SEARCH
PART 10 · PROJECT CONTROL
PART 10
PROJECT CONTROL
4 classes. Each class holds its own standalone lessons.
PART 10 CLASSES
CLASSES, THEN LESSONS.
CLASS 41Git and GitHub for AI Builders
A Repository Is a Project With a Memory
Learn the difference between a project folder, Git, GitHub, a repository, and the history that keeps AI work recoverable.
READ LESSON →September 12, 2026 · CLASS 41Commits, Diffs, and the Moment a Change Becomes a Checkpoint
Turn one coherent change into a readable checkpoint with a reviewed diff and an honest commit message.
READ LESSON →September 12, 2026 · CLASS 41Branches and Worktrees: Parallel Work Without Shared-Folder Collisions
Use named branches and worktrees to let parallel AI work move without sharing one fragile folder.
READ LESSON →September 12, 2026 · CLASS 41Pull Requests, Merges, Reverts, and the Review Gate
Review, merge, preview, and deliberately revert AI-made changes without rewriting shared history.
READ LESSON →CLASS 42Specifications
A Spec Is the Project's Shared Memory
Turn a vague feature request into a durable agreement about behavior, boundaries, and proof.
READ LESSON →September 12, 2026 · CLASS 42The Small File Set That Explains a Real Project
Build the small, clear set of files that gives people and AI one source of truth for a project.
READ LESSON →September 12, 2026 · CLASS 42Decisions, Assumptions, and Change Control
Separate verified facts, assumptions, choices, and open questions so the project stays honest as it changes.
READ LESSON →CLASS 43Plan First, Code Second
From Product Requirement to Buildable Roadmap
Turn an approved product idea into a sequenced roadmap with milestones, dependencies, validation, and stop points.
READ LESSON →September 12, 2026 · CLASS 43Plan the Interfaces Before Parallel Workers Build Them
Write the contracts that let separate human and AI workers build compatible pieces without guessing.
READ LESSON →September 12, 2026 · CLASS 43Risk-Based Planning: Spend Attention Where Being Wrong Is Expensive
Assign the design, review, testing, and approval a task needs based on its consequences, not its novelty.
READ LESSON →CLASS 44Break Large Projects Into Small Tasks
Build Vertical Slices, Not a Pile of Layers
Break ambitious work into user-visible slices that cross the interface, logic, and data layers to reach a real finish line.
READ LESSON →September 12, 2026 · CLASS 44The Task Card: One Clear Job for One Human-AI Team
Give every task a bounded outcome, scope, checks, evidence, and escalation path before an AI starts.
READ LESSON →September 12, 2026 · CLASS 44Dependencies, Parallel Lanes, and the Integration Owner
Map dependencies, run genuinely independent work in parallel, and give one person ownership of integration.
READ LESSON →September 12, 2026 · CLASS 44The Controlled AI Delivery Loop
Run a task from context and inspection through evidence, checkpoint, review, and a handoff another person can trust.
READ LESSON →PRACTICAL APPENDIX
USEFUL TOOLS.
Appendix: An AGENTS.md Starter Every Agent Reads First
A practical starting file that tells every AI worker what the project is, what it may touch, and what proof ends a task.
READ APPENDIX →September 12, 2026 · PART 10 APPENDIXAppendix: A .gitignore Starter for AI Projects
Keep secrets, local data, dependency folders, and operating-system clutter out of your repository history.
READ APPENDIX →September 12, 2026 · PART 10 APPENDIXAppendix: The Weekly Control Review (20 Minutes)
A short weekly ritual to keep your roadmap, task board, repository history, evidence, and next tasks aligned.
READ APPENDIX →