Jennifer Nguyen

Bootwitch

Scientist building AI and research tools
10+ years in immunology research · Building with AI since 2024
Bootwitch / Interactive demo
Simulated session
Files
Terminal
Welcome to Bootwitch.
Projects / bootwitch-cli-architecture

Bootwitch CLI Scaffold — Architecture

bootwitch-clidevtools-and-workflows

There are two layers: the toolkit that creates a project, and the project it leaves you with.

README · ARCHITECTURE · TECHNICAL · source

Bootwitch toolkit and generated projectBootwitch toolkit and generated project

The toolkit

PieceWhat it does
bin/bootwitchFinds the installation and starts the dispatcher
lib/bootwitch/core.shHandles commands, prepares templates, and guides project creation
lib/bootwitch/config.shReads the project's metadata fields
lib/bootwitch/publish.pyGives a prepared project its final directory name
templates/base/Supplies the common starting files
templates/shell/Adds the shell workflow, wrappers, modules, tests, and documentation builder

Creating a project

The direct init command and guided summon setup follow the same path. Bootwitch checks the inputs, prepares a directory beside the destination, combines the templates, fills in project information, and sets the executable permissions. It can initialize local Git as part of that preparation.

Then the publication helper renames the completed directory to its final name. If that final step needs attention, the prepared project remains available and the tool reports where it is. The technical page explains the write behavior in more detail.

Bootwitch project creation flowBootwitch project creation flow

Inside the generated project

FolderWhat you'll find
wrappers/Friendly entry points, including the greeting and README builder
scripts/Workflows such as new-script generation
modules/Shared path, logging, platform, permission, and documentation behavior
src/The starter application script
tests/Checks for the generated project
.bootwitch/Project metadata and the reusable script template

The generated project can run on its own. Its path helpers locate the project, and its wrappers lead into the scripts and modules that do the work.

From a script to its README entry

A new script receives an annotated header. The generator creates it, then calls the README wrapper in a separate Bash process. The documentation module reads the metadata and marked comments, builds the reference, and updates the generated section of the README.

Human-written text stays around that section. If the script is ready but the README refresh needs another try, the tool reports those two outcomes separately.

The source section has excerpts for the entry point, metadata reader, publication helper, and documentation handoff.

The technical page explains how the toolkit checks its own output. The source section follows the entry point, metadata reader, publication helper, and documentation handoff.

All projects