Explorar skills

Use this skill when building evaluation frameworks to measure LLM quality, safety, accuracy, or alignment including test suites, human eval rubrics, automated evals, and metrics design. Not for training or fine-tuning models. Not for dataset curation or benchmark comparison across publicly available models.

Native macOS GUI automation via peekaboo 3.0+. AX-first hybrid with VLM fallback, FSM-based workflows with bug-specific transitions. Safari web automation, native app control, social media operations, system dialogs, cross-app workflows.

Manage Confluence collaboration surface via the confluence-as wrapper: comments, attachments, labels, watchers. Triggers: 'add comment to page 12345', 'upload attachment to page 12345', 'label page 12345 as urgent', 'watch page 12345'. Inline-vs-footer comment safety enforced. NOT for page CRUD (mk:confluence-page); NOT for bulk ops (mk:confluence-bulk).

Detect breaking changes in Claude Code updates by comparing current API surface against changelog. Maps impacts to specific plugin files.

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always

Background utilities Tailwind CSS v4.1. Colors (bg-{color}, palette OKLCH P3), Gradients (bg-linear-*, bg-radial-*, bg-conic-* NEW), Images (bg-cover, bg-contain, bg-repeat), Blend modes.

Set up parallel agent team for COMPLEX tasks. Creates worktree structure, generates ownership map, and initializes task queue. Use when orchestrator decomposes a task into parallel subtasks, or when asked to "set up team", "parallel setup", or "configure worktrees".

Deep learning framework (PyTorch Lightning). Organize PyTorch code into LightningModules, configure Trainers for multi-GPU/TPU, implement data pipelines, callbacks, logging (W&B, TensorBoard), distributed training (DDP, FSDP, DeepSpeed), for scalable neural network training.

Frontend React hooks for AI-powered chat interfaces, completions, and streaming UIs with Vercel AI SDK v5. Includes useChat, useCompletion, and useObject hooks for building interactive AI applications. Use when: building React chat interfaces, implementing AI completions in UI, streaming AI responses to frontend, handling chat message state, building Next.js AI apps, managing file attachments with

Use when refining a draft user prompt before sending it to a coding agent. Decomposes goal/context/constraints/acceptance/output-format, detects ambiguity and model-coupled framing, then emits a model-agnostic rewrite. Supports --analyze, --score, and --deep. NOT for prompts from scratch (mk:brainstorming), plans/reviews (mk:elicit), implementation plans (mk:plan-creator), or general codebase scou

Confluence page CRUD via the confluence-as wrapper. Triggers: 'create page in SPACE', 'show me page 12345', 'update title of page 12345', 'delete page 12345', 'page hierarchy under 12345', 'copy page 12345 to SPACE'. NOT for bulk ops on 10+ pages (mk:confluence-bulk); NOT for spec analysis (mk:confluence-spec-analyst); NOT for comments/attachments/labels (mk:confluence-collaborate).

Sizing utilities Tailwind CSS v4.1. Width (w-*, w-screen, w-full, w-auto), Height (h-*, h-screen, h-full, h-dvh NEW), Min/Max (min-w-*, max-w-*, min-h-*, max-h-*), Aspect ratio (aspect-video, aspect-square).