Skip to main content

Quickstart

Go from nothing to a running app in three phases: analyze → generate → dev loop.

1. Install & bootstrap​

brew install kindling-sh/tap/kindling
kindling init
tip

brew install installs kindling, kind, and kubectl automatically. kindling init creates a local Kubernetes cluster with a container registry, ingress controller, and the kindling operator — all in one shot.

2. Connect a CI runner​

kindling supports GitHub Actions and GitLab CI.

# GitHub (needs a PAT with repo scope)
kindling runners -u <user> -r <owner/repo> -t <pat>

# GitLab
kindling runners --ci-provider gitlab -u <user> -r <group/project> -t <token>

3. Analyze your project​

Before generating anything, check your project's readiness:

kindling analyze

This scans your repo and reports:

  • Dockerfiles found and Kaniko compatibility
  • Dependencies detected (Postgres, Redis, etc.)
  • Secrets and credentials your app needs
  • Agent frameworks (LangChain, CrewAI, etc.) and MCP servers
  • Build context alignment between Dockerfiles and workflow

Fix any issues it flags, then move to generate.

4. Generate a workflow​

kindling generate -k <api-key> -r .

AI-generates a complete CI workflow (.github/workflows/dev-deploy.yml or .gitlab-ci.yml). Detects services, languages, ports, dependencies, health checks, and secrets.

note

Works with OpenAI (default) or Anthropic (--ai-provider anthropic). Preview first with --dry-run.

5. Push and deploy​

git add -A && git commit -m "add kindling workflow" && git push

The runner picks up the job, Kaniko builds images in-cluster, the operator provisions dependencies, and ingress routes go live:

kindling status
curl http://<your-user>-my-app.localhost

6. Start the dev loop​

Now iterate without pushing to git:

# Sub-second live sync — edit, save, see changes instantly
kindling sync -d <your-user>-my-app --restart

# Or open the visual dashboard
kindling dashboard

When you stop sync (Ctrl+C), the deployment automatically rolls back to its pre-sync state.

Need a public URL?​

kindling expose

Creates an HTTPS tunnel instantly — useful for OAuth callbacks, webhooks, or sharing with teammates.

Try the demo app (optional)​

Don't have a project handy? Use the included microservices demo:

cp -r ~/.kindling/examples/microservices ~/kindling-demo
cd ~/kindling-demo
git init && git add -A && git commit -m "initial commit"
gh repo create kindling-demo --private --source . --push

This gives you a 4-service app (Go, Python, Node.js, React) with Postgres, Redis, and MongoDB — plus a pre-built workflow. No AI key needed.

The journey​

analyze → generate → dev loop → promote
   ↓          ↓          ↓          ↓
 readiness  workflow   push/sync  production
 check      via AI     iterate    (coming soon)

Every git push rebuilds and redeploys. kindling sync gives you sub-second iteration. No cloud CI minutes. No Docker Hub. No YAML by hand.

Next steps​

Want to...Guide
Manage API keys and secretsSecrets Management
Set up OAuth callbacksOAuth & Tunnels
Deploy without GitHub ActionsManual Deploy
See all 15 dependency typesDependency Reference
Understand the internalsArchitecture