-
Notifications
You must be signed in to change notification settings - Fork 2
AI Tools Claude Code Setup
Pick the surface that matches your work.
- Terminal, if you already work in command lines
- Desktop app, if you want a standard Mac or Windows window
- VS Code extension, if your work lives inside an editor
Everything below gets easier once the install is right.
Claude needs one onboarding document before it can work properly.
- Create
~/CLAUDE.mdas the file Claude reads before each session - Add your role, voice, banned words, defaults, and output rules
- Treat it like the briefing note you would give a new hire
Without context, every session starts from zero.
Corrections should become system rules, not one-off feedback.
- Save repeated corrections as separate
.mdfiles - Store project memory inside
.claude/projects/{project}/memory/ - Index preferences, past mistakes, and key decisions in
MEMORY.md
This is how the same mistake stops happening twice.
A skill turns a repeated prompt into a single reusable command.
- Use
/your-skillinstead of pasting the same 200-word prompt each time - Connect the tools your work depends on, such as Notion, Gmail, and Drive
- Lock the output format so every run produces consistent results
This is where Claude stops feeling like chat and starts feeling like a workflow you can rely on.
One agent should not do every job.
- Strategist uses Opus to analyse the task
- Builder uses Sonnet to produce the output
- QA Gate rejects anything scoring below 95/100
The principle is separation of concerns: one part thinks, one part builds, and one part checks the work before it reaches you.
Routines move work off your machine and into the background.
- Schedule daily runs in Anthropic's cloud
- Set weekly outputs to drop finished work automatically into Notion
- Configure the schedule once, then stop managing it manually
This is the difference between an AI user and an AI operator. A user asks Claude for help. An operator builds the system Claude runs.
- Claude Code docs - docs.anthropic.com/en/docs/claude-code - Official reference for installation, slash commands, and configuration
- Anthropic GitHub - github.com/anthropics - Official repos including the Claude Code CLI and prompt engineering guides
- Anthropic Prompt Library - docs.anthropic.com/en/prompt-library - Curated prompts for common use cases
- Model Context Protocol (MCP) - modelcontextprotocol.io - Standard for connecting Claude to external tools and data sources
- All About AI - https://www.youtube.com/@AllAboutAI - Deep dives into Claude models, API usage, and agentic pipelines; consistently high quality and kept up to date
- IndyDevDan - https://www.youtube.com/@IndyDevDan - Developer-focused Claude Code tutorials covering terminal usage, CLAUDE.md setup, and multi-agent patterns
- Cole Medin - https://www.youtube.com/@ColeMedin - Practical AI agent builds using Claude with a focus on pushing the limits of what agentic workflows can do; uploads at least twice a week
- Matt Wolfe - https://www.youtube.com/@mreflow - Covers the latest Claude releases, tool comparisons, and real-world use cases in a clear and accessible style; good for staying current
- Liam Ottley - https://www.youtube.com/@LiamOttley - Focused on building AI automation systems and agents with Claude; useful for understanding how to operationalize AI workflows at scale
- Anthropic on YouTube - https://www.youtube.com/@anthropic-ai - Official channel with model announcements, research walkthroughs, and developer demos