Build real projects with a localized AI agent.

Gihanga brings local-agent project building to Rwandan learners, makers, schools, and teams — with a desktop app, CLI workflow, and local-first coding assistant.

folder ~/projects/gihanga-app
robot_2

Muraho, let's build your project.

I see you're setting up the initial CLI structure. I can generate the base commands for `gihanga tangira` and set up the interactive terminal UI. Shall I draft the Commander setup first?

person

Yes, let's start with Commander. Ensure we have local file writing permissions configured.

robot_2

Drafting `index.ts`...

import { Command } from 'commander';
import { setupLocalEnv } from './lib/env';

const program = new Command();

program
  .name('gihanga')
  .description('Localized AI Agent CLI for Rwanda')
  .version('0.1.0');

program.command('init')
  .description('Initialize a new Gihanga project workspace')
  .action(async () => {
    await setupLocalEnv();
    console.log('Workspace ready. Agent standing by.');
  });

Desktop + CLI, not another browser IDE

Built for deep work. Gihanga runs where your code lives, giving the agent direct but safe access to your local environment.

Gihanga project workspace illustration
folder_open

Local project folders

No need to upload zips or copy-paste code back and forth. The agent reads and writes directly to your local workspace, respecting your Git setup.

autorenew
psychology

Agent-guided build loops

Describe your goal. The agent breaks it down, writes the code, runs the tests, and iterates until the feature works. You stay in the loop to review and approve.

terminal

Terminal-first control

Prefer the command line? Run gihanga umukozi:rimwe alongside your existing tools for seamless, keyboard-driven AI assistance.

Made for Rwanda's builders

school

Learners

Stop getting stuck on config. Focus on logic while the agent explains local syntax and best practices.

domain

Schools

Deploy a consistent, reliable AI assistant across lab computers without worrying about individual API keys.

groups

Teams

Standardize your workflow. Share agent prompts and project contexts across your development team.

rocket_launch

Founders

Prototype faster. Go from idea to MVP quickly by leveraging an AI partner that writes boilerplate for you.

How Gihanga works

1

Open a folder

Point Gihanga to any local directory to set context.

2

Describe the goal

Chat naturally about what you want to build or fix.

3

Let the agent plan

Gihanga generates a step-by-step technical execution plan.

4

Review code

Approve file changes before they are written to disk.

5

Ship and learn

Run the code locally and learn from the agent's logic.

Early Access Preview

Terminal native

bash - gihanga-cli
gihanga ubufasha
added 42 packages in 3s
✨ Done in 3.14s.
gihanga tangira
[Gihanga] Initializing local workspace...
[Gihanga] Checking local git status... Clean.
[Gihanga] Agent ready.
gihanga umukozi:kurikirana

Roadmap

check_circle Desktop prototype
DONE
sync CLI runner
IN PROGRESS
schedule Local language prompts
PLANNED
schedule School lab mode
PLANNED
cloud_off Cloud sync optional
Future

Start building with Gihanga

Join the early access program to get your hands on the localized AI agent built for Rwanda's tech ecosystem.

Join early access