Claude Code tips (from Claude Code creator)

cc_tips.md

Claude Code | Tips (from its creator)

set-up

• /allowed-tools: customize tool perms
• /terminal-setup: enable shift + enter to insert new lines
• /install-github-app: Tag @claude on issues and PRs
• /theme: enable light/dark mode
• /config: turn on notifications (turn on MacOS dictation)

tips

• plan mode
  • hit shift + tab (twice) to activate
  • use Opus for Plan Mode: /model to select
• have CC use Plan Mode to write an implementation plan or specification (to save time and resume context)
  • this is useful to not eat up tokens, but also to go back and forth such that it has clarity on what you're building
• /compact: clear conversation history but keep a summary in context.
• answer questions about code
  • "Why does recoverFromException take so many arguments? Look through git history to answer"
  • "Why did we fix issue #18363 by adding the if/else in @src/login.ts API?"
  • "In what version did we release the new @api/ext/PreHooks.php API?"
  • "Examine PR #9383, then carefully verify which app versions were impacted"
  • "What did I ship last week?"
• steer tools my way
  • "Propose a few fixes for issue #8732, then implement the one that I pick"
  • "Identify edge cases that are not covered in @app/tests/signupTest.ts, then update the tests to cover them. Think hard!"
  • "commit, push, PR" (incantation)
    • it'll look through code and history to obtain context and normalize
  • "Use three parallel agents to brainstorm ideas for how to clean up `@services/aggregator/feed_service.cpp"
• plug in team's tools
  • "Use xyz CLI to check the error logs in the last run. Use -h for help on how to use this utility"
  • MCP (model context protocol) tools
    • standard for connecting AI assistants to the systems where data lives -- content repos, biz tools, dev environments
    • MCP servers
    • "Use the xyz MCP server to check for error logs in the latest run"
• common workflows
  • explore > plan > confirm > code > commit
    • "Figure out the root cause for issue #983, then propose a few fixes. Let me choose an approach before you code. Ultrathink"
  • write test > commit > code > iterate > commit
    • "Write tests for @utils/markdown.ts to ensure links render properly (note the tests won't pass yet, since links aren't yet implemented). Then commit. Then update the code to make the tests pass."
  • code > screenshot result > iterate
    • "Implement [mock png] and then screenshot it with xyz tool and iterate until it looks exactly like the mock"
• provide Claude with context
  • CLAUDE.md
    • /<enterprise root>/CLAUDE.md: shared across all projects within enterprise
    • ~/.claude/CLAUDE.md: universal
    • project-root/
      • CLAUDE.md is (version-controlled)
      • CLAUDE.local.md (not version-controlled)
    • NOTE: utilize smaller CLAUDE.md for sub-directories where that context only applies, e.g. /project-root/domains/payments/CLAUDE.md
  • slash commands
    • customize in ~/.claude/commands
      • my commands
        • ~/.claude/commands/foo.md command can be run by typing: /user:foo
      • project-specific commands
        • project-root/
          • .claude/commands/bar.md can be run with /project:bar
    • memory files: /memory
  • @-mentions
  • MCP resources
• key bindings
  • shift + tab auto-accepts edits
  • # to create a memory
  • ! to enter bash mode
  • @ to add file/folder to context
  • escape key can pause/cancel anything and it won't mess anything up
    • --resume or --continue to pick up where it left off
  • double escape key to jump into history
  • ctrl + r for verbose output
  • /vibe (unsure)