HomeLabCore

Independent software & AI lab

We build ideas.

Apps, games and AI development experiments.

The HomeLabCore flask: coding symbols move gently in its blue liquid while steam rises from the neck.

AppsClosed testing · Android

CookFrom

Cook with what you already have. Choose or photograph your ingredients and get an AI-assisted recipe you can adjust, save, and cook from step by step.

Open CookFrom

CookFrom promotional banner: the CookFrom logo with the tagline “Good food. Less waste. More ideas.”, four features — AI recipes from what you have, pantry, shopping list and preferences — and a phone showing an ingredient search, beside fresh tomatoes, basil and garlic.

AppsClosed testing · Android

Ops Planner

A weekly planner for sharing out a department's tasks, duties, and events among employees — with workload tracking, vacations, and schedules you can email. All data stays on your device.

Open Ops Planner

Ops Planner promotional banner in Dutch: the Ops Planner logo, five features — clear planning, managing employees, tasks and shifts, reports and export, overview and progress — and two phones showing the weekly planner and the employee editor.

AppsIn development · Android

AI Tutor

An Android app in development that helps students work through maths problems one step at a time — with Hint and Explain Differently — instead of revealing the final answer straight away.

Screens from the app's built-in sample mode.

Open AI Tutor

  1. 1 · Step
    The first tutoring step for the built-in sample equation.
  2. 2 · Hint
    Level 1 Hint in the built-in sample tutoring flow.
  3. 3 · Explain differently
    Explain Differently in the built-in sample mode; no live AI response.

GamesIn development · Web

Keyboard Trainer

A desktop-first typing game for children aged 6–12: ten keyboard challenges of increasing difficulty, with stars, XP and coins to earn on the way. Progress is stored only in the browser.

Open Keyboard Trainer

The first typing challenge with its highlighted target key.
The real ten-level quest map with its initial unlock state.

Lab · AI workflowsExperiment · Self-hosted

AgentLab

An experiment in AI development infrastructure: planning, building and review roles, with local inference behind a gateway that makes routing and validation explicit.

Open AgentLab

Current AgentLab gateway architecture: deterministic L0 tasks, L1/L2 local inference through Ollama, schema validation, bounded repair, and metadata-only telemetry. Guard rejection requests escalation without making a cloud call.
Watercolor concept art of an Antarctic expedition crossing a frozen landscape toward monumental ancient ruins in the mountains.

GamesConcept art · Not gameplay

At the Mountains of Madness

A game in active development. Its intended visual direction: watercolor, restrained psychological horror, and an Antarctic expedition dwarfed by monumental ancient ruins.

Open the project

Benchmarks

Benchmarks

Public comparisons of AI-assisted development workflows: which models and role combinations build what, how fast, at what cost, and how often they need repair.

How runs will be reported

In preparation · no results published

Each published run will fill these columns, with its setup, method and raw data.

  • Model
  • Planner · builder · reviewer
  • Time
  • Cost
  • Repair loops
  • Success rate

Lab log

Recent updates

About

A small, independent lab

HomeLabCore builds a short list of focused apps and experiments, keeps them maintained, and is clear about how each one handles your data.

More about HomeLabCore · Support · support@homelabcore.dev