devkit

A unified developer workflow toolkit for Rust, Python, and TypeScript projects.

Components

dev CLI

A single-binary Rust CLI for unified developer workflows:

• Config-driven task runner (pipelines: fmt, lint, type, test, fix, check, ci)
• Language scaffolding + provisioning (dev install)
• Git flows (dev git ...)
• Versioning (dev version ...)
• .env helpers (dev env ...)
• System setup (dev setup ...)
• Dockerized GPU dev containers (dev docker ...)
• Isolated research project scaffolding (dev research init)
• Review overlays and directory manifests (dev review, dev walk)

devkey (Windows only)

A hotkey-triggered popup for quick access to environment variables and dev tasks. Press Ctrl+; to open.

For the authoritative spec, see docs/spec.md.

Install

From crates.io

cargo install devkit-cli

From source

# Clone the repo
git clone https://github.com/bakobiibizo/devkit
cd devkit

# Install dev CLI
cargo install --path crates/dev

# (Windows only) Build devkey
cargo build --release -p devkey
# Binary at target/release/devkey.exe

Prebuilt binaries

Download from GitHub Releases.

Quick usage

Tasks and pipelines

dev list
dev run <task>

dev fmt
dev lint
dev type
dev test
dev fix
dev check
dev ci

dev all <fmt|lint|type|test|fix|check|ci>

Config

dev config
dev config check
dev config generate [PATH] --force

Language tooling

dev install [rust|python|typescript]
dev language set <name>

.env management

dev env [--raw]
dev env get <KEY>
dev env add <KEY> <VALUE>
dev env rm <KEY>

dev env profiles
dev env switch <PROFILE>
dev env save <NAME>

dev env check
dev env init
dev env template

dev env diff [REF]
dev env sync [REF]

Research management

# Scaffold an isolated, project-local research workspace
dev research init [DIR] --name <project-name> --package <python_package>

# Preview without writing files
dev --dry-run research init ./research/heu-exp-01

dev research init creates a clean-room research layout (project.yaml, experiments/, src/<package>/bindings/, reports/templates/, .harness/) and sets HARNESS_HOME=.harness so runs remain project-local and commit-friendly.

Docker workflow (GPU dev container)

This is designed for “build inside containers” workflows (including NVIDIA GPU containers).

1) Scaffold Docker files in your project

Run this in the project directory:

dev docker init

This generates:

• docker/Dockerfile.core
• docker-compose.yml
• .env (includes CORE_IMAGE, UID, GID)

2) Build the core image

dev docker build reads CORE_IMAGE from your project .env by default:

dev docker build

Override tag if needed:

dev docker build --image bakobiibizo/devkit:cuda13

3) Enter the container with an interactive shell

This is the primary “put me in the container” command:

dev docker dev

It runs:

• docker compose up -d --build
• then docker compose exec <service> bash -l

Options:

dev docker dev --service core
dev docker dev --no-up

4) Compose helpers

dev docker compose up build
# or detached:
dev docker compose up build -d

Setup system

dev setup

dev setup run <components...>
dev setup all

dev setup status
dev setup list
dev setup config

Review + walk

dev review [--output <path>] [--include-working] [--main]

dev walk [DIR] -o manifest.md --max-depth 10 --extensions .rs .py