Repository work
Setup, authentication, commands, sessions, memory, planning, implementation, verification, code review, and troubleshooting.
Read guideClaude Code Guide
For developers and technical teams who need reproducible Claude Code workflows, failure handling, security boundaries, and versioned sources rather than another feature summary.
Who this guide is for and what it adds
Use this site when you are adopting Claude Code in a real repository or reviewing a team workflow. Official documentation remains the authority for product behavior. This independent guide adds bounded exercises, observable validation, failure paths, rollback steps, security limits, and a verification date so readers can test claims instead of trusting summaries.
1. Establish accessInstall one supported build, identify the authentication path, and verify the active client before opening a project.2. Complete one bounded taskStart in a trusted repository, plan the change, preserve a baseline, and prove one observable result.3. Control context and toolsLearn commands, memory, permissions, settings, and sessions before adding automation.4. Add integrations deliberatelyEvaluate MCP, hooks, skills, plugins, subagents, IDE, browser, desktop, and CI with least privilege.
Content scope
Controls and reusable procedures
CLI and in-session commands, permissions, settings, hooks, skills, plugins, subagents, and context management.
Read guideSurfaces and automation
MCP, IDE, Chrome, Desktop, GitHub Actions, CI/CD, and the boundaries between local checks and account-dependent integrations.
Read guideRecommended learning path
Foundation
Read Setup, Authentication, Getting started, and Commands. Finish one task whose result can be checked independently.
Read guideDaily repository loop
Continue with Workflows, Sessions, Memory, Permissions, Settings, and Code review to build a repeatable evidence loop.
Read guideReusable automation
Add Slash commands and Skills, then evaluate Hooks, Plugins, Subagents, and MCP only when the manual workflow is stable.
Read guideAdditional surfaces
Finish with IDE, Chrome, Desktop, CI/CD, and Troubleshooting; treat account and GUI flows as environment-specific.
Read guideAdvanced features to learn
MCP and hooks
MCP connects Claude Code to external tools. Hooks run checks, commands, prompts, or handlers at lifecycle points.
Read guideSkills and subagents
Skills package repeatable workflows. Subagents let separate agents handle specialized or parallel tasks with scoped context.
Read guidePermissions and integrations
Permissions control tool access. IDE, GitHub Actions, and CI integrations bring Claude Code into existing development loops.
Read guideHow to use and verify this guide
Check the verification record
Each topic distinguishes locally checked CLI behavior, official-documentation claims, and external actions that were not tested.
Read guideRun examples in disposable scope
Commands that create configuration use temporary directories or require a clean, reviewed branch; do not paste them over existing settings.
Read guideReport stale content
Use the correction channel when a command, version, redirect, or product boundary no longer matches the linked official source.
Read guideAll topics
The index covers 22 technical topics plus About, Contact, and Privacy pages in English and Chinese. Topic pages use one shared evidence model: applicable version, verification method, official sources, reproducible workflow, failure handling, rollback, and boundaries.