Muscle memory for your AI workflow. That's what I've been building.

Muscle memory for your AI workflow. That's what I've been building.

The idea: stop losing your AI setup every time you move between workstations. Everything — skills, MCP server configs, environment templates, OAuth scripts — lives in a repo. Clone it, run setup.sh, and you're back to full capability in minutes.

The skills are where it gets fun. I've built slash commands for the stuff I do repeatedly: /linkedin-post to draft and publish posts like this one, /splunk-dashboard-gen to generate a full Dashboard Studio dashboard with an AI-generated background image and deploy it live, /new-post-andrewriley-info to write and publish Hugo blog posts, /webex-update for end-of-session team updates, and /repo-status to check branch sync across local and remote at a glance.

MCP servers tie it together — Splunk for search and query, GitHub for repo ops, HuggingFace for image generation, plus Gmail, Google Calendar, and Slack via claude.ai's OAuth.

Recent projects that came out of this: a live Splunk Dashboard Studio pipeline with AI wallpapers, a scripted input app to ingest NSW traffic camera data into Splunk, and a skills system with CI/CD guard rails (Gitleaks + ShellCheck on every push).

Turns out treating your AI config like production infrastructure — versioned, tested, documented — pays off pretty quickly.

Repo is open if you want to take a look or steal ideas: https://lnkd.in/gYaetF_h

What does your Claude or AI tooling setup look like? Are you scripting repeatable workflows yet?

#Claude #AI #HomeLab #Splunk