# Advanced prompt engineering for journalists — starter kit

A starter repository for a Knight Center course on advanced prompt engineering for journalists, giving students a working Claude Code workspace with worked examples of context files, memory, skills, commands, hooks and a staged pipeline.

This is the workspace students fork and fill in across all four modules of the Knight Center MOOC on using AI coding tools from the command line for journalism, taught by Joe Amditis of the Center for Cooperative Media at Montclair State University. By the end of the course it becomes a working AI-assisted journalism toolkit built around the student's own beat.

Contents include a CLAUDE.md context-file scaffold with teaching comments and a fully worked example, memory and lessons files demonstrating recall patterns, real Claude Code skills with YAML frontmatter, slash commands for beat briefings, claim checks and cross-model review, a confirmation hook, a minimal plugin skeleton, staged pipeline scripts following a fetch-clean-process-save pattern, a GitHub Actions weekly-digest workflow demonstrating a cloud-detection pattern, a worked mini example with synthetic data, fictional sample journalism documents for practice, and stub files for students using Codex or Gemini instead of Claude Code.

Built with Python and Node.js 20 or newer, requiring one of Claude Code, Gemini CLI or another supported AI coding CLI.

Website: https://mooc.amditis.tech/starter-kit/