sinzin91/awesome-sre-skills

# Awesome SRE 技能 [](https://awesome.re)

一份精心整理的关于 SRE、可观测性、监控和事件响应的 AI agent 技能清单。

入门指南 • 兼容性 • 技能 • 技能 vs MCP • 贡献指南

## 为什么有这个清单 像 [ClawHub](https://clawhub.com/)（5,700+ 个技能）和 awesome-openclaw-skills（3,000+ 个技能）这样的大型技能注册中心非常适合用于发现新事物，但也可能包含很多噪音。本清单专注于跨平台（Claude Code、OpenClaw 和 SkillMD）的 SRE、可观测性和事件响应技能。 ## 🚀 入门指南 ### Claude Code CLI ``` # 从 GitHub 安装 claude mcp add-skill https://github.com/OWNER/SKILL-REPO # 或者从本地目录安装 claude mcp add-skill /path/to/skill-directory ``` ### OpenClaw ``` # 将 skill 添加到你的 OpenClaw config 中 openclaw skill add https://github.com/OWNER/SKILL-REPO # 或者复制到你的 skills directory cp -r skill-folder ~/.openclaw/skills/ ``` ## 兼容性 | 平台 | 指令文件 | 图标 | | ----------- | ------------------------------- | ---- | | Claude Code | `CLAUDE.md` 或 `commands/*.md` | 🤖 | | OpenClaw | `SKILL.md` | 🦞 | | SkillMD | 跨平台包 | 📦 | 本清单中的大多数技能都适用于 Claude Code。请通过平台图标来检查兼容性。 **其他图标：** - 🎖️ 官方（由供应商维护） - ⭐ 热门（添加时已有 50+ GitHub star） ## 技能 ### APM 与 Tracing *调试应用性能，跨服务追踪请求，分析 AI agent 行为。* | Skill | Creator | Description | Platform | | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------- | ------------------------------------------------------------------------------------------- | -------- | | [Datadog Auto-Detector](https://github.com/schovi/claude-schovi/tree/main/plugins/schovi/skills/datadog-auto-detector) | @schovi | Auto-detects Datadog URLs/queries, spawns analyzer subagent for context | 📦 | | [LangSmith Fetch](https://github.com/OthmanAdi/langsmith-fetch-skill) | @OthmanAdi | Debug LangChain/LangGraph agents via LangSmith traces | 🤖 | | [Retell AI Observability](https://github.com/jeremylongshore/claude-code-plugins-plus-skills/tree/main/plugins/saas-packs/retellai-pack/skills/retellai-observability) | @jeremylongshore | Metrics, traces, alerts for Retell AI voice integrations | 📦 | | [Honeycomb Agent Skills](https://github.com/honeycombio/agent-skill) 🎖️ | @honeycombio | Official Honeycomb skills: querying, SLOs/triggers, OTel instrumentation, prod-debug agents | 🤖 | | [Langfuse Skills](https://github.com/langfuse/skills) 🎖️ ⭐ | @langfuse | Query/manage Langfuse traces, prompts, datasets & scores for LLM-app observability | 📦 | | [LangSmith Skills](https://github.com/langchain-ai/langsmith-skills) 🎖️ ⭐ | @langchain-ai | Query LLM traces, build eval datasets & custom evaluators in LangSmith | 📦 | | [Datadog Agent Skills](https://github.com/datadog-labs/agent-skills) 🎖️ ⭐ | Datadog | APM/trace analysis and LLM-app observability (RCA, eval, session tracing) | 🤖 | | [Elastic Agent Skills](https://github.com/elastic/agent-skills) 🎖️ ⭐ | Elastic | Observability skills: EDOT/OTel, APM, SLOs, Kubernetes investigation, ES|QL log search | 📦 | ### Logging 与日志分析 *搜索和分析日志，检测异常，跨系统关联事件。* | Skill | Creator | Description | Platform | | ----------------------------------------------------------------------------------------------------------------------- | ------------------- | ---------------------------------------------------------------------------------------- | -------- | | [Datadog CLI Skill](https://github.com/DataDog/pup) 🎖️ ⭐ | Datadog | Official Datadog `pup` CLI + skills: `/sre-investigate` for log/trace/metric correlation | 🤖 | | [Datadog Debug Workflow](https://github.com/schovi/claude-schovi) | @schovi | Debug command with Datadog analyzer for log/metric/trace correlation | 🤖 | | [Splunk Assistant Skills](https://github.com/grandcamel/splunk-assistant-skills) | @grandcamel | Splunk REST API automation, search queries, threat hunting | 🤖 | | [DevOps Monitor Commands](https://github.com/rohitg00/awesome-claude-code-toolkit/blob/main/commands/devops/monitor.md) | @rohitg00 | Monitoring-setup playbook: Prometheus/Grafana/Datadog config, alert rules, OTel tracing | 🤖 | | [Coralogix cx-cli Skills](https://github.com/coralogix/cx-cli/tree/master/skills) 🎖️ ⭐ | Coralogix | Official skills to query logs, spans, metrics, RUM, alerts & SLOs via the cx CLI | 🤖 | | [OpenSearch Agent Skills](https://github.com/opensearch-project/opensearch-agent-skills) 🎖️ | @opensearch-project | Log & trace analytics with PPL queries, error and trace investigation | 📦 | | [kstack](https://github.com/kubetail-org/kstack) ⭐ | @kubetail-org | Kubernetes log/metrics investigation + security/cost/network audits via slash commands | 🤖 | ### Metrics 与监控 *查询指标，构建仪表盘，设置告警规则，追踪 SLO。* | Skill | Creator | Description | Platform | | --------------------------------------------------------------------------------------------------------------------- | --------------- | ----------------------------------------------------------------------------------------- | -------- | | [Grafana Skill](https://github.com/julianobarbosa/claude-code-skills/tree/main/skills/grafana) | @julianobarbosa | Grafana HTTP API for dashboards, alerts, data sources, annotations | 🤖 | | [Prometheus Skill](https://github.com/julianobarbosa/claude-code-skills/tree/main/skills/prometheus) | @julianobarbosa | Query Prometheus HTTP API, execute PromQL, analyze time series | 🤖 | | [Loki Skill](https://github.com/julianobarbosa/claude-code-skills/tree/main/skills/loki) | @julianobarbosa | Grafana Loki log aggregation, LogQL queries | 🤖 | | [Mimir Skill](https://github.com/julianobarbosa/claude-code-skills/tree/main/skills/mimir) | @julianobarbosa | Grafana Mimir for long-term Prometheus metrics storage | 🤖 | | [Tempo Skill](https://github.com/julianobarbosa/claude-code-skills/tree/main/skills/tempo) | @julianobarbosa | Grafana Tempo distributed tracing | 🤖 | | [Pyroscope Skill](https://github.com/julianobarbosa/claude-code-skills/tree/main/skills/pyroscope) | @julianobarbosa | Continuous profiling with Grafana Pyroscope | 🤖 | | [Monitoring Expert](https://skillsmp.com/de/skills/jeffallan-claude-skills-skills-monitoring-expert-skill-md) | @Jeffallan | Build monitoring systems (Prometheus, Grafana, Datadog), incident detection | 📦 | | [Monitoring Config Generator](https://openclaw.army/skills/lxgicstudios/monitor-gen/) | lxgicstudios | Generate Prometheus alert rules + Grafana dashboards | 🦞 | | [Production Monitoring](https://skillmd.ai/how-to-build/production-monitoring/) | SkillMD | Daily health checks, container status, error scanning | 📦 | | [Uptime Kuma](https://openclaw.army/skills/msarheed/uptime-kuma/) | @msarheed | Manage uptime monitors, heartbeats, pause/resume endpoints | 🦞 | | [DevOps Claude Skills](https://github.com/ahmedasmar/devops-claude-skills) ⭐ | @ahmedasmar | 6 scripts: metrics analysis, SLO calculation, log analysis, dashboards | 🤖 | | [Grafana Skills](https://github.com/grafana/skills) 🎖️ ⭐ | @grafana | 44 official skills for Grafana, PromQL, Loki, Tempo, Mimir, Pyroscope, k6 & on-call/IRM | 🤖 | | [Checkly Skill](https://github.com/checkly/checkly-cli/tree/main/skills/checkly) 🎖️ ⭐ | Checkly | Synthetic & uptime monitoring-as-code: API/browser/multistep checks, alerts, status pages | 📦 | | [Observability & Monitoring Plugin](https://github.com/wshobson/agents/tree/main/plugins/observability-monitoring) ⭐⭐ | @wshobson | Prometheus config, Grafana dashboards (RED/USE), SLO implementation & distributed tracing | 🤖 | ### 事件响应与告警 *对事件进行分类，执行 runbook，划分严重等级，生成复盘报告。* | Skill | Creator | Description | Platform | | ------------------------------------------------------------------------------------------------------------------------------------------- | ------------ | ------------------------------------------------------------------------------------ | -------- | | [Incident Triage](https://skillmd.ai/how-to-build/incident-triage-1/) | SkillMD | SEV classification, first 5-15 min response framework | 📦 | | [On-Call Playbooks](https://skillmd.ai/how-to-build/on-call-playbooks/) | SkillMD | Comprehensive runbooks for common incidents | 📦 | | [IT Operations](https://skillsmp.com/es/skills/davila7-claude-code-templates-cli-tool-components-skills-development-it-operations-skill-md) | @davila7 | ITIL service management, observability strategies, incident response | 📦 | | [Incident Response (Anthropic)](https://github.com/anthropics/knowledge-work-plugins/tree/main/engineering/skills/incident-response) 🎖️ ⭐⭐ | Anthropic | Incident triage (SEV1-4), status comms & blameless postmortems with 5-whys RCA | 🤖 | | [PagerDuty Claude Code Plugins](https://github.com/PagerDuty/claude-code-plugins) 🎖️ | PagerDuty | Official PagerDuty plugins: commit-risk scoring from incident history, skill-builder | 🤖 | | [Axiom SRE](https://github.com/axiomhq/skills/tree/main/skills/sre) 🎖️ | @axiomhq | Official Axiom skill: query logs/metrics & investigate incidents via APL | 🤖 | | [Incident Commander](https://github.com/borghei/Claude-Skills/blob/main/engineering/incident-commander/SKILL.md) ⭐ | @borghei | SEV scoring, timeline reconstruction & postmortem/RCA generation | 📦 | | [Incident Response Plugin](https://github.com/wshobson/agents/tree/main/plugins/incident-response) ⭐⭐ | @wshobson | Postmortem writing and on-call handoff patterns | 🤖 | | [Claude Tabletop](https://github.com/cjcsecurity/claude-tabletop) | @cjcsecurity | Generate project-aware IR tabletop exercises, facilitator runbook & AAR | 🤖 | ### 可靠性与混沌工程 *定义 SLO 和错误预算，运行混沌实验，并在故障影响生产环境之前验证系统弹性。* | Skill | Creator | Description | Platform | | ---------------------------------------------------------------------------------------------------------------- | -------- | ------------------------------------------------------------------------------------------ | -------- | | [AWS Resilience Skills](https://github.com/aws-samples/sample-aws-resilience-skill) 🎖️ | AWS | Resilience maturity/modeling, chaos engineering (FIS), EKS checks, Well-Architected review | 📦 | | [Chaos Engineering](https://github.com/borghei/Claude-Skills/blob/main/engineering/chaos-engineering/SKILL.md) ⭐ | @borghei | Design & run fault-injection experiments with hypotheses and blast-radius control | 📦 | ### 基础设施与云 *Kubernetes 故障排查，云平台运维，DevOps 工具包。* | Skill | Creator | Description | Platform | | -------------------------------------------------------------------------------------------------- | --------------- | ---------------------------------------------------------------------------------------- | -------- | | [HolmesGPT Skill](https://github.com/julianobarbosa/claude-code-skills/tree/main/skills/holmesgpt) | @julianobarbosa | Investigate Kubernetes issues, analyze alerts, troubleshoot cloud-native apps | 🤖 | | [Microsoft Azure Skills](https://github.com/microsoft/skills) 🎖️ ⭐⭐ | Microsoft | Azure SDKs, AI Foundry integration, cloud resource management | 🤖 | | AWS Cost & Operations | @zxkane | CloudWatch metrics, Cost Explorer, Well-Architected reviews | 🤖 | | [Cloudflare Troubleshooting](https://github.com/daymade/claude-code-skills) ⭐⭐ | @daymade | Diagnose Cloudflare issues (SSL, DNS, redirects, caching) | 🤖 | | [ServiceNow Agent](https://openclaw.army/skills/thesethrose/servicenow-agent/) | @thesethrose | Read-only queries to ServiceNow ITSM (tables, CMDB, incidents) | 🦞 | | [DevOps Skills](https://github.com/lgbarn/devops-skills) | @lgbarn | Terraform/OpenTofu workflows, AWS infrastructure, safety-first IaC | 🤖 | | [Kubernetes Skill](https://github.com/LukasNiessen/kubernetes-skill) ⭐ | @LukasNiessen | Grounds agents in official Kubernetes/Helm/Kustomize best practices & security hardening | 🤖 | | [Flux CD Agent Skills](https://github.com/fluxcd/agent-skills) 🎖️ ⭐ | @fluxcd | Debug live Flux/Kubernetes GitOps clusters, audit repo readiness, generate manifests | 🤖 | | [Terraform Skill](https://github.com/antonbabenko/terraform-skill) ⭐⭐ | @antonbabenko | Terraform/OpenTofu best practices: testing, modules, state, CI/CD, security scanning | 📦 | | [Neon PostgreSQL Skills](https://github.com/neondatabase/postgres-skills) ️ | @neondatabase | PostgreSQL reliability best practices: performance, HA and operational tuning | 🤖 | ### 安全与合规 *威胁检测，漏洞扫描，凭证泄露防范，审计追踪。* | Skill | Creator | Description | Platform | | ------------------------------------------------------------------------- | --------------- | ------------------------------------------------------------------------------------------ | -------- | | [ClawSec](https://github.com/prompt-security/clawsec) ⭐⭐ | Prompt Security | Security skill suite: prompt injection, supply chain, credential leak detection | 🦞 | | [Skill Guard](https://github.com/UseAI-pro/openclaw-skills-security) | UseAI-pro | Monitor skill behavior, flag permission violations, sandbox unsafe actions | 🦞 | | [Trail of Bits Security Skills](https://github.com/trailofbits/skills) ⭐⭐ | Trail of Bits | Security research, vulnerability detection, audit workflows (fix review, variant analysis) | 🤖 | ## 技能注册中心 | Registry | Skills | Notes | | ---------------------------------------------------------------------------------- | ---------- | ---------------------------------- | | [awesome-claude-skills](https://github.com/ComposioHQ/awesome-claude-skills) ⭐⭐ | 1000+ | ComposioHQ curated list | | [awesome-openclaw-skills](https://github.com/VoltAgent/awesome-openclaw-skills) ⭐⭐ | 5,200+ | VoltAgent curated list | | [SkillsMP](https://skillsmp.com) | 2,000,000+ | Cross-platform marketplace | | [ClawHub](https://github.com/openclaw/clawhub) ⭐⭐ | varies | OpenClaw official registry | | [Anthropic Skills](https://github.com/anthropics/skills) 🎖️ ⭐⭐ | ~17 | Official reference implementations | | [agent-skills.md](https://agent-skills.md/) | varies | Skill discovery site | ## 💡 SRE 场景下的 Skills 与 MCP 对比 | 方面 | Skills | MCP Servers | | --------------- | ------------------------------------------------------ | --------------------------------------------------- | | **目的** | 流程化工作流、runbook、分析模式 | 数据访问、API 集成 | | **最适用于** | 事件分类、生成复盘报告、告警分析 | 查询指标、获取日志、创建仪表盘 | | **示例** | “分析此错误并提出修复建议” | “获取过去一小时的 CPU 指标” | | **执行方式** | 指令 + 可选脚本 | 运行中的服务器进程 | | **可移植性** | 可将 `SKILL.md` 复制到任何位置 | 需要服务器设置 | **何时使用 Skills：** 当你需要 Claude 遵循特定的工作流，或通过脚本访问实时数据时——例如分类事件、查询 API 或生成报告。 **何时使用 MCP：** 当你想要一个用于数据访问的标准化协议时——MCP 跨不同的工具和数据源提供了一致的接口。 **结合使用：** Skills 可以编排 MCP 调用。例如，一个“事件调查” Skill 可能会使用 Datadog MCP 获取相关日志，或者直接通过 shell 脚本调用 API。 ## 资源 ### 相关清单 - [awesome-devops-mcp-servers](https://github.com/rohitg00/awesome-devops-mcp-servers) - 包含可观测性工具的 DevOps MCP 服务器。 - [awesome-mcp-servers](https://github.com/punkpeye/awesome-mcp-servers) ⭐⭐ - 通用 MCP 服务器 - [awesome-observability](https://github.com/adriannovegil/awesome-observability) ⭐⭐ - 通用可观测性工具 - [awesome-sre](https://github.com/dastergon/awesome-sre) ⭐⭐ - 站点可靠性工程 (SRE) 资源 ### Skill 开发 - [Agent Skills Spec](https://github.com/agentskills/agentskills) ⭐⭐ - 开放的可移植技能标准 - [Claude Code Skills Guide](https://code.claude.com/docs/en/skills) - Anthropic 官方文档。 ### 安全研究 - [Bitdefender AI Skills Checker](https://www.bitdefender.com/en-us/consumer/ai-skills-checker) - 验证 OpenClaw skill 的安全性。 - [1Password: OpenClaw Attack Surface](https://1password.com/blog/from-magic-to-malware-how-openclaws-agent-skills-become-an-attack-surface) - 关于恶意 skills 的安全研究。 ## ❓ 常见问题

SRE skills 和 MCP 服务器有什么区别？

Skills 教会 Claude *如何*做某事（工作流、分析模式）。MCP 为 Claude 提供*访问*某事物的能力（API、数据库）。对于 SRE：使用 skills 来执行事件 runbook 和分析工作流；使用 MCP 来查询你实际的监控数据。

我可以将这些 skills 与我现有的监控技术栈一起使用吗？

是的！大多数 skills 不受技术栈限制——它们教授的是模式和工作流。部分 skills 是特定平台的（例如 Datadog、Grafana），需要访问这些平台。请查看每个 skill 的文档了解前置条件。

如何创建自定义 SRE skill？

1. 创建一个包含 `SKILL.md`（OpenClaw）或 `CLAUDE.md`（Claude Code）的文件夹 2. 为你的工作流编写指令（例如“如何分类 P1 事件”） 3. 可选择添加脚本以实现自动化 4. 在本地测试，然后通过 GitHub 分享 请查阅官方文档中的[创建 Skills](#creating-skills)。

skills 支持所有 Claude 模型吗？

Skills 支持 Claude Pro、Max、Team 和 Enterprise 版本。免费层级用户无法使用 Skills。关于 API 使用，请查阅 [Skills API 文档](https://docs.anthropic.com)。

为什么有些链接带有不同的图标？

- 🤖 = 兼容 Claude Code（`CLAUDE.md` 格式） - 🦞 = 兼容 OpenClaw（`SKILL.md` 格式） - 📦 = SkillMD 跨平台包 - 🎖️ = 官方/供应商维护

标签：AI智能体, API集成, 可观测性, 子域名突变, 故障响应, 用户代理, 监控, 站点可靠性工程, 自定义请求头, 资源清单, 运维