Skip to main content
đŸ”Ĩ

kindling

GitHub ActionsGitLab CI

Layers SDLC best practices into your project

Minimize the time from idea to prod. Designed for builders, Kindling layers SDLC best practices in, so your project — multi-agent architectures, microservices or any flavor of containerized apps — runs everywhere. Your project just works. In prod.

$brew install kindling-sh/tap/kindlingCopy
Get Started → Star on GitHub
terminal
$ kindling init ▸ Creating Kind cluster ✅ Cluster, registry, operator ready $ kindling analyze ✅ 2 Dockerfiles, 3 dependencies ℹī¸ Agent frameworks: LangChain ✅ Ready for 'kindling generate' $ kindling generate -k sk-... -r . 🤖 Provider: openai, Model: o3 ✅ Workflow written to dev-deploy.yml $ git push 🏗ī¸ Building → registry:5000/app:abc123 ✅ http://you-app.localhost $ kindling sync -d you-app --restart 🔄 Watching for changes... ✅ Synced 3 files → restarted

Everything you need for local dev staging

One operator. One CLI. From git push to running app on localhost.

Analyze Before You Build

kindling analyze checks your repo's readiness — Dockerfiles, dependencies, secrets, agent architecture, Kaniko compatibility — before you generate a single line of CI.

AI-Generated Workflows

kindling generate scans your repo and produces a complete GitHub Actions or GitLab CI workflow. Detects agent frameworks, MCP servers, inter-service calls, and secrets.

Two-Speed Dev Loop

Outer loop: git push → build → deploy. Inner loop: edit → sync → reload in under a second. Both run on your laptop, zero cloud CI minutes.

15 Auto-Provisioned Dependencies

Declare postgres, redis, kafka, elasticsearch, and 11 more. The operator provisions them and injects connection URLs automatically.

Explain On-Demand

kindling explain gives your coding agent concise, always-current workflow guidance (hot reload, dependencies, builds, secrets) — pulled on demand, not injected into every session.

Localhost to Staging — Multi-Tenant, GitOps-Ready

Dev on localhost with instant staging. Need OAuth callbacks? kindling expose creates a public HTTPS tunnel. Ready to ship? kindling snapshot --deploy graduates to a shared staging cluster with branch-scoped naming, then --render-prod-values hands off a production-ready Helm chart for your GitOps controller.

Secrets & Credentials Built In

Manage API keys with kindling secrets. Automatic detection during analyze and generate. Local backup survives cluster rebuilds.

How it works

1

Analyze

kindling analyze checks your project — Dockerfiles, dependencies, secrets, agent architecture — and tells you exactly what's ready.

→
2

Generate

kindling generate scans your repo and uses AI to produce a complete CI workflow with builds, deploys, and dependencies.

→
3

Dev Loop

git push builds and deploys.kindling sync gives you sub-second live reload. Iterate until it's right.

→
4

Snapshot & Deploy

kindling snapshot --deploy copies images to your registry and deploys to a shared, multi-tenant staging cluster — branch-scoped so concurrent PRs never collide, and non-interactive-safe so it can run from CI. Add--render-prod-values for a production-ready Helm chart your GitOps controller picks up at the git boundary.

Quick start

# Install
brew install kindling-sh/tap/kindling

# Bootstrap a local cluster
kindling init

# Register a CI runner (GitHub or GitLab)
kindling runners -u <user> \
  -r <owner/repo> -t <pat>

# Check your project's readiness
kindling analyze

# AI-generate a workflow for your app
kindling generate -k <api-key> -r /path/to/app

# Push and watch it deploy
git push origin main

# Start live sync
kindling sync -d <user>-<app> --restart

Prerequisites

  • Docker Desktop — container runtime
  • Kind — local Kubernetes clusters
  • kubectl — Kubernetes CLI

Supported languages

GoTypeScriptPythonJavaRustRubyPHPC#Elixir

Supported AI providers

OpenAI (o3, gpt-4o)Anthropic (Claude)