Claude Code Tips

Collection of my favorite Claude Code tips as I explore it.

Table of Contents

General

• Tip 1: Name and resume sessions like git branches
• Tip 2: Match your effort level to the task
• Tip 3: Skip local file discovery for scripted runs
• Tip 4: Turn on Explanatory or Learning mode to study a codebase
• Tip 5: Move sessions between your laptop, phone, and the cloud
• Tip 6: Turn a recurring task into a loop or a scheduled job
• Tip 7: Understand what auto mode decides for you
• Tip 8: Track your token spend with ccusage
• Tip 9: Organize multiple projects with a terminal multiplexer app
• Tip 10: View all background sessions with agent view
• Tip 11: Install a language server plugin for real code navigation
• Tip 12: Check your context usage with /context
• Tip 13: Analyze your usage patterns with /insights

Command

• Tip 1: Give Claude a standing goal for the whole session
• Tip 2: Let /batch run a migration across hundreds of files
• Tip 3: Run parallel sessions in isolated worktrees
• Tip 4: Pre-approve the commands you already trust
• Tip 5: Sandbox risky commands instead of prompting for each one
• Tip 6: Script Claude into CI and pre-commit hooks
• Tip 7: Bootstrap CLAUDE.md with /init instead of starting from a blank file
• Tip 8: Be aware of ultra features

Agent

• Tip 1: Define reusable subagents instead of re-explaining a role every time
• Tip 2: Delegate research to a subagent to keep it out of your context
• Tip 3: Have a fresh subagent grade the work before you call it done
• Tip 4: Batch large changes across worktree-isolated agents
• Tip 5: Use agent teams when work needs multiple roles, not just parallel copies

Skill

• Tip 1: Restrict a skill to manual invocation so Claude doesn't guess when to use it
• Tip 2: Use another CLI when WebFetch can't fetch a page
• Tip 3: Let a plugin enforce TDD and root-cause debugging discipline
• Tip 4: Give Claude a real browser for pages WebFetch can't handle
• Tip 5: Browse skills.sh before writing one from scratch
• Tip 6: Use skill-creator to create new skill
• Tip 7: Install a skill to curb overengineering and unrequested changes

Mcp

• Tip 1: Bundle MCP servers, skills, and hooks together as a plugin
• Tip 2: Tune how aggressively Claude loads your MCP tool definitions
• Tip 3: Keep raw tool output out of your context with a sandboxed MCP server
• Tip 4: Give Claude persistent memory across sessions with claude-mem
• Tip 5: Bridge Claude Code and Codex with a plugin
• Tip 6: Browse docs with Context7

Prompt

• Tip 1: Have Claude interview you before building something big
• Tip 2: Write the code in one session, review it in another
• Tip 3: Put your planning effort into plan mode, not into micromanaging
• Tip 4: Move plans out of the 30-day cache and into your repo
• Tip 5: Keep exploration out of the plan you're about to approve

Hooks

• Tip 1: Hook into specific moments in Claude's lifecycle
• Tip 2: Auto-format on every edit with a PostToolUse hook
• Tip 3: Use a Stop hook as a deterministic gate for unattended runs
• Tip 4: Get worktree isolation on non-git version control
• Tip 5: Turn hooks into ambient audio cues

Workflow

• Tip 1: Treat CLAUDE.md as a file you keep improving, not a one-time README
• Tip 2: Reserve "IMPORTANT" for the one rule Claude keeps missing
• Tip 3: Let /doctor prune your CLAUDE.md for you
• Tip 4: Give Claude a way to check its own work
• Tip 5: Run risky, unsupervised work in a container

Context

• Tip 1: Manage context deliberately instead of just filling it up

General

Tip 1: Name and resume sessions like git branches

Claude Code saves every conversation locally, so a task that spans multiple sittings doesn't need re-explaining from scratch. Run /rename to give a session a descriptive name, the same way you'd name a branch. Use claude --continue to pick up your most recent session, or claude --resume to choose from a list of past ones. Treat each workstream as its own persistent context: one session per feature, not one giant thread for everything you touch this week. Descriptive names pay off weeks later, when you're trying to remember which session already has the context you need.

Example:

claude --resume

Then pick "oauth-migration" from the list to jump back into that work with full context intact.

Reference: Manage sessions

Tip 2: Match your effort level to the task

Run /effort to control how hard Claude thinks before responding. Levels range from low (fewer tokens, faster) through medium, high, xhigh, max, and auto, where Claude picks per request. The default is high on Team, Enterprise, and direct API access, and medium everywhere else. Reach for xhigh on complex coding or agentic work when you want deeper reasoning without paying the full cost of max. Save max for the genuinely hard cases: a gnarly debugging session or an architecture decision where you want Claude to think for as long as it needs. Max burns through usage limits faster, so turn it on for the session that needs it rather than leaving it as your default.

Example:

/effort xhigh
Refactor the auth module to support multi-tenant sessions without breaking existing token validation.

Reference: Power user tips

Tip 3: Skip local file discovery for scripted runs

By default, claude -p and the SDKs search your filesystem for CLAUDE.md files, settings, and MCP configs before every run. That's the right behavior interactively, but for scripted or CI usage you already know exactly what should load. Add --bare and pass --system-prompt, --mcp-config, and --settings explicitly instead, and startup gets roughly 10x faster. This local search was a default set early on, and the Claude Code team plans to flip it in a future version, so --bare is the flag worth reaching for today whenever you're calling Claude from a script.

Example:

claude -p "summarize this codebase" \
  --output-format=stream-json \
  --verbose \
  --bare

Reference: Power user tips

Tip 4: Turn on Explanatory or Learning mode to study a codebase

Output styles change how Claude talks, not just what it does. Set one in /config. Explanatory mode has Claude narrate the frameworks and patterns behind its own changes as it works, which is useful when you're onboarding onto code you didn't write. Learning mode goes a step