chrisblattman/claudeblattman

Claude Blattman – A No‑Code Guide to Building AI Workflows

What it is – An open‑source, markdown‑based resource that teaches non‑developers (especially academics and busy professionals) how to set up and run AI‑powered desktop agents using Anthropic’s Claude platform. The repo bundles documentation, a ready‑to‑install Claude plugin (the starter‑kit), and optional Python‑based “Codex” starter packs that provide pre‑written skills and agents.

Key parts of the repo

• docs/ – MkDocs‑Material site source that powers the public website claudeblattman.com.
• plugins/starter-kit/ – The Claude plugin that users add from the Claude Code marketplace. It contains the “Desktop Agent Starter Kit” which wires up basic chatbot, transcription, and prompt‑engineering capabilities.
• codex-starter/ – A small Python installer (install-codex-starter.py) that drops additional example skills into a local Claude Code environment.
• skills/ & agents/ – Collections of ready‑made markdown skill files and agent definitions that can be imported into Claude Code without writing code.
• templates/ – Boilerplate files (e.g., CLAUDE.md, goals.yaml) that help users create their own custom agents.

How you use it

1. Install Claude Code (Anthropic’s desktop‑agent IDE) and sign in.
2. In Claude Code run the marketplace command:

   /plugin marketplace add chrisblattman/claudeblattman
   /plugin install starter-kit@claudeblattman
   

   This pulls the starter‑kit plugin into your workspace.
3. Optionally run the Python installer to add the “Codex starter pack”:

   git clone https://github.com/chrisblattman/claudeblattman.git
   cd claudeblattman/codex-starter
   python3 install-codex-starter.py install
   

4. Follow the on‑site “Start Here” guide to configure a desktop agent, add skills, and begin building more complex, self‑improving workflows.

Who it’s for – Researchers, professors, and any professional who wants AI assistance (email drafting, meeting summarisation, etc.) but does not want to write code. Everything is driven by markdown files and Claude’s natural‑language configuration, so the learning curve is low.

License – MIT, allowing free reuse and modification of the docs, plugins, and skill templates.

Why it matters – It lowers the barrier to creating functional AI agents on a personal computer, turning Claude from a pure chat model into a programmable assistant that can run custom “skills” and orchestrate multi‑step workflows without traditional software development.