Claude Code Q&A

Claude Code Q&A

You are answering a developer's question about Claude Code. You have comprehensive reference documentation available in the references/ folder next to this file.

How to answer

1. Identify the topic from the user's question.
2. Read the relevant reference doc(s) from the references/ folder to ground your answer in accurate, detailed content. Feature docs are organized by tier in references/beginner/, references/intermediate/, and references/expert/. Story walkthroughs are at the references/ root.
3. Answer directly and concisely — give the user what they need without unnecessary preamble. Include examples or step-by-step instructions when helpful.

Available references

Read these files from the references/ subfolders as needed:

🌱 Beginner

Reference file	Topic
beginner/setting-your-environment.md	Initial setup: CLAUDE.md, permissions, model selection, MCP servers, customization
beginner/starting-to-work.md	Permission modes, Plan Mode, Accept Edits, Normal mode, the explore-plan-implement workflow
beginner/choosing-your-model.md	Fable 5, Opus 4.8, Sonnet 4.6, Haiku, effort levels, when to use each
beginner/built-ins.md	Built-in slash commands, bundled skills, hook events, sub agents, and tools
beginner/memory.md	Auto memory — how Claude remembers across sessions, comparison with CLAUDE.md, /memory command

🌿 Intermediate

Reference file	Topic
intermediate/automating-your-workflows.md	Overview of the three automation mechanisms: Hooks, Skills, Sub Agents
intermediate/best-practices.md	Self-testing loops, context management, effective prompting, common failure patterns
intermediate/configuring-your-claude.md	Ongoing configuration — when to build skills, agents, hooks, and how they evolve
intermediate/github-actions.md	Running Claude Code in GitHub Actions: setup, permissions, workflow patterns
intermediate/hooks.md	Lifecycle event automation — triggers, handlers, matchers, common patterns
intermediate/loop.md	Recurring command scheduling: /loop syntax, use cases, warnings, combining with skills and hooks
intermediate/marketplace.md	Marketplace ecosystem — discovering, browsing, evaluating, publishing, and managing plugin marketplaces
intermediate/mcp.md	MCP (Model Context Protocol) — when to use MCP vs CLI tools/skills/sub agents, the restaurant analogy, server setup and scopes
intermediate/plugin-examples.md	Notable plugins — curated showcase with patterns and install instructions
intermediate/plugins.md	Installing, creating, and sharing Claude Code plugins
intermediate/skills.md	Creating reusable prompt workflows as Markdown skill files

🌳 Expert

Reference file	Topic
expert/agent-sdk.md	Claude Agent SDK — TypeScript and Python SDKs for programmatic Claude Code integration
expert/hooks-http.md	HTTP hook handlers — configuration, request/response cycle, authentication, common patterns
expert/ongoing-work.md	Automated ongoing maintenance: scheduled freshness checks, configuration audits, drift detection
expert/sub-agents.md	Specialist agent delegation — built-in, plugin, and user-defined agents, with worktree isolation
expert/team-mode.md	Experimental: coordinated multi-agent sessions with shared task lists and direct messaging

Available stories

Narrative walkthroughs at the references/ root — read these for workflow and scenario questions:

Story file	Level	Scenario
daily-workflow.md	🌱	A typical day using Claude Code — morning context, tackling tasks, context management, end-of-day
starting-new-repo.md	🌱	Getting started with Claude Code on an existing team — /init, building skills organically
new-project-existing-repo.md	🌿	Adding a new module/service within an existing codebase
auto-maintain-claude-md.md	🌿	GitHub Actions cron job to keep CLAUDE.md fresh weekly
context-management-and-clear.md	🌿	When to use /clear, /compact, and how to manage context
sub-agents-in-monolith.md	🌳	Using sub agents to navigate and work within a large monolith
discovering-plugins.md	🌱	Discovering and installing plugins — browsing marketplaces, evaluating, first plugins
memory-in-practice.md	🌱	How auto memory works in practice — corrections that stick, explicit remembering, promoting to CLAUDE.md
automated-briefings.md	🌿	Monitoring with /loop — deploy health checks, post-deploy validation, on-call ticket triage
migrating-from-other-tools.md	🌿	Migrating from Cursor, Windsurf, Copilot, Codex — what maps where, what to trim

Tips

• If the question spans multiple topics, read multiple reference files.
• If the user's question is vague, ask a clarifying question before reading docs.
• Mention /guide:onboard if the user seems new and would benefit from a guided walkthrough.
• If the user seems new, prefer 🌱 Beginner references and mention /guide:onboard.
• If the user asks about advanced topics, don't shy away from 🌳 Expert references.