic through self-serviceuse-cases/app-developer/create-topic.md
Produce and consume through Gateway (client configs, schemas)use-cases/app-developer/produce-consume.md
Request access to another team's topicuse-cases/app-developer/request-access.md
我想要...参考文档
通过Conduktor入门Kafka（连接、发现主题）use-cases/app-developer/onboard-to-kafka.md
通过自助服务创建主题use-cases/app-developer/create-topic.md
通过Gateway生产和消费数据（客户端配置、Schema）use-cases/app-developer/produce-consume.md
申请访问其他团队的主题use-cases/app-developer/request-access.md

Agent behavior

Agent行为规范

IMPORTANT: Do NOT call any documentation MCP tool before completing steps 1-2 below. These skill files + CLI discovery contain everything you need. MCP docs are a last resort (step 5).
When a user asks about Conduktor, do not just explain how things work. Be an active assistant:
  1. Discover first with CLI — read the matching use-case file, then run the exact 
    conduktor get
    commands from its "Agent workflow" section. The CLI gives you real state; docs give you generic examples.
  2. Ask with options — use discovery results to offer concrete choices instead of open-ended questions. If an 
    AskUserQuestion
    tool is available, use it with predefined options (e.g. topic name suggestions, partition counts, retention presets). Ask one question at a time, not a list of numbered questions. Prefer sensible defaults — only ask when the choice genuinely matters.
  3. Generate ready-to-use output — produce complete YAML, HCL, or client configs with real names from discovery. Never give templates with placeholders when you can fill in real values.
  4. Execute with confirmation — offer to run 
    conduktor apply -f --dry-run
    first, then 
    conduktor apply -f
    on approval. For Terraform, offer 
    terraform plan
    then 
    terraform apply
    .
  5. Look up what you don't know — only after checking skill files AND CLI discovery, if the answer is still missing, check your available tools for a 
    search_conduktor_documentation
    tool (Conduktor's MCP docs server). This is a fallback, not a first step. Never invent env var names, config fields, or API endpoints that are not in these files. If the MCP tool is not available, tell the user to add it with this config:
    json
    { "mcpServers": { "conduktor-docs": { "type": "url", "url": "https://docs.conduktor.io/mcp" } } }
The CLI requires auth. If commands fail with 401/connection errors, help the user configure 
CDK_BASE_URL
+ 
CDK_API_KEY
(Console) or 
CDK_GATEWAY_BASE_URL
+ 
CDK_GATEWAY_USER
/
CDK_GATEWAY_PASSWORD
(Gateway).
重要提示：完成以下1-2步之前，请勿调用任何文档MCP工具。这些技能文件 + CLI发现功能已涵盖你所需的全部内容。MCP文档是最后的备选方案（第5步）。
当用户询问Conduktor相关问题时，不要仅解释工作原理，要做主动的助手：
  1. 优先通过CLI发现信息 —— 阅读匹配的用例文件，然后运行其「Agent工作流」部分给出的准确
    conduktor get
    命令。CLI可为你提供真实状态，文档仅提供通用示例。
  2. 提供选项式询问 —— 使用发现结果提供具体选择，而非开放式问题。如果有
    AskUserQuestion
    工具可用，请搭配预定义选项使用（例如主题名称建议、分区数量、保留期预设）。一次只问一个问题，不要给出编号问题列表。优先使用合理默认值 —— 仅当选择确实会产生影响时才询问用户。
  3. 生成可直接使用的输出 —— 结合发现的真实名称生成完整的YAML、HCL或客户端配置。如果可以填充真实值，绝对不要提供带占位符的模板。
  4. 确认后再执行 —— 主动提议先运行
    conduktor apply -f --dry-run
    ，获得批准后再执行
    conduktor apply -f
    。如果是Terraform场景，提议先运行
    terraform plan
    再执行
    terraform apply
    。
  5. 查询未知内容 —— 仅当检查完技能文件和CLI发现结果后仍找不到答案时，再检查可用工具中是否有
    search_conduktor_documentation
    工具（Conduktor的MCP文档服务器）。这是降级方案，不是第一步操作。绝对不要编造这些文件中没有的环境变量名、配置字段或API端点。如果没有MCP工具，告知用户可通过以下配置添加：
    json
    { "mcpServers": { "conduktor-docs": { "type": "url", "url": "https://docs.conduktor.io/mcp" } } }
CLI需要身份认证。如果命令执行失败返回401/连接错误，帮助用户配置
CDK_BASE_URL
+ 
CDK_API_KEY
（Console）或
CDK_GATEWAY_BASE_URL
+ 
CDK_GATEWAY_USER
/
CDK_GATEWAY_PASSWORD
（Gateway）。

Intent routing

意图路由

When a user mentions these keywords, load the corresponding file:
  • encrypt, mask, PII, GDPR, shield -> 
    use-cases/platform/encrypt-kafka-data.md
  • data quality, CEL, validate, enforce schema -> 
    use-cases/platform/enforce-data-quality.md
  • virtual cluster, tenant, isolation, team namespace -> 
    use-cases/platform/multi-tenancy.md
  • rate limit, throttle, safeguard, quota, traffic -> 
    use-cases/platform/traffic-control.md
  • deploy, Docker, Helm, Kubernetes, install -> 
    use-cases/platform/deploy-conduktor.md
  • conduktor CLI, apply, GitOps, CI/CD, pipeline, automation -> 
    use-cases/platform/gitops-automation.md
  • Terraform, IaC, HCL, provider -> 
    use-cases/platform/terraform.md
  • onboard, connect, credentials, getting started, bootstrap -> 
    use-cases/app-developer/onboard-to-kafka.md
  • create topic, new topic, self-service topic -> 
    use-cases/app-developer/create-topic.md
  • produce, consume, schema registry, consumer group -> 
    use-cases/app-developer/produce-consume.md
  • access, permission, request, share topic -> 
    use-cases/app-developer/request-access.md
当用户提到以下关键词时，加载对应的文件：
  • encrypt, mask, PII, GDPR, shield -> 
    use-cases/platform/encrypt-kafka-data.md
  • data quality, CEL, validate, enforce schema -> 
    use-cases/platform/enforce-data-quality.md
  • virtual cluster, tenant, isolation, team namespace -> 
    use-cases/platform/multi-tenancy.md
  • rate limit, throttle, safeguard, quota, traffic -> 
    use-cases/platform/traffic-control.md
  • deploy, Docker, Helm, Kubernetes, install -> 
    use-cases/platform/deploy-conduktor.md
  • conduktor CLI, apply, GitOps, CI/CD, pipeline, automation -> 
    use-cases/platform/gitops-automation.md
  • Terraform, IaC, HCL, provider -> 
    use-cases/platform/terraform.md
  • onboard, connect, credentials, getting started, bootstrap -> 
    use-cases/app-developer/onboard-to-kafka.md
  • create topic, new topic, self-service topic -> 
    use-cases/app-developer/create-topic.md
  • produce, consume, schema registry, consumer group -> 
    use-cases/app-developer/produce-consume.md
  • access, permission, request, share topic -> 
    use-cases/app-developer/request-access.md