Qovery Skills

AI agent skills for deploying and troubleshooting applications on Kubernetes using Qovery. Compatible with 30+ AI coding tools that support the Agent Skills open standard.

Quick Install

One command — installs both skills globally for all your projects:

curl -fsSL https://skill.qovery.com/install.sh | bash

Install in the current project only:

curl -fsSL https://skill.qovery.com/install.sh | bash -s -- --project

Uninstall:

curl -fsSL https://skill.qovery.com/install.sh | bash -s -- --uninstall

That's it. The installer automatically places both skills in all the right directories so they're discovered by any compatible tool.

Update to the latest version — just run the install command again. It overwrites the previous versions:

curl -fsSL https://skill.qovery.com/install.sh | bash

Compatible Tools

This skill follows the Agent Skills open standard and works with:

And any other tool that discovers skills from .claude/skills/ or .agents/skills/ directories.

What it does

qovery-onboard

When you tell your AI agent "I'm new to Qovery, help me get started":

1. Understands who you are — asks about your role (developer, DevOps, CTO, founder, non-technical), experience level with cloud infrastructure, and whether you know what Kubernetes is
2. Understands what you have — existing cloud provider account, existing K8s cluster (BYOK path), applications to deploy, migration from another platform
3. Understands what you need — primary goal, industry compliance (HIPAA, PCI-DSS, SOC2, GDPR), constraints (region, budget, private networking), team size
4. Recommends the optimal setup — cloud provider + region, cluster type (managed or BYOK), environment structure (dev/staging/prod), security defaults, cost defaults, RBAC roles
5. Executes the full setup with Console links at every step — cloud credentials, cluster creation (with progress updates while waiting), project/environments, deployment rules, git provider connection
6. Invites team members with appropriate RBAC roles (Admin, DevOps, Viewer, or custom roles for enterprise)
7. Handles BYOK (Bring Your Own Kubernetes) — guides through qovery cluster install for existing clusters
8. Handles migrations — detailed guides for Heroku (env var import via qovery env parse --heroku-json, concept mapping, FAQ), Vercel/Netlify, Render/Railway, and manual Kubernetes
9. Encodes best practices by default — private databases, internal networking, secrets encryption, deployment stages, deployment rules for cost savings — NOT optional recommendations, but the default setup

qovery-deploy

When you tell your AI agent "deploy my application with Qovery":

1. Analyzes your codebase — detects language, framework, ports, database needs, environment variables
2. Creates a Dockerfile if one is missing — production-ready templates for 12+ frameworks
3. Asks the right questions — dev vs production, database type, deployment method
4. Sets up infrastructure — cluster creation from scratch if needed (AWS, GCP, Azure, Scaleway)
5. Deploys via CLI + API (quick path) or Terraform provider (recommended for production)
6. Provisions databases — container mode for dev/test, managed mode (e.g. AWS RDS) for production, or Terraform services for advanced setups like RDS Aurora
7. Sets up environment variables, secrets, service interconnection, health checks, deployment stages
8. Handles Helm charts, Terraform modules, lifecycle jobs, and cron jobs
9. Watches deployments and auto-fixes failures — diagnoses build errors, port mismatches, health check failures, missing env vars, and OOM issues

qovery-troubleshoot

When you tell your AI agent "my Qovery deployment is failing, can you help?":

1. Integrates with the Qovery MCP Server for fast, structured diagnostics (falls back to CLI/API if MCP is not configured)
2. Runs an 8-layer systematic diagnosis — deployment status, build logs, runtime logs, health checks, environment variables, network/connectivity, resources/performance, cluster infrastructure
3. Matches 20+ error patterns — OOM kills, crash loops, connection refused, missing env vars, port mismatches, DNS failures, auth errors, database issues, and more
4. Includes 10 pre-built playbooks — App Won't Start, App Is Slow, Database Connection Fails, Deployment Stuck, Custom Domain Not Working, Terraform/Helm Errors, High Costs, OOM/Resource Exhaustion, Build Failing
5. Auto-fixes Qovery configuration (ports, health checks, memory, CPU, env vars, deployment stages) without permission; asks before modifying user code
6. Generates runbooks in .qovery/runbooks/ to document what happened and how it was fixed — builds institutional knowledge over time
7. Provides prevention recommendations tailored to the specific issue that was fixed

qovery-optimize

When you tell your AI agent "optimize my Qovery costs":

1. Gathers business context first — asks about application type, traffic patterns (seasonal spikes, business-hours, steady), growth expectations, and reliability requirements before touching any metrics
2. Analyzes historical resource consumption — compares allocated CPU/memory vs actual peak usage over 7-day and 30-day windows, with safety buffers adjusted by environment type
3. Analyzes 7 optimization dimensions — service right-sizing, autoscaling, database mode, environment scheduling, cluster optimization (spot instances, instance types), build optimization, external resource costs
4. Estimates external cloud resource costs — calculates costs for RDS, ElastiCache, NAT Gateways, load balancers, and other infrastructure from configuration parameters and public cloud pricing, with clear methodology disclaimers
5. Generates detailed cost reports — markdown report with executive summary, per-cluster/environment/service breakdown, external resource estimates, and sorted recommendations with expected savings and risks. Also generates CSV for spreadsheet analysis
6. Applies changes via the right tool — uses Qovery API for immediate changes or generates Terraform diffs if the user manages infrastructure as code. Asks the user which tool they prefer
7. Accounts for seasonal patterns — never right-sizes below seasonal peaks, recommends pre-scaling before known peaks, suggests post-peak review
8. Offers Kubecost deployment for ongoing real-time cost visibility, and suggests sharing the report with Qovery support for professional review

qovery-speedup

When you tell your AI agent "my deployments are slow, can you speed them up?":

1. Measures the full deployment pipeline using the V2 deployment history API — total time, per-stage, per-service, with structured durations
2. Breaks down sub-step timing by parsing deployment logs: git clone, Docker build, image push, pod scheduling, app startup, health check pass
3. Generates build runner usage reports (Grafana snapshots showing CPU, memory, network I/O during builds) using the deployment build usage API
4. Classifies each bottleneck as user-controllable or Qovery infrastructure — clear ownership for every slow step
5. Proposes Dockerfile optimizations — layer ordering, .dockerignore, multi-stage builds, build cache mounts, alpine base images, dev dependency removal (10+ patterns with before/after diffs)
6. Tunes health check configuration — right-sizes initial_delay_seconds, period_seconds, and failure_threshold based on actual measured startup time
7. Optimizes deployment stage parallelism — identifies independent services in serial stages that can run in parallel
8. Optimizes container image pull time — image size reduction, layer sharing between services, registry proximity
9. Generates diagnostic reports for Qovery support when the bottleneck is infrastructure-side (queue time, build runner capacity, registry push, Karpenter scheduling)

Usage

Just ask your AI agent:

I'm new to Qovery, help me get started

Deploy my application with Qovery

My Qovery deployment is failing, can you help?

The onboard skill acts as your personal cloud architect. The deploy skill handles application deployment. The troubleshoot skill diagnoses and fixes issues.

Prompts that trigger qovery-onboard:

• "I'm new to Qovery, help me get started"
• "Set up Qovery for my organization"
• "Help me onboard onto Qovery"
• "I have an existing Kubernetes cluster, can I use Qovery?"
• "I'm migrating from Heroku to Qovery"
• "Configure Qovery for my company"

Prompts that trigger qovery-deploy:

• "Deploy my application with Qovery"
• "Set up Qovery for my project"
• "Deploy this to Kubernetes with Qovery"
• "Create a Qovery Terraform configuration for my app"

Prompts that trigger qovery-troubleshoot:

• "My deployment is failing"
• "My app is crashing on Qovery"
• "Why is my application not working?"
• "My database connection is failing"
• "My app is slow / out of memory"
• "My cluster is not responding"

Prompts that trigger qovery-optimize:

• "Optimize my Qovery costs"
• "My cloud bill is too high"
• "Right-size my applications"
• "Are my services over-provisioned?"
• "How much is my infrastructure costing me?"
• "Generate a cost report"

Prompts that trigger qovery-speedup:

• "My deployments are slow"
• "Speed up my Qovery deployment"
• "Why does my deployment take so long?"
• "Optimize my build time"
• "My Docker build is slow"
• "My app takes too long to start"

Prompts that trigger qovery-preview:

• "Create a preview environment for the current branch"
• "Create a PR environment for PR-123"
• "Set up preview environments for my project"
• "I want to test my PR in isolation"
• "Preview this branch on Qovery"
• /qovery-preview (slash command)

Prerequisites

Before deploying, you need:

1. A Qovery account — sign up at console.qovery.com
2. A Qovery API token — generate at Organization Settings > API Tokens (or let the skill generate one via the CLI)
3. A Kubernetes cluster — AWS EKS, GCP GKE, Azure AKS, or Scaleway Kapsule
4. A git repository connected to Qovery (GitHub, GitLab, or Bitbucket)

New account? The skill handles cluster creation from scratch — including cloud provider credential setup (AWS CloudFormation, GCP Cloud Shell, Azure Cloud Shell), cluster provisioning via the Qovery Console, API, or Terraform, and waiting for the cluster to be ready. It supports all four cloud providers.

Supported Frameworks

The skill includes production-ready Dockerfile templates for:

If your framework is not listed, the agent will create a custom Dockerfile based on your project structure.

What the Skills Cover

qovery-onboard

qovery-deploy

qovery-troubleshoot

qovery-optimize

qovery-speedup

qovery-preview

Deployment Methods

The skill supports two deployment paths — the user chooses which one:

CLI + API (Quick Start)

Best for development and staging. Uses the Qovery CLI for monitoring and the REST API (https://api.qovery.com) for creating resources. Fast to set up, no files to commit.

Terraform Provider (Recommended for Production)

Creates a qovery.tf file that defines your entire infrastructure as code. Reproducible, version-controlled, CI/CD-friendly. Uses the Qovery Terraform Provider (qovery/qovery v0.54+).

Manual Installation

If you prefer not to use the install script, copy the skill folders manually:

git clone https://github.com/Qovery/qovery-skills.git
cd qovery-skills

# Global install (pick the paths for your tools)
mkdir -p ~/.claude/skills && cp -r qovery-onboard qovery-deploy qovery-troubleshoot qovery-optimize qovery-speedup qovery-preview ~/.claude/skills/
mkdir -p ~/.config/opencode/skills && cp -r qovery-onboard qovery-deploy qovery-troubleshoot qovery-optimize qovery-speedup qovery-preview ~/.config/opencode/skills/
mkdir -p ~/.agents/skills && cp -r qovery-onboard qovery-deploy qovery-troubleshoot qovery-optimize qovery-speedup qovery-preview ~/.agents/skills/

# Install the /qovery-preview slash command (OpenCode)
mkdir -p ~/.config/opencode/commands && cp qovery-preview/commands/qovery-preview.md ~/.config/opencode/commands/

# Or project-local install
mkdir -p .claude/skills && cp -r qovery-onboard qovery-deploy qovery-troubleshoot qovery-optimize qovery-speedup qovery-preview .claude/skills/

Verify the skills are discovered by checking if your tool lists all six Qovery skills.

Links

• Qovery Documentation
• Qovery Console
• Qovery CLI Reference
• Qovery API Reference
• Qovery Terraform Provider
• Real-World Example (Doktolib)
• Agent Skills Standard

License

MIT