AI coding collaboration manual
Turn ideas, context, and acceptance criteria
into a shippable workflow
Install Claude Code and Codex CLI, write better prompts, feed project context, split tasks, review diffs, run verification, and move AI coding from experiments into repeatable delivery.
turn the publishing flow into reversible small tasksread routes, build scripts, and sidebar structureplan first, edit second, verify every stepsummarize the change after the build passesDelivery path
- read context
- split tasks
- edit code
- verify
Toolchain
Install the AI coding tools first
A terminal-native AI coding assistant for understanding repos and editing across files.
Open guideCodex CLIOpenAI local coding agent with sandboxing, approval flow, and controlled execution.
Open guideQuick startRun one low-risk task through prompting, implementation, and verification.
Start practiceLearning Path
Jump in by real development stage
Practice the full AI collaboration loop with one small feature.
02Prompting and contextWrite prompts that help AI understand the task and the project.
03Collaboration managementClarify requirements, acceptance, boundaries, and human roles.
04Delivery troubleshootingHandle drift, failing tests, quality regressions, and launch checks.
Workflow
The page is the doorway. The rhythm is the product.
Every AI coding task works better when goal, context, constraints, and acceptance criteria live in one loop. You steer the direction; AI advances the verifiable steps.
Name the feature, relevant files, and boundaries that must stay intact.
Ask AI to summarize the existing pattern before proposing implementation steps.
Review diffs, run checks, and keep the work aligned with the goal.