What problem does it solve? Writing a product requirements document and then getting it reviewed normally requires two separate commands with manual handoff in between. This skill removes that seam by authoring the PRD and immediately running the dual-reviewer (main agent plus reviewer agent, default Claude plus Codex) inline-fix review loop in a single invocation. ## Core Features & Use Cases - Two-phase orchestration: Phase A authors the PRD into openspec/changes/<id>/prd.md via the ptp-prd skill; Phase B runs the full ptp-review-prd-full dual-reviewer loop without re-confirming scope. - PRD gate: If Phase A fails to write the PRD, the skill stops before Phase B and reports the failure with a recovery recommendation. - Pre-resolved inputs: Consumes the already-resolved epic selector, codex.mode decision, branch guard, and model target from the outer command session without re-resolving them. - Multi-epic support: Runs author, gate, and review sequentially per epic, with one review tally table per epic. - Use Case: A product engineer types /ptp:prd-full <epic> and receives a written PRD plus a converged Claude-and-Codex review, with a terminal report recommending /ptp:plan <change-id> as the next step. ## Quick Start Run the /ptp:prd-full command with an epic selector to author the PRD and immediately review it with both reviewers.