I build reusable skills for AI coding agents, with a focus on clear goals, practical collaboration, and reviewable results.
Turn a broad request into a Goal with an observable outcome, a bounded work loop, acceptance evidence, and clear stop conditions. Useful for coding, game development, and research that spans several iterations.
Start with a loop preview, keep a durable checkpoint, and make the next step explicit. Try a short read-only task before adapting it to a longer coding or game-development workflow.
Start in two minutes · Three prompts to try · English demo
If the workflow helps you, star Goal Instruction Writer to keep it handy and follow new examples.
Coordinate native Codex and Claude Code agents around a shared outcome. Define roles, task dependencies, write boundaries, handoffs, review, and integration.
Choose Solo, Delegation, or Team to suit the work, with optional repository checkpoints for longer tasks.
Goal Instruction Writer defines the outcome and the evidence needed to finish. Agent TaskGraph organizes who does the work and how their contributions come together.
The focus is on practical workflows that leave a clear result and a usable handoff.


