tinyprompts 0.0.2

A dependency free prompts (not the LLM prompts) library built for CLI.

tinyprompts

A dependency-free interactive CLI prompts library for Python. Zero pip install dependencies -- just the stdlib.

tinyprompts does one job: collect user input interactively with a pretty TUI, then hand back a typed result. It does not replace argparse, click, or typer -- it complements them.

Note: The API may change before 1.0. Requires Python 3.9+ and a Unix TTY (macOS / Linux).

Dependency

None. Your only dependency is Python 3.9+.

Key Features

• Interactive prompts -- Text input (with regex validation), searchable multi-select, and yes/no confirmation, rendered in-terminal with keyboard navigation.
• Type-hint-first API -- Define prompts as a dataclass with Annotated fields. One dataclass, fully typed output.
• Bring your own CLI -- tinyprompts returns a dataclass (or dict). Wire it into argparse, click, typer, or nothing at all.
• Tiny footprint -- 4 source files, ~500 lines total. Easy to vendor, audit, or fork.
• (Not implemented) - Defer preload. Python has a slow start problem which make it bad for CLI. Instead of fighting it, start the user interactive part first, and load the heavy things in the background.

Quickstart

from dataclasses import dataclass, field
from typing import Annotated
from tinyprompts import Text, Confirm, MultiSelect, Choice, prompt

@dataclass
class ProjectSetup:
    name: Annotated[str, Text("Project name")] = "my-project"
    tools: Annotated[list[str], MultiSelect("Pick tools", choices=[
        Choice("ruff", "Linting"),
        Choice("pytest", "Testing"),
    ])] = field(default_factory=list)
    with_example: Annotated[bool, Confirm("Include example?")] = False

config = prompt(ProjectSetup)  # interactive TUI, returns ProjectSetup or None

Pairing with argparse

tinyprompts doesn't own your CLI -- you do. Build your parser normally, then pass CLI values as defaults so prompt() only asks for what's missing:

import argparse
from tinyprompts import prompt

parser = argparse.ArgumentParser()
parser.add_argument("--name", default=None)
parser.add_argument("--tools", nargs="*", choices=["ruff", "pytest"])
args = parser.parse_args()

defaults = {k: v for k, v in vars(args).items() if v is not None}
config = prompt(ProjectSetup, defaults=defaults)

The same pattern works with click, typer, or any framework that produces a dict of values.

Install

pip install tinyprompts