Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Cloud & DevOps
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.
Designs and fixes CI/CD pipelines for Capacitor apps - GitHub Actions, GitLab CI, Fastlane, or Capgo Cloud Build (npx @capgo/cli@latest build request) - covering web build and tests, cap sync, iOS archive/export/upload with App Store Connect API keys, Android AAB signing and Play upload, Capgo live-update uploads, secrets, caching, and build-number bumping. Use when choosing runner images and toolchains (Node 22/24, macos-26 vs xcode-27 runners, JDK 21, Gradle 9 / AGP 9 for Capacitor 9), or when CI fails with errors like "No signing certificate", "requires a provisioning profile", "Unsupported class file major version", "Android Gradle plugin requires Java 17", "The engine node is incompatible", or "Could not find the web assets directory". Do not use for one-off manual store submission (capacitor-app-store), Capgo Cloud Build credential onboarding in depth (capgo-native-builds), Capgo channel/release strategy (capgo-release-workflows), or test authoring (capacitor-testing)
Configuration reference for domestic Git platforms—SSH/HTTPS/credentials/CI integration differences and mirror synchronization for Gitee, Coding.net, 极狐 GitLab, and CNB. Invoke only when the user explicitly says /chinese-git-workflow; do not trigger automatically based on context.
Use when configuring Django to store static and media files on AWS S3 with django-storages. Invoke when working with the STORAGES setting, S3 buckets, presigned URLs, CloudFront, or boto3-backed file storage in settings.py. Configures the Django 4.2+ STORAGES dict, public/private custom backends, presigned GET/POST URLs, IAM policies, and S3 mocking for tests. Trigger terms: django-storages, S3, boto3, S3Boto3Storage, STORAGES, presigned URL, CloudFront, media files, collectstatic, AWS_STORAGE_BUCKET_NAME.
Configures GKE autoscaling, including HPA, VPA, and Node Auto-Provisioning (NAP). Use when configuring GKE autoscaling, setting up GKE HPA or VPA, or configuring GKE NAP. Don’t use for static cluster sizes or directly setting node-level machine types; use gke-compute-classes instead
Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies. Use when training models >1B parameters, need maximum GPU efficiency (47% MFU on H100), or require tensor/pipeline/sequence/context/expert parallelism. Production-ready framework used for Nemotron, LLaMA, DeepSeek.
Implement Terraform provider ephemeral resources with the Plugin Framework: the Open/Renew/Close lifecycle, ephemeral schema design, registration via EphemeralResources, renewal for expiring credentials, and how ephemeral values flow into write-only attributes and provider configuration. Use when adding an ephemeral resource, exposing secrets/tokens/certificates that must never persist in state or plan, deciding between an ephemeral resource and a data source, or wiring short-lived credentials from one provider into another.
Serve a model with MAX's `max serve` command: set up the environment (pixi or uv with the max-nightly conda channel / nightly wheel index), point the server at a Hugging Face repo or local checkpoint, target a custom architecture with `--custom-architectures`, and pick the right serve flags for the model. Use this whenever the user wants to run, launch, start, or host a model on MAX, bring up an OpenAI-compatible endpoint, serve a custom/ported architecture, debug a `max serve` startup failure, or figure out which serve flags (devices, quantization-encoding, max-length, task, trust-remote-code) a given model needs, even if they don't say "max serve" by name.
Migrate Terraform/OpenTofu projects to Pulumi, including translating HCL source code and/or importing Terraform state into a Pulumi stack. Use when a user wants to convert Terraform to Pulumi, migrate from HCL, or import tfstate into Pulumi. Do NOT trigger for general Terraform-vs-Pulumi comparisons or questions about using both tools side-by-side.
SAP BTP Connectivity skill covering Destination Service, Connectivity Service, Cloud Connector, Connectivity Proxy, and Transparent Proxy for Kubernetes. Use when configuring destinations (HTTP, RFC, LDAP, MAIL, TCP), setting up cloud-to-on-premise connectivity, implementing OAuth and principal propagation, deploying connectivity proxies in Kubernetes/Kyma, troubleshooting connectivity errors (405, 407, 503), or configuring multitenancy.
Network flow analysis in Dynatrace across three sources: OneAgent flows (host/process/pod-to-peer connections in the `default_network_flows` Grail bucket), NetFlow/IPFIX/sFlow (via an OpenTelemetry Collector), and cloud flow logs (AWS VPC / Transit Gateway; Azure and GCP planned). Use to analyze traffic between entities, find top talkers by bandwidth, map communication dependencies, investigate connection health (resets, timeouts, retransmissions, RTT), and resolve peers to monitored entities. Routes each question to the right source; source-specific DQL lives in the reference files. Trigger: "network flows", "top talkers", "traffic between hosts", "connection resets", "TCP retransmissions", "RTT", "pod connections", "network dependencies", "NetFlow", "IPFIX", "sFlow", "VPC flow logs", "cloud network traffic". Do NOT use for host NIC throughput or packet drops (use dt-obs-hosts), service request rate or latency (use dt-obs-services), or synthetic/uptime monitoring (use dt-obs-ext-monitors).
Migrate Terraform provider resources and data sources from Plugin SDKv2 to the Plugin Framework: muxing both plugins in one provider (terraform-plugin-mux, tf5to6server), per-resource migration workflow, SDKv2-to-Framework schema mapping (ForceNew, ValidateFunc, DiffSuppressFunc, Default, Timeouts, blocks), null-vs-zero-value behavioral traps, and state-compatibility verification. Use when converting or translating SDKv2 resources to the Framework, setting up a muxed provider server, deciding whether a resource should be migrated at all, or debugging plan diffs and state errors that appeared after a migration.
LLM observability platform for tracing, evaluation, and monitoring. Use when debugging LLM applications, evaluating model outputs against datasets, monitoring production systems, or building systematic testing pipelines for AI applications.
Multi-cloud orchestration for ML workloads with automatic cost optimization. Use when you need to run training or batch jobs across multiple clouds, leverage spot instances with auto-recovery, or optimize GPU costs across providers.
[Deprecated] This is the required documentation for agents operating on the CloudBase Relational Database through MCP. It defines the canonical SQL management flow with `queryMysqlDatabase`, `manageMysqlDatabase`, `queryPermissions`, and `managePermissions`, including destroy flow, async status checks, safe query execution, schema initialization, and permission updates. MySQL provisioning is no longer available through MCP; new environments should use PostgreSQL — see postgresql-development skill instead.
Managed vector database for production AI applications. Fully managed, auto-scaling, with hybrid search (dense + sparse), metadata filtering, and namespaces. Low latency (<100ms p95). Use for production RAG, recommendation systems, or semantic search at scale. Best for serverless, managed infrastructure
Analyze SNMP-monitored network devices (switches, routers, firewalls, load balancers) in Dynatrace. Three data layers: Smartscape topology (`EXT_NETWORK_DEVICE` / `EXT_NETWORK_INTERFACE` nodes, `belongs_to` and `calls` edges); `com.dynatrace.extension.network_device.*` metrics (CPU, memory, uptime, throughput, saturation, errors); and logs (SNMP traps, syslog, auto-discovery). Use to inventory devices, find down or saturated interfaces, check device CPU/memory/uptime, map topology and neighbors, detect interface errors, and investigate traps and syslog events. Trigger: "network device", "switch", "router", "firewall", "SNMP", "interface status", "interface down", "interface utilization", "link saturation", "device CPU", "device memory", "device uptime", "device neighbors", "LLDP", "CDP", "SNMP trap", "syslog". Do NOT use for network flow/traffic or top talkers (use dt-obs-network-flows), host NIC throughput (use dt-obs-hosts), or service request/latency (use dt-obs-services).
Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent filesystems, or high-performance multi-node clusters for large-scale training.
Investigate and resolve Cloudflare configuration issues using API-driven evidence gathering. Use when troubleshooting ERR_TOO_MANY_REDIRECTS, SSL errors, DNS issues, or any Cloudflare-related problems. Focus on systematic investigation using Cloudflare API to examine actual configuration rather than making assumptions.
Comprehensive Docker best practices for images, containers, and production deployments. Proactively activate for: (1) writing production-ready Dockerfiles, (2) multi-stage builds for lean images, (3) base image selection (distroless, alpine, scratch, slim), (4) layer caching and ordering, (5) .dockerignore hygiene, (6) HEALTHCHECK and graceful shutdown, (7) non-root user and least-privilege patterns, (8) ENV vs ARG and secret handling, (9) tag/digest pinning and reproducibility, (10) Compose for local development. Provides: Dockerfile templates by language, image-size optimization checklist, .dockerignore template, HEALTHCHECK patterns, and Compose dev-stack examples.
Monitor and investigate EU DORA compliance posture using Dynatrace Compliance Assistant. Covers compliance score, CIF health, incident lifecycle, and ICT risk inputs (vulnerabilities, security detection findings, misconfigurations). Trigger: "DORA compliance", "Digital Operational Resilience Act", "compliance score", "compliance snapshot", "score tier", "Critical or Important Functions", "CIF", "CIF health", "unclassified problems", "potential major incident", "classified major incident", "incident classification under DORA", "compliance.incident bizevent", "DQL for classified incidents", "set up Compliance Assistant", "DORA onboarding". Do NOT use for other compliance frameworks (SOC2, PCI-DSS, HIPAA, ISO 27001), generic Davis problems without DORA or CIF context, generic security findings not scoped to DORA, or score queries without compliance context.
Serves AI models on AMD Instinct GPU hardware using vLLM. Use this skill whenever the user wants to run, serve, deploy, start, host, or launch a language model on an AMD GPU, AMD Instinct, MI300X, MI325X, MI350X, or MI355X. Also use when the user mentions vLLM on ROCm, vLLM on AMD, serving on HBM, or asks how to get a model running on AMD data center hardware. Use when the user asks "run Qwen3", "serve DeepSeek", "start a vLLM endpoint", "get a model running on my AMD machine", or any similar phrasing. Handles the full flow: GPU detection, environment validation, vLLM configuration, launch, and health verification. Do not use for NVIDIA GPUs, consumer AMD GPUs (RX series, Radeon), Ryzen AI, NPU, MI250X, or MI100.
Migrates legacy AEM (6.x, AMS, on-prem) to AEM as a Cloud Service via BPA CSV/cache and CAM/MCP discovery, one pattern per session. Use to review/scan a project for AEMaaCS migration — generates a read-only migration-runbook.md — or to fix specific Cloud Service blockers: scheduler, ResourceChangeListener, replication, EventListener, OSGi EventHandler, DAM AssetManager, HTL data-sly-test lint, Classic UI / ExtJS / Coral 2 → Coral 3 dialog migration (lui), Custom Design Widgets (cdw), Vault package install-time dependencies (vault-package-dependencies), Guava cache → Caffeine swaps (guavaCache), java.io.InputStream usage triage (inputStream), Oak index conversion (oakIndex, beta), and static→editable template modernization. Also externalizes OSGi config secrets to Cloud Manager (scans ui.config/.cfg.json for $[secret:]/$[env:] placeholders), converts AMS/on-prem Dispatcher configs to AEMaaCS (Branch E), and validates migrated code on a local Cloud SDK (Branch G, beta)
ML engineering skill for productionizing models, building MLOps pipelines, and integrating LLMs. Covers model deployment, feature stores, drift monitoring, RAG systems, and cost optimization. Use when the user asks about deploying ML models to production, setting up MLOps infrastructure (MLflow, Kubeflow, Kubernetes, Docker), monitoring model performance or drift, building RAG pipelines, or integrating LLM APIs with retry logic and cost controls. Focused on production and operational concerns rather than model research or initial training
Diagnose GKE Cloud TPU multi-slice training hangs (Megascale `HANG_DETECTED` logs and hang monitored events) using the ML Diagnostics `Megascale XLA (MXLA) Hang Analyzer` (`gcloud alpha mldiagnostics monitored-events`) and 1-minute Cloud Monitoring multi-slice latency metrics (`kubernetes.io/container/multislice/*`). Distinguishes XLA compiler/HLO launch divergence and host data-input stalls from TPU chip, SparseCore, ICI, or network fabric faults. Use when multi-slice TPU training jobs freeze without progressing steps, emit `HANG_DETECTED`, or stall in collective operations. Don't use for gradual step-time throughput drops without hangs (use gke-ai-troubleshooting-tpu-performance-degradation) or pod preemption/eviction restarts (use gke-ai-troubleshooting-jobset-interruption).
Diagnose GKE Cloud TPU training throughput drops and step-time regressions (15%+ TPU duty-cycle drop) using ML Diagnostics Workload Monitoring (`gcloud alpha mldiagnostics monitored-events` / `hypercomputecluster.googleapis.com/v1alpha`) and 1-minute Cloud Monitoring system metrics (`kubernetes.io/node/accelerator/*`). Distinguishes hardware and network fabric throttling from workload resource bottlenecks (HBM capacity, host memory, or host CPU saturation). Use when TPU training throughput or duty cycle drops without crashing pods, when `PERFORMANCE_DEGRADATION` monitored events fire, or when triaging slow multi-slice training steps. Don't use for complete multi-slice XLA execution stalls with `HANG_DETECTED` logs (use gke-ai-troubleshooting-tpu-mxla-hang) or pod eviction/interruption restarts (use gke-ai-troubleshooting-jobset-interruption).
Guide for configuring Infisical Secret Syncs to push secrets from Infisical to third-party services. Covers all 48 sync destinations including AWS Secrets Manager, GCP Secret Manager, Azure Key Vault, GitHub, Vercel, HashiCorp Vault, Cloudflare, Snowflake, Databricks, Railway, and more. Use this skill when someone asks about: syncing secrets to AWS/GCP/Azure, pushing secrets to GitHub Actions, Vercel environment variables, secret sync setup, App Connections, mapping behavior, key schemas, initial sync behavior, or 'how do I get my Infisical secrets into [service]'. Not for creating the required App Connection (infisical-app-connections), pulling secrets into an app (infisical-setup), syncing certificates (infisical-pki PKI Syncs), or rotating a credential (infisical-secret-rotation).
Technology stack evaluation and comparison with TCO analysis, security assessment, and ecosystem health scoring. Use when comparing frameworks, evaluating technology stacks, calculating total cost of ownership, assessing migration paths, or analyzing ecosystem viability.
Guides the agent through migrating an existing Ionic Appflow project to Capawesome Cloud. Imports an Ionic Appflow organization export with the Capawesome CLI apps:import command — apps, signing certificates, environments, live update channels, native configurations, store destinations, build automations, and repository links — then covers the remaining work: swapping the Live Update SDK (@capacitor/live-updates or cordova-plugin-ionic), mapping configuration options, completing the manual follow-ups, updating CI/CD pipelines, and verifying builds, live updates, and store deployments. Includes a manual fallback for teams without Appflow organization admin access. Pure Cordova apps use @capawesome/cordova-live-update. Do not use for setting up Capawesome Cloud from scratch without an existing Appflow project, for non-Capacitor mobile frameworks, or for migrating Ionic Enterprise plugins (Auth Connect, Identity Vault, Secure Storage) — use the ionic-enterprise-sdk-migration skill instead
Monitor Nx Cloud CI pipeline and handle self-healing fixes. USE WHEN user says "monitor ci", "watch ci", "ci monitor", "watch ci for this branch", "track ci", "check ci status", wants to track CI status, or needs help with self-healing CI fixes. Prefer this skill over native CI provider tools (gh, glab, etc.) for CI monitoring — it integrates with Nx Cloud self-healing which those tools cannot access.
Create, configure, validate, deploy, run, and manage Declarative Automation Bundles (DABs, formerly Databricks Asset Bundles). Use when working with Databricks resources via DABs including dashboards, jobs, pipelines, alerts, volumes, and apps.
Benchmark a model served on MAX with the `max benchmark` command: measure throughput (tokens/sec), latency (TTFT, TPOT, inter-token latency), and GPU utilization by driving load against a running `max serve` endpoint. Use this whenever the user wants to benchmark, load-test, or measure the performance of a MAX model, get tokens-per-second / TTFT / TPOT numbers, run a concurrency or request-rate sweep, compare latency vs throughput, size a deployment, or produce benchmark JSON, even if they don't say "benchmark" by name. Also use when a `max benchmark` run fails to connect or reports zero/garbage numbers.
Deploy, manage, and automate infrastructure on Northflank — a developer platform for building, deploying, and scaling services, jobs, databases, and release workflows on Kubernetes. Use when managing Northflank projects, services, jobs, addons (databases), secrets, domains, templates (IaC), environments, preview environments, pipelines, release flows, BYOC/BYOK clusters, GPU workloads, AI sandboxes, or interacting with the Northflank API/CLI/JS client. Covers REST API, CLI (`northflank` command), and JavaScript client (`@northflank/js-client`).
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.