Clarify
ask-discover
Convert a fuzzy request into a brief with constraints, unknowns, and acceptance criteria.
Playbooks
Portable prompts for teams that need Codex discipline everywhere.
Playbooks are copy-ready prompt structures. They complement local skills and keep the operating model portable across ChatGPT, Code Review, and other places where you still need the same operating standard.
Execute
code-implement
Push Codex into implementation with explicit verification and limited edit scope.
Review
review-risk
Drive a review that prioritizes correctness and regression risk over taste.
Execute
debug-production
Work from observed versus expected behavior toward a safe fix.
Delegation
delegate-cloud-task
Split a large task into safe delegated units with clear outputs.
Review
prepare-release
Check build, tests, docs, rollback path, and release notes before shipping.
Decision
write-decision-memo
Turn a messy business choice into a concise decision memo with recommendation, tradeoffs, and asks.
Execute
meeting-to-actions
Convert notes, transcripts, or bullets into decisions, owners, dates, and a tight recap.
Execute
status-synthesis
Collapse scattered updates into a clear status narrative, risks, and next-step summary.
What a good playbook does
- makes the expected output explicit
- defines the lens, such as review versus execution
- reduces back-and-forth by naming constraints up front
- stays short enough that contributors will actually use it
Automations
Turn repeated playbook patterns into low-noise recurring tasks.
MCP
Add integrations only after the playbook and ownership are clear.