Claude Code for Developers – Advanced

1. Kickoff Briefing

• Introduction to the realistic feature request or incident.
• Project constraints: no production write access, branch-and-PR workflow only, and limited credentials.
• Available resources: repository CLAUDE.md, pre-authorized MCP servers, existing subagents and Skills, configured hooks, and automated PR review.
• Definition of the required outcome: a merge-ready pull request rather than only a working local change.
• Project success criteria: appropriate tests, reviewable changes, documented assumptions where appropriate, and no unrelated modifications.
• Guidance on where to look when blocked.

2. Project Scoping and Diagnostic Workflow

• Narrowing scope deliberately within a large repository.
• Connecting to pre-authorized MCP servers.
• Reviewing the permissions granted to each MCP tool.
• Orchestrating a multi-tool diagnostic workflow across issue context, database schema, logs, and source files.
• Hands-on exercise: diagnose a realistic cross-cutting software issue.

3. Implementation with Claude Code

• Applying the repository's existing CLAUDE.md guidance, Skills, subagents, and hooks.
• Delegating noisy or parallel subtasks to subagents.
• Using or extending existing project Skills and hooks.
• Deciding when Claude Code should execute commands directly and when commands should be run manually.
• Hands-on exercise: implement the selected fix or feature on a branch.

4. Database Changes and Testing

• Generating a database migration and tests on a branch.
• Reviewing all generated SQL before execution.
• Applying unit tests, integration tests, and mocking as appropriate.
• Validating the implementation against the project's success criteria.
• Hands-on exercise: produce and validate migrations, SQL, and automated tests.

5. Code Freeze Checkpoint

• Reviewing each team's diagnosis.
• Evaluating each team's implementation approach.
• Identifying blockers before structured engineering review begins.
• Confirming readiness against the Definition of Done and engineering quality rubric.

6. Engineering Playoffs – Round 1

• Pairing eight teams to review repositories, pull requests, architecture, and testing strategies.
• Evaluating solutions using the shared Definition of Done and engineering quality rubric.
• Selecting the strongest base solution.
• Identifying the best engineering ideas and weaknesses from each implementation.
• Hands-on exercise: integrate the strongest compatible ideas into four improved solutions.

7. Engineering Playoffs – Round 2

• Pairing the four advancing solutions for another structured engineering review.
• Selecting the strongest implementation from each pair.
• Incorporating the best compatible ideas from both solutions.
• Validating the integrated implementation.
• Hands-on exercise: prepare two solutions for final review.

8. Engineering Playoffs – Final Round

• Presenting the two finalist solutions to the class.
• Explaining the problem diagnosis and engineering decisions.
• Presenting the testing strategy and Claude Code workflow.
• Identifying ideas adopted from earlier review rounds.
• Defending tradeoffs and rejected approaches.
• Evaluating both solutions with the shared engineering rubric.
• Selecting the preferred class implementation.
• Instructor validation of the final outcome.

9. Pull-Request Review and Retrospective

• Opening a pull request and triaging automated review findings before merging.
• Participant or pair walkthroughs covering the diagnosis, fix, and tests.
• Facilitator-led review of generated SQL, migrations, and commands requiring manual approval.
• Group discussion of where Claude Code accelerated the work.
• Discussion of where Claude Code required correction.
• Reflection on where engineering judgment remained essential.
• Identification of engineering practices that consistently produced the strongest results.