How I built a second brain that runs my business.

What you need:

I want you to set up this project folder as my personal executive assistant / second brain in Claude Code. You're going to do this in 3 phases. Complete each phase fully before moving to the next. ## Phase 1: Create the folder structure First, initialize a git repo in this folder if one doesn't exist already. Then create the following template structure. Don't put any content in the files yet -- just create the skeleton with placeholder files where noted. ``` CLAUDE.md # Main brain file (we'll fill this in Phase 3) CLAUDE.local.md # My personal overrides (git-ignored) .gitignore # Ignore .env, CLAUDE.local.md, settings.local.json .claude/ settings.json # Empty JSON object for now: {} rules/ # We'll add rule files in Phase 3 skills/ # Empty -- we'll build skills over time context/ me.md # About me (filled in Phase 3) work.md # My business/work details (filled in Phase 3) team.md # My team (filled in Phase 3) current-priorities.md # What I'm focused on right now (filled in Phase 3) goals.md # Quarterly goals and milestones (filled in Phase 3) templates/ session-summary.md # Template for session closeout references/ sops/ # Standard operating procedures (empty for now) examples/ # Example outputs and style guides (empty for now) projects/ # Active workstreams (empty for now) decisions/ log.md # Decision log (append-only, empty for now) archives/ # Completed/outdated material (empty for now) ``` For empty directories, create a .gitkeep file inside them so git tracks them. For templates/session-summary.md, use this starter template: ```markdown # Session Summary **Date:** Focus: ## What Got Done - ## Decisions Made - ## Open Items / Next Steps - ## Memory Updates - Preferences learned: - Decisions to log: ``` For decisions/log.md, use this starter: ```markdown # Decision Log Append-only. When a meaningful decision is made, log it here. Format: [YYYY-MM-DD] DECISION: ... | REASONING: ... | CONTEXT: ... --- ``` For .gitignore: ``` .env CLAUDE.local.md .claude/settings.local.json node_modules/ ``` For CLAUDE.local.md: ```markdown # Local Overrides Personal preferences and overrides that don't get shared via git. Add anything here that's specific to your local setup. ``` ## Phase 2: Ask me onboarding questions Before filling in the context files, rules, and CLAUDE.md, interview me. Ask these questions one section at a time. Don't dump all questions at once -- go section by section and wait for my answers before moving on. ### Section 1: About You - What's your name? - What's your role/title? (e.g., CEO, freelancer, content creator, developer) - What's your timezone? - In one sentence, what do you do? - What's your #1 priority -- the thing everything else should support? ### Section 2: Your Business / Work - What's your company or business called? (or describe your work if you're not a business owner) - What are your products, services, or revenue streams? (List each with a one-liner) - Roughly how much revenue does each generate? (optional, but helps me prioritize) - What tools do you use day-to-day? (ClickUp, Notion, Slack, Google Workspace, etc.) - Do you have any MCP servers already connected to Claude Code? (If you don't know what this means, just say no) ### Section 3: Your Team - Do you have a team? If yes, how many people? - Who are the 2-3 key people I should know about? (name, role, when to loop them in) - Where does your team communicate? (ClickUp, Slack, etc.) - What's your biggest pain point with managing your team? ### Section 4: Priorities, Goals & Projects - What are the 3-5 things you're most focused on right now? - Are there any deadlines or time-sensitive items I should know about? - Do you have quarterly goals or milestones you're tracking? (If yes, list them. If you don't do formal goals, that's fine -- your priorities above will work.) - What active projects or workstreams are you managing right now? (These are specific initiatives, not ongoing responsibilities. E.g., "launching a new course", "hiring a VA", "redesigning the website") ### Section 5: Communication Preferences - How do you like information presented? (bullet points, detailed paragraphs, etc.) - Any writing pet peeves? (e.g., "never use emojis", "no em dashes", "keep it under 3 sentences") - What tone do you want internally (casual, professional, etc.)? - What tone do you want for external/public-facing content? ### Section 6: What Do You Want Help With? - What are the recurring tasks that eat up your time? (List as many as you want) - What would you hand off to an assistant first if you could? - Are there any specific workflows you want to automate or templatize? ## Phase 3: Build out the files Based on my answers, fill in all the files: ### Context files - context/me.md -- My profile based on Section 1 answers - context/work.md -- My business/work details based on Section 2 answers - context/team.md -- Team structure based on Section 3 answers (if I have a team; if solo, note that and skip this file) - context/current-priorities.md -- Priorities from Section 4, dated today - context/goals.md -- Quarterly goals and milestones from Section 4. If no formal goals, use their top priorities as informal goals. Date it with the current quarter (e.g., "Q2 2026"). Include a note at the top: "Update this file at the start of each quarter." ### Projects If the user listed active projects in Section 4, create a folder for each inside projects/. Each project folder gets a brief README.md with: - One-line description of the project - Current status (active, planning, on hold) - Key dates or deadlines mentioned If no specific projects were mentioned, leave projects/ empty. ### Rule files in .claude/rules/ Create rule files based on my communication preferences (Section 5). Suggested files: - communication-style.md -- Writing tone, formatting preferences, pet peeves - Any other domain-specific rules that emerged from my answers (e.g., if I mentioned specific tool conventions, content guidelines, etc.) Keep each rule file focused on ONE topic. Don't create more than 3-4 rule files to start. ### CLAUDE.md (the main brain file) This is the most important file. Keep it UNDER 150 lines. It should contain: 1. One-line identity -- Who you are to me (e.g., "You are [Name]'s executive assistant.") 2. Top priority -- The #1 thing everything supports 3. Context imports -- Use @context/me.md, @context/team.md, @context/goals.md, etc. to reference files instead of repeating their content 4. Tool integrations -- What tools are connected (ClickUp, etc.) 5. Skills directory -- Point to .claude/skills/ and explain how skills work 6. Decision log -- Point to decisions/log.md and explain append-only convention 7. Memory -- Add a section explaining how memory works: - "Claude Code maintains a persistent memory across conversations. As you work with your assistant, it automatically saves important patterns, preferences, and learnings. You don't need to configure this -- it works out of the box." - "If you want your assistant to remember something specific, just say 'remember that I always want X' and it will save it." - "Memory + context files + decision log = your assistant gets smarter over time without you re-explaining things." 8. Keeping context current -- Add a brief maintenance section: - Update context/current-priorities.md when your focus shifts - Update context/goals.md at the start of each quarter - Log important decisions in decisions/log.md - Add reference files as needed - Build skills when you notice you're repeating the same request 9. Projects -- Point to projects/ and explain that active workstreams live there 10. Templates -- Point to templates/ 11. References -- Point to references/ 12. Archives rule -- Don't delete, archive Do NOT put communication style rules in CLAUDE.md -- those go in .claude/rules/communication-style.md. Do NOT repeat context from the context files -- just import them with @. Do NOT include a "Session Start Protocol" that tells Claude to read a bunch of files -- the import system and rules handle that automatically now. ### Skills directory Leave .claude/skills/ empty. Don't create any skills yet. Just make sure CLAUDE.md mentions that skills live there and explains the pattern: - Each skill gets a folder: .claude/skills/skill-name/SKILL.md - Skills are built organically as recurring workflows emerge - Include a note about the answers from Section 6 (what they want help with) as a "Skills to Build" backlog in CLAUDE.md or a separate file ## Final Step After everything is created: 1. Show me a tree view of every file and folder you created 2. Show me a brief summary of what's in each file (one line per file) 3. List the "Skills to Build" backlog based on my Section 6 answers -- these are the workflows we'll turn into skills over time 4. Show me this maintenance cheat sheet: Keeping Your Assistant Sharp - Weekly: Nothing required. Auto-memory handles daily learnings for you. - Monthly: Glance at context/current-priorities.md. If your focus has shifted, update it. - Quarterly: Update context/goals.md with new goals and milestones. - As needed: Log decisions in decisions/log.md. Add reference files. Build new skills. - Pro tip: If you want your assistant to remember something permanently, just tell it: "Remember that I always prefer X." It will save it across all future conversations. 5. Create the first git commit with all the files 6. Ask me if I want to build any of those skills right now ## Important Rules for You (Claude) - Do NOT create any skills yet. The skills directory should be empty after setup. - Keep CLAUDE.md UNDER 150 lines. If it's getting long, you're putting too much in it. - Use @ imports (e.g., @context/me.md) in CLAUDE.md instead of repeating information. - One rule file = one topic. Max 3-4 rule files to start. - Ask onboarding questions ONE SECTION AT A TIME. Wait for my response before asking the next section. - If I say "skip" or "I'll fill that in later" for any section, create the file with a placeholder and move on.