Appearance
Adding Tools and Customizing
~5 minutes. Connect more tools, customize your config, and go beyond the starter skills.
Your workspace starts useful, but it gets powerful when Claude can see more of your world. Every tool you connect gives it more context to work with.
Adding More Tools (MCPs)
MCP (Model Context Protocol) is how Claude connects to external tools. Each MCP server gives Claude a new capability: searching docs, reading tickets, querying data, browsing code.
🔍
Search Tools
Enterprise search, Google Drive, Notion, Confluence. Lets Claude search your docs, wikis, and chat threads.
🎫
Project Trackers
Jira, Linear, GitHub Issues, Asana. Claude reads your tickets, sprint boards, and backlog.
🗄️
Data Warehouses
Snowflake, BigQuery, PostgreSQL. Run queries, check metrics, pull data for reports.
🔀
Code Repositories
GitLab, GitHub, Sourcegraph. Browse repos, read PRs, search across codebases.
📧
Google Workspace
Gmail, Calendar, Docs, Sheets. Read emails, check schedules, create documents.
💬
Messaging
Slack, Teams, Discord. Read channels, search threads, post updates.
Check What You Have
Ask Claude directly:
"What tools do you have access to? List everything that's connected."
Claude will enumerate every MCP server and its capabilities. If something's missing, ask Claude to help you set it up — it can read your config files and walk you through the connection.
Adding a New MCP
MCP servers are configured in your Claude Code settings. The exact method depends on your setup, but typically:
- Find the MCP server for the tool you want (check the MCP directory or your team's documentation)
- Add it to your config — Claude can help with this: "Help me connect a Jira MCP server"
- Restart Claude Code to pick up the new connection
- Verify by asking Claude to use the new tool
Customizing CLAUDE.md
Your CLAUDE.md file is Claude's instruction manual for working with you. It's a living document — not something you set once and forget.
What belongs in CLAUDE.md
Keep It Updated
When Claude gets something wrong — a name, a preference, a convention — tell it:
"You got that wrong. My weekly reviews go in
_System/, notMeetings/. Update CLAUDE.md so you remember."
Claude will edit the file directly. Over time, your CLAUDE.md becomes a precise instruction set tuned to how you work.
Beyond the Starter Skills
The setup gives you the basics: /process-meeting, /pkb-file, /note. But the system supports many more workflows:
📋
meeting-prep
Reads your KB to prepare briefings before meetings — attendee profiles, open action items, last discussion context.
✉️
email-draft
Drafts emails using your KB context — knows your projects, stakeholders, and communication style.
📊
stakeholder-update
Generates project updates from accumulated meeting notes, decisions, and action items.
🔬
data-check
Queries your data warehouse to verify assumptions, check metrics, or pull numbers for a brief.
Skills Are Just Markdown Files
Every skill lives in .claude/skills/ as a markdown file. You can:
- Read them to understand what they do
- Modify them to change behavior
- Write your own from scratch
Open any skill file in your editor. It's markdown with frontmatter (metadata) and instructions (what Claude should do). No special syntax, no compilation step. Edit, save, and it takes effect immediately.
Domain Packs (Coming Soon)
Domain packs are curated context bundles designed for specific team types:
- Product Management — sprint rituals, PRD templates, stakeholder mapping
- Engineering — incident response, architecture decisions, code review patterns
- Data Science — experiment tracking, analysis templates, metric definitions
- Design — research synthesis, design critique, handoff documentation
When available, domain packs pre-load relevant skills, CLAUDE.md sections, and folder conventions for your role. Think of them as a head start — you'll still customize everything to fit how you actually work.
Next: Workflows and Patterns — Common workflows, prompt templates, and when to create a skill vs. keep it as a pattern.