yzin-17/yzin-skills

Use when generating reviewed TypeScript DTOs, VOs, mappers, and request functions from OpenAPI. Use for API-code artifact review, single-file or split output plans, and safe code generation; do not use for Mockoon/Whistle exports or reverse synchronization from TypeScript.

Use only when the active parent/orchestrator model is not Luna, for coding tasks involving substantial code reading, cross-file implementation, testing, debugging, log analysis, or browser validation. Do not trigger this skill when the active model is Luna, including when Luna is running as a delegated worker. Delegate only when the overall benefit justifies it; when no worker is specified, default non-UI work to Luna, while frontend visual, layout, or styling work must not be assigned to Luna. Only one worker may execute at a time. The parent always owns scope, key decisions, and final review.

Use when turning loose API docs or reviewed OpenAPI into page-local Mockoon and Whistle mock artifacts. Use for mock-artifact review, Mockoon export, Whistle forwarding rules, list mock policies, and delayed Whistle JSON/CJS export selection; API client generation belongs to api-code-gen.

Use when creating, updating, reviewing, implementing, or closing software Specs and linked task documents, including re-scoping tasks that become difficult to verify during execution, tracking confirmed follow-up implementation in a repository TODO index, and archiving completed task documents without losing historical traceability. Also use for non-trivial changes, refactors, or migrations that need stable acceptance criteria, bounded deliverables, verification evidence, consistency review, and document lifecycle handling. Do not use for general documentation, simple questions, routine low-risk edits, one-line changes, or explicitly plan-free work.

Explicit-only workflow for delegating a task to a user-selected ChatGPT model, mode, and reasoning level, defaulting only unspecified selection dimensions to the strongest option actually selectable after opening the user's ChatGPT page. Use ONLY when the user explicitly invokes $to-chatgpt. Never activate automatically because a task is difficult, large, risky, architectural, or likely to benefit from a stronger model.

Apply Yzin's preferred style to external-facing Chinese technical articles. Invoke this skill only when the user explicitly references $writing-style-yzin; never trigger it automatically, including for technical articles, blog posts, tutorials, or case studies.