Back to Details

cekura-coordinator

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Cekura Coordinator

Cekura协调器

Purpose

用途

Route users to the right Cekura skill or command based on what they need. This is the "front desk" — it knows everything available across all Cekura plugins and helps users find the right tool.
根据用户需求将其引导至合适的Cekura技能或命令。它就像“前台”——了解所有Cekura插件中的可用功能，并帮助用户找到合适的工具。

Performing Platform Actions

执行平台操作

When this skill suggests creating, listing, updating, or evaluating something on Cekura, prefer using available platform tools over describing API calls or dashboard steps. In Claude Code with the Cekura plugin installed, these tools are auto-configured and handle authentication, parameter validation, and error handling for you. Fall back to direct API endpoints or dashboard guidance only when no tools are available in the current session.
当该技能建议在Cekura上创建、列出、更新或评估内容时，优先使用可用的平台工具，而非描述API调用或控制台步骤。在安装了Cekura插件的Claude Code中，这些工具已自动配置，可为你处理身份验证、参数验证和错误处理。仅当当前会话中没有可用工具时，才回退到直接使用API端点或控制台指引。

When This Skill Loads

该技能加载的场景

  • User just installed cekura-skills and asks "what can I do?"
  • User asks for help or doesn't know which command to use
  • User describes a task and you need to route them
  • 用户刚安装cekura-skills并询问“我能做什么？”
  • 用户寻求帮助或不知道该使用哪个命令
  • 用户描述了一项任务，你需要将其引导至合适的工具

Available Skills & Commands

可用技能与命令

Skills activate when the user describes a relevant task. Commands are slash commands available only in Claude Code with the plugin installed.
当用户描述相关任务时，技能会激活。命令是仅在安装了插件的Claude Code中可用的斜杠命令。

Skills

技能

SkillUse when
cekura-onboarding
New to Cekura, first-time setup, platform walkthrough
cekura-create-agent
Set up a voice AI agent — provider, mock tools, KB, dynamic vars
cekura-self-improving-agent
Auto-tune agent prompts from eval results — diagnose → propose → apply → re-validate
cekura-metric-design
Design new metrics, improve existing ones, metric best practices
cekura-metric-improvement
Improve metric accuracy through feedback cycle (labs workflow)
cekura-predefined-metrics
Catalog of built-in metrics — what each does, costs, constraints, configuration
cekura-eval-design
Design evaluators, test suites, coverage strategy, conditional actions
cekura-fixing-prod-issues
Debug a failing production call, reproduce with evals, fix, regression test, raise PR
技能使用场景
cekura-onboarding
首次使用Cekura、初始设置、平台导览
cekura-create-agent
设置语音AI Agent——提供商、模拟工具、知识库（KB）、动态变量
cekura-self-improving-agent
根据评估结果自动优化Agent提示词——诊断→提议→应用→重新验证
cekura-metric-design
设计新指标、改进现有指标、指标最佳实践
cekura-metric-improvement
通过反馈循环提升指标准确性（实验室工作流）
cekura-predefined-metrics
内置指标目录——每个指标的功能、成本、约束、配置
cekura-eval-design
设计评估器、测试套件、覆盖策略、条件操作
cekura-fixing-prod-issues
调试失败的生产调用、通过评估重现问题、修复、回归测试、提交PR

Commands

命令

CommandUse when
/cekura-onboarding
Run the guided onboarding flow (state-aware, picks up where you left off)
/setup-mcp
Configure the Cekura MCP server (Claude Code only)
/upgrade-skills
Update all Cekura skills to the latest version
/report-bug
Report a bug — files a GitHub issue, optionally attempts a fix
/create-metric
Create or update a metric via API
/list-metrics
List metrics for an agent or project
/evaluate-calls
Run metrics on specific calls
/improve-metric
Full improvement cycle: collect feedback, run labs, auto-improve
/manual-create-update-eval
Create or update a single evaluator with full field walkthrough
/autogen-eval
Auto-generate evaluators (or bulk create from CSV/JSON)
/list-evals
List evaluators for an agent or project
/run-evals
Execute evaluators (run test scenarios)
/eval-results
Check results from a test run
/cekura-report
Full end-to-end quality report — generates 10 evals, runs them, produces structured analysis
命令使用场景
/cekura-onboarding
运行引导式入门流程（感知状态，可从上次中断处继续）
/setup-mcp
配置Cekura MCP服务器（仅Claude Code可用）
/upgrade-skills
将所有Cekura技能更新至最新版本
/report-bug
报告Bug——提交GitHub Issue，可选尝试修复
/create-metric
通过API创建或更新指标
/list-metrics
列出Agent或项目的指标
/evaluate-calls
针对特定调用运行指标评估
/improve-metric
完整改进周期：收集反馈、运行实验室测试、自动优化
/manual-create-update-eval
创建或更新单个评估器，包含完整字段引导
/autogen-eval
自动生成评估器（或从CSV/JSON批量创建）
/list-evals
列出Agent或项目的评估器
/run-evals
执行评估器（运行测试场景）
/eval-results
查看测试运行的结果
/cekura-report
完整端到端质量报告——生成10个评估器、运行评估、生成结构化分析

Routing Guide

路由指引

When the user describes what they need, route them:
User NeedRoute To
"I'm new to Cekura" / first-time setupcekura-onboarding skill
"Set up my agent" / "connect my voice agent"cekura-create-agent skill
"Configure MCP" / "MCP not working"
/setup-mcp
command
"Create metrics for my agent"cekura-metric-design skill
"My metrics are giving wrong results"
/improve-metric
command (or cekura-metric-improvement skill for full cycle)
"I need to test my agent"cekura-eval-design skill
"Generate test scenarios"
/autogen-eval
command
"Create a specific test scenario"
/manual-create-update-eval
command
"Run my tests"
/run-evals
command
"Check test results"
/eval-results
command
"Create a metric that checks X"
/create-metric
command (or cekura-metric-design skill for complex metrics)
"Update this metric"
/create-metric
command (handles both create and update)
"Evaluate calls against metrics"
/evaluate-calls
command
"Update my skills"
/upgrade-skills
command
"What metrics should I have?"cekura-metric-design skill (baseline metrics section)
"Help me improve this metric"
/improve-metric
command
"Leave feedback on a metric result"
/improve-metric
command (Phase 1: feedback collection)
"Set up production monitoring"cekura-onboarding skill (Phase 6) + observability docs
"Add mock tools" / "set up tools"cekura-create-agent skill (Phase 4)
"Upload knowledge base"cekura-create-agent skill (Phase 5)
"Something's broken" / "file a bug"
/report-bug
command
"Improve my agent" / "auto-tune from eval results"cekura-self-improving-agent skill
"Which built-in metrics are available?" / "what does Hallucination Detection cost?"cekura-predefined-metrics skill
"Fix this prod call bug" / "reproduce and test a fix"cekura-fixing-prod-issues skill
"Run a full quality report" / "generate evals and run them end-to-end"
/cekura-report
command
当用户描述需求时，将其引导至：
用户需求引导至
“我是Cekura新手” / 首次设置cekura-onboarding技能
“设置我的Agent” / “连接我的语音Agent”cekura-create-agent技能
“配置MCP” / “MCP无法工作”
/setup-mcp
命令
“为我的Agent创建指标”cekura-metric-design技能
“我的指标结果不正确”
/improve-metric
命令（或完整周期使用cekura-metric-improvement技能）
“我需要测试我的Agent”cekura-eval-design技能
“生成测试场景”
/autogen-eval
命令
“创建特定测试场景”
/manual-create-update-eval
命令
“运行我的测试”
/run-evals
命令
“查看测试结果”
/eval-results
命令
“创建一个检查X的指标”
/create-metric
命令（复杂指标使用cekura-metric-design技能）
“更新这个指标”
/create-metric
命令（同时处理创建和更新）
“针对指标评估调用”
/evaluate-calls
命令
“更新我的技能”
/upgrade-skills
命令
“我应该使用哪些指标？”cekura-metric-design技能（基线指标部分）
“帮我改进这个指标”
/improve-metric
命令
“对指标结果提供反馈”
/improve-metric
命令（第一阶段：反馈收集）
“设置生产监控”cekura-onboarding技能（第六阶段） + 可观测性文档
“添加模拟工具” / “设置工具”cekura-create-agent技能（第四阶段）
“上传知识库”cekura-create-agent技能（第五阶段）
“某个功能坏了” / “提交Bug”
/report-bug
命令
“改进我的Agent” / “根据评估结果自动优化”cekura-self-improving-agent技能
“有哪些可用的内置指标？” / “幻觉检测的成本是多少？”cekura-predefined-metrics技能
“修复这个生产调用Bug” / “重现并测试修复方案”cekura-fixing-prod-issues技能
“运行完整质量报告” / “生成评估器并端到端运行”
/cekura-report
命令

Typical User Journeys

典型用户旅程

Journey 1: Brand New User

旅程1：全新用户

  1. /setup-mcp
    → Configure MCP server for API access (Claude Code plugin users)
  2. cekura-onboarding → Set up account and project
  3. cekura-create-agent → Add agent with provider, mock tools, KB, dynamic vars
  4. cekura-onboarding → Enable pre-defined metrics, generate first evaluators
  5. cekura-onboarding → Run first tests, review results
  6. cekura-metric-design → Create custom metrics based on what they learned
  7. cekura-eval-design → Build targeted test suites
  1. /setup-mcp
    → 配置MCP服务器以实现API访问（Claude Code插件用户）
  2. cekura-onboarding → 设置账户和项目
  3. cekura-create-agent → 添加包含提供商、模拟工具、知识库、动态变量的Agent
  4. cekura-onboarding → 启用预定义指标，生成首个评估器
  5. cekura-onboarding → 运行首次测试，查看结果
  6. cekura-metric-design → 根据所学内容创建自定义指标
  7. cekura-eval-design → 构建针对性测试套件

Journey 2: Has Agent, Needs Testing

旅程2：已有Agent，需要测试

  1. cekura-eval-design → Design test suite
  2. /autogen-eval
    → Auto-generate evaluators
  3. /run-evals
    → Execute tests
  4. /eval-results
    → Review results
  1. cekura-eval-design → 设计测试套件
  2. /autogen-eval
    → 自动生成评估器
  3. /run-evals
    → 执行测试
  4. /eval-results
    → 查看结果

Journey 3: Has Metrics, Needs Improvement

旅程3：已有指标，需要改进

  1. /improve-metric
    → Full cycle: collect feedback, check readiness, auto-improve
  2. /evaluate-calls
    → Validate changes
  1. /improve-metric
    → 完整周期：收集反馈、检查就绪状态、自动优化
  2. /evaluate-calls
    → 验证更改

Journey 4: Production Monitoring

旅程4：生产监控

  1. cekura-metric-design → Design observability metrics
  2. /create-metric
    → Deploy metrics (create or update)
  3. /evaluate-calls
    → Validate on sample calls
  1. cekura-metric-design → 设计可观测性指标
  2. /create-metric
    → 部署指标（创建或更新）
  3. /evaluate-calls
    → 在示例调用上验证

API Access

API访问

For Claude Code plugin users: each plugin auto-configures access to the Cekura API. If commands or platform operations aren't working, run 
/setup-mcp
to configure the connection.
For other clients (Cursor, Codex, npx skills installs, etc.): use the Cekura dashboard at https://dashboard.cekura.ai or call the API directly using your API key.
对于Claude Code插件用户：每个插件都会自动配置对Cekura API的访问权限。如果命令或平台操作无法正常工作，请运行
/setup-mcp
来配置连接。
对于其他客户端（Cursor、Codex、npx skills安装等）：使用Cekura控制台https://dashboard.cekura.ai，或使用你的API密钥直接调用API。

Next Steps

下一步

This skill routes — it doesn't perform tasks itself. After confirming the user's need, invoke the matching skill:
  • New to Cekura → cekura-onboarding
  • Connecting an agent → cekura-create-agent
  • Auto-tuning an agent prompt from eval results → cekura-self-improving-agent
  • Designing metrics → cekura-metric-design
  • Improving metric accuracy → cekura-metric-improvement
  • Picking which built-in metrics to use → cekura-predefined-metrics
  • Designing test scenarios → cekura-eval-design
  • Fixing a production call bug end-to-end → cekura-fixing-prod-issues
该技能仅负责路由——不执行任务本身。确认用户需求后，调用匹配的技能：
  • 首次使用Cekura → cekura-onboarding
  • 连接Agent → cekura-create-agent
  • 根据评估结果自动优化Agent提示词 → cekura-self-improving-agent
  • 设计指标 → cekura-metric-design
  • 提升指标准确性 → cekura-metric-improvement
  • 选择内置指标 → cekura-predefined-metrics
  • 设计测试场景 → cekura-eval-design
  • 端到端修复生产调用Bug → cekura-fixing-prod-issues

Documentation

文档