CLAUDE.md Setup Guide

## Projects: 
This project is an AI assistant helping manage daily notes, planning, and marketing for a local health clinic. 

## Before responding to anything — including the very first message —
silently read these files in this order:
1. ~/.claude/primer.md
   Short-term memory. Active projects, last session, what's next.
   Read it completely before doing anything.

2. ~/.claude/memory.md
   Long-term memory. Facts about me, my work, key decisions,
   important context that doesn't change often.

3. ~/.claude/memory/learnings/claude-cli.md
   Behavioural rules from past corrections. Apply every rule
   from the very first response. Do not wait to be reminded.

Rules for reading:
- Do it silently. 
- If a file doesn't exist yet, skip it and continue.
- You are responsible for
  keeping them updated per the rules below.

## Workflow
- Enter plan mode for any non-trivial task (3+ steps or architectural decisions). Stop and re-plan if something goes sideways.
- Use subagents liberally — offload research, exploration, and parallel analysis to keep the main context clean.
- After any correction do a self-improvement loop: note the pattern to avoid repeating it. Add to learnings.md  
- Never mark a task done without verifying it works.
- Bug reports: just fix them. Use logs, errors, and tests — don't ask for hand-holding.

When wrapping up a work session, update the "Last Session" section of primer.md.

**Learnings rule:** When user corrects something or says "remember this", write it immediately to `~/.claude/memory/learnings/claude-cli.md` using this format:

I want you to set up a personal memory and context system for my
Claude Code sessions. This system uses flat markdown files to give
you persistent memory across sessions without any hooks or
technical setup beyond this conversation.

Here is what I need you to build:

FOLDER STRUCTURE to create:
- ~/.claude/memory/learnings

FILES to create (with proper starter content):
- ~/.claude/CLAUDE.md
- ~/.claude/primer.md
- ~/.claude/memory.md
- ~/.claude/memory/learnings/claude-cli.md

Before creating any files, ask me these questions ONE AT A TIME
and wait for my answer before asking the next:

1. What is your name and what do you do for work?

2. What is the main project or goal you want Claude to help you
   with? (e.g. running a business, content creation, software
   development, managing a team, personal productivity)

3. What platforms or tools do you use regularly that Claude
   should know about? (e.g. Notion, social media handles,
   software tools, specific workflows)

After I answer all questions, do the following:

STEP 1: Create the folder structure using the Bash tool.

STEP 2: Create ~/.claude/CLAUDE.md with these exact sections:
- READ THIS FIRST block that instructs you to silently read
  primer.md, memory.md, and learnings/claude-cli.md at session
  start before responding to anything
- primer.md rules (short-term memory — update Last Session only,
  keep short)
- memory.md rules (long-term memory — append facts, never delete)
- learnings rules (behavioural — append corrections, never delete)
- Session wrap-up protocol (update primer, check learnings,
  check memory, confirm in one line)

STEP 3: Create ~/.claude/primer.md with this structure:
# Session Primer
## Active Projects
[filled from my answers]
## Last Session
Date: —
What we worked on: —
Completed: —
Next step: —
Open blockers: —
## Open Items
- (nothing yet)
## Notes

STEP 4: Create ~/.claude/memory.md with this structure:
# Long-Term Memory
Facts about me and my work that should persist indefinitely.
Claude appends here when important facts emerge. Never overwrites.
---
## About Me
[filled from my answers — name, work, goals, tools]
## Key Decisions
(none yet)
## Important Context
(none yet)

STEP 5: Create ~/.claude/memory/learnings/claude-cli.md with:
# Claude CLI — Learnings
Rules from past corrections. Append new entries at top. Never delete.
---

STEP 6: After all files are created, show me a summary of what
was built and tell me the three things I need to do to maintain
the system going forward.

Then add the following into the claude.md file according to my preferences too:

## Projects: 
This project is an AI assistant helping manage daily notes, planning, and marketing for a local health clinic. 

## Before responding to anything — including the very first message —
silently read these files in this order:
1. ~/.claude/primer.md
   Short-term memory. Active projects, last session, what's next.
   Read it completely before doing anything.

2. ~/.claude/memory.md
   Long-term memory. Facts about me, my work, key decisions,
   important context that doesn't change often.

3. ~/.claude/memory/learnings/claude-cli.md
   Behavioural rules from past corrections. Apply every rule
   from the very first response. Do not wait to be reminded.

Rules for reading:
- Do it silently. 
- If a file doesn't exist yet, skip it and continue.
- You are responsible for
  keeping them updated per the rules below.

## Workflow
- Enter plan mode for any non-trivial task (3+ steps or architectural decisions). Stop and re-plan if something goes sideways.
- Use subagents liberally — offload research, exploration, and parallel analysis to keep the main context clean.
- After any correction do a self-improvement loop: note the pattern to avoid repeating it. Add to learnings.md  
- Never mark a task done without verifying it works.
- Bug reports: just fix them. Use logs, errors, and tests — don't ask for hand-holding.

When wrapping up a work session, update the "Last Session" section of primer.md.

**Learnings rule:** When user corrects something or says "remember this", write it immediately to `~/.claude/memory/learnings/claude-cli.md` using this format:

Start by asking question 1 now.