cert-prep

A terminal UI application for practicing certification exam questions. Load a question bank from a JSON file, answer multi-select questions with immediate color-coded feedback, and see your final score.

Installation

Linux / macOS

curl -fsSL https://raw.githubusercontent.com/waveFrontSet/cert-prep/main/install.sh | sh

This installs the latest release to ~/.local/bin. To choose a different directory or pin a version:

curl -fsSL https://raw.githubusercontent.com/waveFrontSet/cert-prep/main/install.sh | sh -s -- -b /usr/local/bin v0.5.0

Note: the prebuilt macOS binary is Apple Silicon (aarch64) only. On Intel Macs, build from source.

Windows

powershell -ExecutionPolicy Bypass -c "irm https://raw.githubusercontent.com/waveFrontSet/cert-prep/main/install.ps1 | iex"

This installs cert-prep.exe to %LOCALAPPDATA%\Programs\cert-prep (override with the CERT_PREP_INSTALL_DIR environment variable) and prints a hint if that directory is not on your PATH.

Manual download

Grab the archive for your platform from the releases page, extract it, and put the cert-prep binary somewhere on your PATH.

Usage

cert-prep <config.json> [-n N] [-w CATEGORY:WEIGHT ...]

On the first run, you must provide a path to a question bank JSON file:

cert-prep ./my-questions.json

The config is then registered locally. On subsequent runs, you can omit the path and pick from previously used configs via an interactive selector:

cert-prep

The selector lists all previously used question banks sorted by most recently used.

Options

Flag	Description
<config.json>	Path to a question bank JSON file (required on first use)
-n N, --sample-amount N	Override the number of questions to sample
-w CATEGORY:WEIGHT, --weight CATEGORY:WEIGHT	Set category weight for stratified sampling (repeatable)
--version	Print the version and exit

Examples

# First run: load a question bank
cert-prep ./aws-sa-questions.json

# Subsequent runs: select from previously used configs
cert-prep

# Sample only 10 questions
cert-prep ./aws-sa-questions.json -n 10

# Emphasize a category
cert-prep ./aws-sa-questions.json -w "AWS Storage:3" -w "AWS Security:2"

Controls

Key	Action
Space	Toggle answer selection
Enter	Submit answer / Next question
Arrow keys / j / k	Navigate answer choices
h / l	When reviewing, travel to previous / next question
a	When reviewing, request an AI explanation (see below)
Ctrl-f / Ctrl-b	Scroll the explanation page-wise down / up
e	When finished, export a report of your results
q / Esc	Quit

Mouse clicks on checkboxes and buttons are also supported (not in all terminals).

After submitting an answer, selections are color-coded:

• Green [+] -- correct selection
• Yellow [O] -- missed correct answer
• Red [X] -- incorrect selection

AI explanations (optional)

With an API key, pressing a while reviewing an answer streams an AI-generated explanation of the correct answer (and why the incorrect choices are wrong) into the TUI:

export GEMINI_API_KEY=your-key
cert-prep

Without the environment variable, the feature is simply disabled.

By default, explanations use the gemini-2.5-flash model via Gemini's OpenAI-compatible endpoint. Model, endpoint, and system prompt can be overridden in settings.json (see Configuration files):

{
  "aiModel": "gemini-2.5-flash",
  "aiBaseUrl": "https://generativelanguage.googleapis.com/v1beta/openai",
  "aiSystemPrompt": "You are a friendly, concise mentor ..."
}

All keys are optional; any OpenAI-compatible endpoint should work.

Trophies

You can earn trophies for feats such as answer streaks, fast answers, or a flawless session. Trophies are tracked per question bank and shown in the TUI when earned.

Configuration files

State lives in the XDG config directory (~/.config/cert-prep/ on Linux/macOS, %APPDATA%\cert-prep\ on Windows):

File	Purpose
registry.json	Previously used question banks (feeds the config selector)
settings.json	Optional AI explanation settings
trophies/	Earned trophies, one file per question bank

Question bank format

Question banks are JSON files with the following structure:

{
  "title": "My Certification Exam",
  "sample_amount": 10,
  "category_weights": {
    "Category A": 2,
    "Category B": 1
  },
  "questions": [
    {
      "text": "Which option is correct?",
      "answer_choices": ["Option A", "Option B", "Option C"],
      "correct_answer": [1],
      "category": "Category A"
    }
  ]
}

Field	Description
title	Display name for the question bank
sample_amount	Number of questions to randomly sample per session
category_weights	Relative weights for stratified sampling across categories
questions[].text	The question prompt
questions[].answer_choices	List of answer options
questions[].correct_answer	Zero-indexed list of correct answer indices
questions[].category	Category label for the question

Building from source

Requires GHC and Cabal. Uses hpack (package.yaml).

cabal build
cabal run cert-prep -- ./config.json

Development

The development environment is managed with devenv. It provides GHC, Cabal, Haskell Language Server, linters, and git hooks.

devenv shell

Entering the shell automatically installs and configures the git hooks (hpack, fourmolu, hlint, yamllint).

Common tasks:

devenv tasks run cert-prep:build   # build using cabal.fast.project
devenv tasks run cert-prep:test    # run test suite
devenv tasks run cert-prep:lint    # run hlint

For AI explanations during development, set GEMINI_API_KEY in a .env file (managed via Secretspec).