Helps turn low-leverage Codex interactions into stronger collaborations. Use when the user wants to work better with Codex, feels stuck in a narrow prompt pattern, wants help choosing between explain/plan/implement/review/automate, or wants to codify repeated workflows into skills, agents, commands, or plugins.
Use this skill when the user is underusing Codex, asking how to collaborate better, or repeating a manual pattern that should become a reusable workflow in their global Codex setup.
This skill is for moving the interaction up one level of leverage without inflating scope.
Typical triggers:
Default behavior:
Identify:
Move the user up one step at a time. Prefer the lowest level that creates durable leverage.
Do not stop at advice if a safe next action is available. Move into reading, editing, reviewing, testing, scripting, or scaffolding.
If the work exposed a stable pattern, propose one small durable follow-up:
Do not codify one-off work.
When the workflow should become reusable, choose the lightest asset that fits.
Default to a skill first unless delegation or tool integration is the real bottleneck.
When the user asks how to use Codex better, respond in this order:
Keep the recommendation tied to the current environment and actual task. Avoid generic "you could also" advice.
Example:
This is an inspection task, not a Q&A task. Better use of Codex: inspect the relevant config and tooling first, then decide what to change. Next step: read the files that define the current flow and summarize the constraints. Stretch option: if this keeps recurring, package the workflow as a skill in your global Codex setup.
For prompt patterns and concrete examples, read references/examples.md.