get-started

name	get-started
description	Mandatory initialization workflow for new Keboola Python components — use this skill before writing any code. Contains the exact cookiecutter command, Keboola-specific post-template cleanup steps, KBC_DATADIR directory structure, data/config.json format, and first commit format that are easy to get wrong without this reference. Invoke whenever the user is starting from a fresh or empty repository, has never built a Keboola component before, wants to run the cookiecutter template, or says things like "new extractor for X", "build ex-something from scratch", "fresh repo", "empty git repo", "brand new component", "never done this in keboola", "initialize a component", "scaffolding a new component", "just initialized the git repo", "nothing set up yet". Not for existing components that already have src/component.py.
metadata	{"tools":"Bash, Read, Write, Edit, Glob, AskUserQuestion","model":"sonnet","color":"green"}

Initialize New Keboola Component

Before writing any code, check whether the component has already been initialized.

Step 1: Detect state

Check for signs of an existing cookiecutter setup:

ls src/component.py component_config/ pyproject.toml 2>/dev/null

• If all present → skip initialization, hand off to @develop-component immediately
• If missing or empty repo → proceed with initialization below

Step 2: Gather component info

You may already know the component details from the conversation. If not, ask:

• Component name — kebab-case, no type suffix (e.g. ex-salesforce, not salesforce-extractor)
• Component ID — vendor.component-name format (e.g. keboola.ex-salesforce)
• Type — extractor, writer, or application
• Short description — one sentence of what it does

Naming rules:

• Never include "extractor", "writer", or "application" in the name itself
• Use prefix convention: ex- for extractors, wr- for writers

Step 3: Run cookiecutter

# Install if missing
which cookiecutter || pip install --user cookiecutter

cookiecutter gh:keboola/cookiecutter-python-component

Run interactively so the user can confirm the values. The template creates src/, component_config/, tests/, .github/workflows/, Dockerfile, pyproject.toml, and data/ with example files.

Step 4: Clean up and configure

# Remove the generic example files that come with the template
find data -type f -delete

# Recreate the expected directory structure
mkdir -p data/in/tables data/in/files data/out/tables data/out/files

Then create data/config.json with realistic example parameters based on what the component actually does — not a generic placeholder. Read component_config/configSchema.json if it already has content, and mirror the required fields with plausible example values. Use # prefix for sensitive fields (e.g. "#api_key").

Step 5: Initial commit

git add -A
git commit -m "feat: initialize component from cookiecutter template"

Step 6: Hand off

Once initialized, pass to @develop-component to start the implementation. Share what you know about the component's purpose so it has context.

---

Key resources

• Cookiecutter template: gh:keboola/cookiecutter-python-component
• Developer Portal: https://components.keboola.com/
• Component tutorial: https://developers.keboola.com/extend/component/tutorial/