What problem does it solve? Writing a change requirement ("ТЗ") for the HubTalk AI Conversation Flow project requires a strict, multi-artifact format: a numbered OpenSpec change directory with a proposal, spec deltas, and acceptance tasks. This Skill enforces that structure so requirements are consistent, reviewable, and traceable across the migration branch and main. ## Core Features & Use Cases - Numbered change authoring: Computes the next free ТЗ number from four sources (changelog, local patches, origin/main patches, existing changes) and creates openspec/changes/tz-NNN-<slug>/ with proposal.md, spec deltas, and tasks.md. - Spec delta rules: Enforces OpenSpec delta conventions (ADDED/MODIFIED Requirements, Scenario blocks, id/entities/enforced metadata whitelist, MODIFIED identity rules) so changes validate and archive cleanly. - Gate-driven workflow: Integrates with the pipeline /tz -> /tz-review -> plan-griller -> test-author -> /tz-implement, including tier selection (light/standard/deep) and pre-submission checks via scripts/sdd/check.sh and lint_brand.py. - Use Case: A developer is asked to write a requirement for a new voice-flow fix; the Skill scouts the codebase, picks number tz-101, drafts the proposal with Why/What Changes/Grill/Тир sections, writes the spec delta against the affected capability, and lists acceptance tasks. ## Quick Start Ask the assistant to write a ТЗ (change requirement) for your feature or defect using the tz skill, providing the ticket or defect context.