Claude Code tutorial: setup, agents, skills and plugins

Krish Naik's course explains using Claude Code within an existing project. It opens with a separate Claude Cowork example that analyzes invoice files and produces a report, then turns to the coding workflow. Installation options cover macOS, Linux, WSL and Windows; terminal login and repository access are demonstrated in Cursor. Model and subscription options reflect the recording.

The course describes gathering context, taking action and checking results, with users reviewing permissions and intervening. Examples include explaining a repository, creating CLAUDE.md instructions, inspecting context usage, resuming a session and planning a Python notebook for vectorless RAG with PageIndex. The notebook is shown as created later; this is not a separate PageIndex evaluation.

A custom code improvement advisor receives read-only tools and project-scoped memory, then returns prioritized findings without applying fixes. Agent Views track parallel README and test-review tasks. The course contrasts subagents with collaborating agent teams, demonstrates a small notebook review and a writer/reviewer workflow, and presents teams as experimental with an environment setting and higher token use.

The final section builds reusable research and documentation skills. Exa authentication fails during the research example, so it falls back to web search. After installing and reloading Context7, a later example retrieves LangChain documentation and uses it through a documentation skill. The walkthrough uses a local coding client connected to hosted models and external services; it does not demonstrate offline inference.