Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Cloud & DevOps
Convert Docker Compose files or installation docs into production-grade Sealos templates. Use when user has a docker-compose.yml and wants a Sealos or Kubernetes template, wants to migrate from Docker Compose to Sealos, needs to convert container orchestration configs to Sealos format, or mentions compose-to-template conversion. Also triggers on "/docker-to-sealos".
Full FlagRelease pipeline orchestrator. Runs the complete LLM deployment, verification, and benchmarking pipeline for multi-chip GPU backends. Executes: install-stack → env-verify → model-verify → perf-test in sequence, passing state between steps and producing a final structured report. Assumes gpu-container-setup (Step 1) is already done — a running container with PyTorch + GPU access must exist.
This skill should be used at the start of any computationally intensive scientific task to detect and report available system resources (CPU cores, GPUs, memory, disk space). It creates a JSON file with resource information and strategic recommendations that inform computational approach decisions such as whether to use parallel processing (joblib, multiprocessing), out-of-core computing (Dask, Zarr), GPU acceleration (PyTorch, JAX), or memory-efficient strategies. Use this skill before running analyses, training models, processing large datasets, or any task where resource constraints matter.
Checks Redis cluster metadata files (nodes-*.conf) for integrity, detecting 7 categories of metadata issues that cause Redis instance startup failures, connection anomalies, and fault alerts. 执行7项检查：slot槽位完整性、master/slave实例个数、文件格式(4个模块)、文件名端口一致性、端口关系、主备关系、myself标记。 Use this skill when the user mentions Redis metadata file checks, nodes-*.conf checks, or Redis cluster configuration checks. Use it whenever the user asks about Redis startup failures or connection anomalies. Trigger: "Redis元数据检查", "Redis nodes检查", "Redis cluster配置检查", "检查Redis元数据", "Redis元数据校验", "Redis启动失败排查"
Install the 5-package multi-chip software stack (vLLM, FlagTree, FlagGems, FlagCX, vllm-plugin-FL) inside a GPU container. Handles network mirror detection, dependency ordering, wheel selection, and per-package validation. Use after gpu-container-setup has produced a running container with PyTorch + GPU access.
Investigate security incidents in Amazon Web Services—reconstruct attacker activity from CloudTrail, VPC Flow Logs, and GuardDuty; anchor the investigation on the compromised principal (access key or role); trace privilege escalation and persistence through IAM API calls; detect data exfiltration and crypto-mining; and contain the incident without destroying evidence or tipping off the attacker. Use when responding to a suspected AWS compromise, exposed access keys, anomalous CloudTrail activity, a GuardDuty finding, unexpected IAM changes, crypto-mining EC2 instances, or S3 data exfiltration.
Kubernetes operational knowledge for accessing clusters, running kubectl, understanding Flux status, and navigating the homelab Kubernetes environment. Use when: (1) Accessing a cluster or checking connectivity, (2) Running kubectl commands or checking resource status, (3) Checking Flux reconciliation status or triggering reconciliation, (4) Finding internal service URLs, (5) Understanding cluster layout or resource types, (6) Researching unfamiliar Helm charts or services. Triggers: "kubectl", "kubeconfig", "flux get", "flux reconcile", "flux status", "cluster access", "internal URL", "service URL", "prometheus URL", "grafana URL", "helm release status", "check flux", "which cluster", "how to access", "port-forward".
Deploy containerized applications (especially Rails) to VPS using Kamal 2. Covers deploy.yml configuration, accessories (PostgreSQL, Redis, Sidekiq), SSL/TLS, secrets management, CI/CD with GitHub Actions, database backups, server hardening, debugging, and scaling. Use when setting up Kamal, configuring deployments, troubleshooting deploy issues, or managing production infrastructure with Kamal.
Manages existing Grafana dashboards via gcx: list, search, get, audit panels, queries, datasources, and variables; apply a finished manifest; delete; versions and restore; pull/push/validate files; folders; PNG snapshots. Not for new dashboards, panels, or variables (use create-dashboard).
Manage Workers/KV/R2/D1/Hyperdrive via Cloudflare MCP, perform observability/build troubleshooting/audit/container sandbox operations. Triggers: worker/KV/R2/D1/logs/build/deploy/screenshot/audit/sandbox. Three permission tiers: Diagnose (read-only), Change (write requires confirmation), Super Admin (isolated environment). Write operations must follow read-first, user confirmation, post-execution verification.
Bootstrap or maintain MkDocs Material docs sites with GitHub Pages, strict two-pass i18n plus default-language llms.txt, copy-to-LLM, page/nav helpers, and zh-TW translation. Use when setting up or publishing docs, adding bilingual or multilingual docs, repairing nearly empty llms.txt or i18n strict-build failures, or migrating an older mkdocs-site-bootstrap site. Consent-gated; never rewrites existing docs without approval.
Verify the serving stack with a user-specified target model. Runs twice: first with FlagGems/FlagCX disabled (isolate model-specific errors), then with full multi-chip stack enabled. Diffs the two runs to pinpoint which layer caused any failure.
NetBox Assurance drift detection and deviation management. Use when working with network state comparison, deviation review/remediation, data source configuration, or understanding how intended vs actual network state is reconciled in NetBox Cloud and Enterprise
Drive the NetBox Labs Platform MCP server (hosted on NetBox Cloud) from an AI agent. Covers connecting/auth, Code Mode (the three tools netbox_search_schema / netbox_execute / rediscover_netbox, the injected helper catalog, the sandbox contract, round-trip minimization) and Discrete per-tool mode. Use when an agent connects to the Platform MCP server, writes code-mode programs against NetBox, or chooses between code and discrete mode.
Use only for offline/local OSMO service-config admin requests involving explicit config folders or values files, or to ask for one when a file-specific config request omits it. Do not inspect the workspace to infer a path. Do not use for live workflow support, resource capacity, pod/node diagnostics, or cluster operations, except live service-config paths that must be refused.
Use when analyzing failures, outages, incidents, or negative outcomes, conducting blameless postmortems, documenting root causes with 5 Whys or fishbone diagrams, identifying corrective actions with owners and timelines, learning from near-misses, establishing prevention strategies, or when user mentions postmortem, incident review, failure analysis, RCA, lessons learned, or after-action review.
Guide users through connecting to a ScyllaDB Cloud cluster, choosing the correct driver and version, and configuring authentication. Use this skill when a user needs to connect to ScyllaDB Cloud, pick a CQL driver, check a driver version, configure credentials, set the default consistency level, size the connection pool, or troubleshoot connection issues. Cluster management through an AI agent goes to scylladb-cloud-mcp. Triggers on "connect to ScyllaDB Cloud", "ScyllaDB connection", "ScyllaDB driver setup", "which driver", "driver version", "Java driver", "Python driver", "CQL connection", "DC-aware load balancing", "ScyllaDB credentials", "connection bundle", "shard-aware", "tablet-aware", "consistency level", "LOCAL_QUORUM", "too many connections".
Deploy any GitHub project to Sealos Cloud in one command. Assesses readiness, generates Dockerfile, builds image, creates Sealos template, and deploys — fully automated. Use when user says "deploy to sealos", "deploy this project", "deploy to cloud", "deploy this repo", mentions Sealos deployment, wants to deploy a GitHub URL or local project to a cloud platform, or asks about one-click deployment. Also triggers on "/sealos-deploy".
Systematic Linux server management. Use BEFORE making changes to capture baseline, then AFTER for verification. Triggers on 'check this server', 'audit', 'set up this machine', 'security audit', 'harden this Pi', 'fresh Pi setup', 'provision this server'. (user)
Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism
Fast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM with prefix sharing. Powers 300,000+ GPUs at xAI, AMD, NVIDIA, and LinkedIn.
Sveltia CMS, a lightweight Git-backed headless CMS and modern successor to Decap/Netlify CMS (about 300 KB bundle, GraphQL-based). Covers admin/config.yml collection config, YAML and TOML frontmatter, OAuth setup including a Cloudflare Workers OAuth proxy, i18n with DeepL translation, image optimization, and deployment on Cloudflare Pages, Netlify, Vercel, or GitHub Pages. Framework-agnostic across Hugo, Jekyll, 11ty, Gatsby, Astro, SvelteKit, and Next.js. Use when adding a Git-based editing interface to a static site for non-technical editors, migrating from Decap or Netlify CMS, or fixing OAuth authentication failures, YAML parse errors, CORS/COOP policy problems, content not listing, script loading errors, or image upload failures.
UiPath Admin via `uip admin` — Identity Server (users, groups, robot accounts, external OAuth2 apps, secrets, PATs, SMTP), Authorization (custom roles, role assignments, permission catalog, effective-access check-access PDP), OMS (org read/update, tenant lifecycle, service provisioning, regions, async op polling), IP Restriction (allowlist, enforcement, bypass rules, lockout safety), and Audit via `uip admin audit` (event sources, paginated queries, JSON-folder or CSV export, exclusion rules). Troubleshoot access-denied, login failures, role misconfig, IP lockout, PAT/app auth. Owns ALL org/tenant/identity audit — use `uip admin audit`, NOT `uip or audit-logs`, for any audit logs / audit trail / audit events / export / login history / who-did-what request. Also owns audit exclusion rules: stop, suppress, or mute recording of chosen audit events (`audit org exclusions`). Orchestrator-specific roles/permissions/folders/jobs→uipath-platform. RPA workflows→uipath-rpa.
Always invoke for `.uipx` files. UiPath Solution lifecycle via `uip solution`: init/download/restore/pack/publish/deploy/activate/upgrade/upload, projects add|import|remove, resources refresh|add|remove|edit, feeds list, `--feed` / `--personal-workspace`. DIAGNOSE lifecycle failures: pack/publish/deploy/activate errors, stale/missing bindings, unimported/unresolved resources, deploy error codes, `--acknowledge-permanent-deletion` consent, publish name+version/processKey collisions, redeploy blocked by an existing deployment (`deploy upgrade`), feed 'not an available publish location', package not in `packages list`, 'unknown command' after new→init rename. Bundles RPA/Flow/Case/Agents/API Workflow/AppV2 projects in one `.uipx`. Routing: PDD→SDD design & task derivation→uipath-planner; non-solution Orchestrator/IS/resources/auth/traces→uipath-platform; .xaml/.cs→uipath-rpa; .flow→uipath-maestro-flow; .bpmn→uipath-maestro-bpmn; agent.json/.py agents→uipath-agents; standalone coded-app deploy→uipath-coded-apps.
Look up VictoriaMetrics, VictoriaLogs, VictoriaTraces, operator, vmanomaly, Helm chart and VictoriaMetrics Cloud documentation before naming a command-line flag, an HTTP API path, or a LogsQL/MetricsQL construct. Use when about to state that a flag or endpoint exists, when a user asks what a flag does, or when a previous answer may have invented an API path. Triggers on: command-line flag, CLI flag, what flag, flag does, API path, HTTP endpoint, does this endpoint exist, MetricsQL function, LogsQL pipe, LogsQL filter, operator flag, is this documented, where are the docs, VictoriaMetrics documentation.
Generate an interactive bash wizard that walks a human through steps only they can perform. Use when provisioning infrastructure, setting up credentials or CI secrets, walking an unfamiliar third-party dashboard, or running a one-off migration or cutover. Don't invoke this for steps the agent can perform itself.
Optional: Use for WP Engine hosting workflows — SSH-based git push, remote WP-CLI via SSH gateway, GitHub Actions CI/CD with safety-gated deploys to dev/staging/production, managing installs/domains/cache/backups via the wpe-labs Claude Code skills, and WP Engine API access. Requires SSH key setup and WPE_USERNAME/WPE_PASSWORD env vars for API operations.
Expert in migrating CI/CD pipelines to Woodpecker CI from GitHub Actions, Semaphore CI, GitLab CI, Jenkins, Drone CI, and other platforms. Helps write and debug .woodpecker.yml configurations, translate pipeline steps, find Woodpecker plugins, set up services, workflows, matrix builds, secrets, volumes, and when-conditions. Use this skill whenever the user mentions Woodpecker CI, .woodpecker.yml, migrating CI pipelines, converting GitHub Actions to Woodpecker, Woodpecker plugins, Woodpecker syntax, or asks for help with any CI/CD configuration involving Woodpecker — even if they don't explicitly name "Woodpecker" but mention migrating away from GitHub Actions or Semaphore CI.
Complete guide to zapret-openwrt, an anti-DPI tool for OpenWrt routers. Use for any zapret-openwrt question: project architecture, configuration, DPI bypass strategies, LuCI web UI, scripts, updates, diagnostics, custom.d hooks, host lists, nfqws parameters, procd service management. Also use for development: adding strategies, changing the UI, writing custom.d scripts, debugging. Triggers: zapret, nfqws, DPI desync, обход блокировок, антиблокировка, zapret strategies, hostlist, autohostlist, стратегии zapret, роутер, блокировки.
Use for cross-module development, tracing which chain something belongs to, deciding which BK-CI module a requirement should land in, or quickly understanding how the full pipeline chain works together. For single-module changes, read the corresponding module skill first rather than staying here.
Use this skill for Node.js backend AI via @cloudbase/node-sdk (>=3.16.0) — cloud functions, CloudRun, Express/Koa/NestJS, serverless APIs, scheduled jobs, LLM proxies, agent orchestration. The only SDK supporting image generation (ai.createImageModel + generateImage). Text via ai.createModel with groups cloudbase, hunyuan-exp, or custom-*; model ids (e.g. deepseek-v4-flash, glm-5, kimi-k2.6) go in the `model` field of generateText/streamText. MUST run two-step preflight before code — see body. NOT for browser/Web (use ai-model-web) or Mini Program (use ai-model-wechat).
Use when working on BK-CI artifact upload and download, artifact metadata, BkRepo or disk backend storage, file tasks, and cleanup chains. Prefer this when users mention build artifacts, artifact archiving, download tokens, report files, BkRepo integration, or artifact cleanup.
Assess managed Kubernetes clusters on EKS, GKE, and AKS by exploiting the seams between cloud IAM and Kubernetes RBAC -- IRSA/OIDC trust abuse, Workload Identity Federation, pod-to-IMDS escalation, aws-auth ConfigMap takeover, node pool service account abuse, and AAD integration weaknesses. Use when pentesting a managed k8s cluster, reviewing RBAC in EKS/GKE/AKS, testing pod-to-cloud escalation, or assessing network policy enforcement across namespaces
Attack and enumerate Azure AD / Entra ID tenants: initial reconnaissance with AADInternals and ROADtools, password spraying, token theft (PRT, CAE, refresh tokens), application and service principal abuse, Conditional Access bypass, cross-tenant pivoting, hybrid identity attacks (PTA agent, Azure AD Connect), and managed identity abuse. Use when penetration-testing Entra ID tenants, assessing Azure AD security posture, or exploiting cloud identity misconfigurations.
Attack serverless compute — AWS Lambda, Azure Functions, GCP Cloud Functions, and edge runtimes like Cloudflare Workers. Enumerate functions, inject through event sources (S3, SQS, SNS, API Gateway, EventBridge), extract credentials from the execution environment, abuse over-privileged IAM roles, exploit cold-start persistence and container reuse, and test trigger misconfigurations. Use when a target runs serverless functions, when API Gateway or function URLs front compute, when event-driven architectures process untrusted input, or when cloud function execution roles need privilege-escalation testing
AWS cost optimization, monitoring, and operational excellence expert. Use when analyzing AWS bills, estimating costs, setting up CloudWatch alarms, querying logs, auditing CloudTrail activity, or assessing security posture. Essential when user mentions AWS costs, spending, billing, budget, pricing, CloudWatch, observability, monitoring, alerting, CloudTrail, audit, or wants to optimize AWS infrastructure costs and operational efficiency.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.