Mastering Claude AI | 2 Days Workshop at IIIT Bangalore

● Claude models, context windows and tokens. ● Installation, authentication and first session. ● Terminal, IDE and web interfaces. ● Understanding usage and cost. ❖ Practical Outcome / Project: Explore and explain an unfamiliar codebase using Claude Code. Module 2: Driving Claude Code Effectively ● Explore, Plan, Code, Commit workflow. ● Plan mode and permission modes. ● Prompting an agent versus prompting a chatbot. ● Context management with clear and compact. ● Rewind, resume and session control. ❖ Practical Outcome / Project: Ship a verified code change using plan mode. Module 3: Development Workflows ● Onboarding to an existing codebase. ● Code generation, refactoring and debugging. ● Test-driven development with Claude. ● Documentation and test case generation. ● Git commits and pull request workflows. ● Verifying agent output and avoiding false confidence. ❖ Practical Outcome / Project: Land a reviewed feature branch end to end. Module 4: Project Memory and Configuration ● CLAUDE.md and the memory hierarchy. ● Settings files and their precedence. ● Permission rules and safe defaults. ● Team configuration in version control. ● Secure and responsible agent usage. ❖ Practical Outcome / Project: Configure a real repository for your team. Module 5: Custom Commands and Skills ● Building custom slash commands. ● Arguments, file references and inline shell. ● Creating and triggering Agent Skills. ● Choosing between commands, skills and memory. ● Packaging scripts and reference files. ❖ Practical Outcome / Project: Build a custom command and a working skill. Module 6: Model Context Protocol (MCP) ● Introduction to MCP and its architecture. ● Connecting local and remote MCP servers. ● Tools, resources and prompts. ● Project-scoped server configuration. ● MCP security considerations. ❖ Practical Outcome / Project: Complete a real task through a connected MCP server. Module 7: Subagents, Hooks and Agent Design ● Creating specialised subagents. ● Parallel delegation and context isolation. ● Hooks and lifecycle events. ● Deterministic automation versus model instructions. ● Human-in-the-loop and reliable agent design. ❖ Practical Outcome / Project: Build a subagent and an enforcing hook. Module 8: Automation and Integration ● Headless mode and scripted runs. ● Structured JSON output for pipelines. ● CI/CD and automated pull request review. ● Claude Agent SDK and the Claude API. ● Choosing the right surface for the job. ❖ Practical Outcome / Project: Automate a real developer workflow in CI. Final Project ● Build one working Claude Code solution end to end. ● Demonstrate and review within the group. ❖ Practical Outcome / Project: A production-ready Claude Code setup you take back to work.