From inside the project
Stories about building games inside AI, with practical notes on working with it.
Can a game help you learn AI? Yes—if you do more than watch a character talk. The useful part is the loop: you give an AI a goal, see what it makes, decide what works, and steer the next move.
Read articleCodex is an AI coding agent: it can help write and review code, run tests and commands, and work with a project or repository. That makes it more useful than a blank chat for software work—and it also means you should be clear about the task and check what changed.
Read articleMaking a game with Codex is not one magic prompt. It is a sequence of design decisions: choose what the player can do, define what the game should remember, and test whether the next screen follows from that action.
Read articlePlan a short AI learning game session with a clear objective, observable learner actions, and a debrief that helps you see what students understood.
Read articleUnderstand Blender Lab’s MCP project, what to inspect before connecting it to Codex, and how to test one small scene change safely.
Read articleA practical Codex vs Claude Code comparison built around your environment, review habits, and task—not a universal winner or stale feature checklist.
Read articleLearn when a Codex skill is useful, what belongs in its instructions, and how to make a small workflow easy to find and verify.
Read articleGive Codex repository guidance that stays useful: define the project, point to sources of truth, and make checks and boundaries explicit.
Read articleA plain-language introduction to MCP: what the client and server do, how a tool call moves, and what to check before connecting one.
Read articleKeep a branching quest understandable by naming its states, transitions, and guard conditions before asking Codex to implement the story logic.
Read articlePractice vibe coding with a small, reversible game change: state the behavior, inspect the diff, check the original path, and fix regressions.
Read articlePrioritize a game prototype by uncertainty, learning value, and effort; use a timebox and stop rule to keep the next experiment focused.
Read articleUse a short acceptance pass to catch broken paths, missing instructions, and setup surprises before another player tries your AI-built game.
Read article