Skip to content
Kindgi

Kindgi - a full-stack AI agent framework and runtime.

Your coding agent writes tools, agents, flows and guardrails in your app's own code. Kindgi runs them in production, durably, on your own infrastructure or hosted by us (private preview). It records every step, so you can audit what they did, trace each answer and its cost, and keep making them better.

Start with your coding agent

Section titled “Start with your coding agent”

Paste this prompt into your coding agent: Claude Code, Cursor, Codex or any other. It takes you from nothing to a first agent that answers. Along the way, you sign in, copy two values and answer one question.

Set up Kindgi for me: read https://docs.kindgi.com/start/agent.md and follow it step by step. Tell me whenever you need me to do something, and wait for me.

What your coding agent follows · How it knows Kindgi

Quickstart

Section titled “Quickstart”

From an empty folder to an agent that answers: you run one command, and your coding agent does the rest. Quickstart

Rather write it yourself? A first pack by hand, running on your machine in ten minutes: TypeScript, Python, Java or Scala.

What you need

Section titled “What you need”
  • Either Node or Python. Which one depends on your pack's language:
    • TypeScript: Node 22.12 or later.
    • Python: Python 3.11 or later, and uv (or Poetry, or pip). The kindgi CLI comes from PyPI with its own copy of Node.
    • Java or Scala (preview): a JDK 17 or later, and Node or uv to run the CLI. A Java pack brings Maven; a Scala pack needs sbt.
  • Docker (Docker Desktop, or the Docker engine on Linux). kindgi dev runs Kindgi's runtime in it on your machine.
  • Access to the runtime image: sign in at access.kindgi.com with GitHub, to get the image's pull credentials and a free non-production license key, for staging and CI.
  • A model: your model provider's API key, or an open model you serve yourself, such as Ollama on your machine, with no key.

Install has the details, and kindgi doctor checks what Kindgi needs.

Free on your machine, with support for local models

Section titled “Free on your machine, with support for local models”

Everything you build and test on your machine is free: kindgi dev runs the runtime in development mode, which needs no license key. Your agents can run on a model on your machine too, with no API key and no cost per call. Connect a model you serve shows how, with Ollama as the example.

Running Kindgi in production needs a commercial license. Check Licensing for more details.

The documentation

Section titled “The documentation”
What is Kindgi?Packs, the runtime, and how your app and your coding agent work with them.
Add Kindgi to an existing appYour app's own code as tools, in TypeScript, Python, Java or Scala.
TutorialsBuild a support desk, step by step, in TypeScript or Python.
GuidesHow to do one thing: tools, agents, models, flows, runs, webhooks, approvals, secrets.
ConceptsHow Kindgi works: packs, runs and the journal, the security model, licensing.
DeployRun Kindgi in your own infrastructure, or let us run it for you on Kindgi Cloud (private preview).
ReferenceEvery API route, SDK export, CLI command, setting and event.
ContributingHow the SDK repository is organized and the rules for changes.