'Visualizes Azure infrastructure from ARM templates, Azure CLI, or descriptions. Use when user has Azure resources to diagram.'

Implement technical plans from thoughts/shared/plans with verification

Generate custom reports (query reports and script reports) for Frappe applications with filters, aggregations, and formatting.

Create and maintain llms.txt files for AI crawlers following the official specification. Use when setting up AI-friendly documentation indexes, creating machine-readable content maps, or helping websites become more discoverable by LLMs like ChatGPT, Perplexity, and Gemini.

Verifies that a PaperMiner change is ready to hand over by checking tests, lints, ripple effects, and documentation sync. Use before declaring a feature, refactor, or bug fix complete.

Provides React Native performance optimization guidelines for FPS, TTI, bundle size, memory leaks, re-renders, and animations. Applies to tasks involving Hermes optimization, JS thread blocking, bridge overhead, FlashList, native modules, or debugging jank and frame drops.

>-

Converts static HTML mockups (especially multi-section pages under docs/inputhtmls) into Next.js App Router pages or React section components for this portfolio. Maps Tailwind-style tokens from the mock (primary, background-light/dark, surface, glass-card, animations) to project CSS variables and existing utilities in src/app/globals.css. Use when the user references docs/inputhtmls/*.html, asks to port a section or full page from HTML, or to sync a Storybook-style HTML export with the live site.

Translates discovery and product design artifacts into implementation-ready backend specifications. Produces API contracts (OpenAPI), schema design, auth model, and integration point specs for human r

High-level PyTorch framework with Trainer class, automatic distributed training (DDP/FSDP/DeepSpeed), callbacks system, and minimal boilerplate. Scales from laptop to supercomputer with same code. Use when you want clean training loops with built-in best practices.

Provides guidance for enterprise-grade RL training using miles, a production-ready fork of slime. Use when training large MoE models with FP8/INT4, needing train-inference alignment, or requiring speculative RL for maximum throughput.

Enforces API documentation using language-standard comments for public interfaces and generates Conventional Commits style messages when asked, clearly summarizing changes. Use when creating or modifying public classes/functions or when the user requests commit messages.

Perform cross-artifact consistency and quality analysis. Automatically

Drives iterative development of the Pet Central trusted marketplace. Use when building features, improving UI/UX, implementing backend modules, testing, or completing any work described in docs/spec.md. Applies to all apps (web-consumer, web-vendor, web-admin, web-partner, web-kiosk), packages (ui, database, auth, etc.), and the api-gateway. Includes complete testing and validation with correction loops until the system is fully functional.

Creates React components following RYLA patterns and file organization. Use when creating components, building UI, implementing features, or when the user mentions components, UI, or React.

Design and implement Concept Encoder PyTorch modules safely. Use when sketching tensor shapes, writing or refactoring nn.Module code, checking memory or DDP behavior, choosing AMP or torch.compile patterns, or improving training-time performance and numerical stability. Not for experiment logging, changelog updates, or high-level research prioritization.

>-

需求分析与结构化专家。从多种来源（飞书 PRD 文档、手动输入、文件）分析业务需求，进行结构化分析（业务场景、功能点、交互流程、数据模型、技术实现），并生成可执行的 todolist。适用于：需要将业务需求转化为结构化分析文档的场景、需要生成开发 todolist 的场景、需要从 PRD 文档提取和分析需求的场景。

Goal: build a skill that finds real vulnerabilities while minimizing false positives.

Verify an E2E test fix by running the test multiple times and checking code quality

Guides writing clear, well-structured technical documentation following Mintlify's best practices (Diátaxis framework, audience awareness, style/tone, navigation, SEO/AEO). Use when writing, reviewing, or improving documentation, guides, tutorials, API references, or any user-facing technical content.

Profile app performance while browsing, collecting Web Vitals and React rerender data via react-scan. Orchestrates parallel profiler subagents via playwright-cli to capture navigation timing, long tasks, layout shifts, LCP, React commit counts, render bursts, and per-component render data. Use when profiling browsing performance, finding bottlenecks, diagnosing excessive rerenders, or auditing page performance.

Maintain a development log (DEVLOG.md) at the project root. Use at the START and END of every coding session, after implementing features, fixing bugs, refactoring code, or making any meaningful code changes. Automatically triggered for all development tasks.

Use this skill when the task explicitly involves:

Reviews local git changes and generates a pull request summary following the project template. Use when the user asks to automate PR summary, generate PR description, fill PR template, or review local changes for a PR.

Implement new MCP tools in the deno-mcp-template project. Provides the exact file structure, type signatures, registration steps, and patterns for standard tools, sampling tools, form and URL elicitation, resource-backed tools, and notification tools. Use when adding a new tool, creating MCP tools, or asking how tools work in this project.

Add or modify tasks in Implementation Plan. Use when user proposes new requirements, changes priorities, or adjusts existing task scope.

Activates whenever referencing backend routes in frontend components. Use when importing from @/actions or @/routes, calling Laravel routes from TypeScript, or working with Wayfinder route functions.

Sync .env from 1Password using env_var_mappings.

>

'Stage files and create a conventional commit with a well-formed message.'

Personalizes a fork of this Cloudflare multi-worker monorepo—worker names in wrangler, package names, README, and UI—so deploys and docs are not stuck on starter kit defaults. Use when the user created a repo from the template, wants to deploy their own app, or asks to rename workers, rebrand, or remove generic starter kit copy. Skip when developing the upstream starter kit (origin URL contains firtoz/cf-multiworker-starter-kit; canonical https://github.com/firtoz/cf-multiworker-starter-kit) unless the user explicitly asks.

>-

'Use when staging and committing code changes with conventional commit messages'

METAINFORMANT rules for directory scripts/ontology. Use when editing, adding tests, or reviewing code under this path. Read the linked AGENTS.md first; use uv only, write outputs to output/, no mocks.

METAINFORMANT rules for directory examples/templates. Use when editing, adding tests, or reviewing code under this path. Read the linked AGENTS.md first; use uv only, write outputs to output/, no mocks.

Create a GitHub Pull Request to merge the current branch into main with a well-structured title and description using GitHub CLI. Use when the user asks to create a PR, open a pull request, submit a merge request, or send changes for review.

METAINFORMANT rules for directory examples/multiomics. Use when editing, adding tests, or reviewing code under this path. Read the linked AGENTS.md first; use uv only, write outputs to output/, no mocks.

Deploy Supabase Edge Functions. Use when deploying process-task or other functions, when Docker mount fails, or when updating edge function code.

>-

office

Security and compliance for MCPAtlas: threat model, RBAC, audit, supply chain, input validation, plugin sandbox. Use when implementing auth, hardening, or reviewing security-sensitive code.

azure-auth

Rosetta ad-hoc adaptive meta-workflow that constructs, tracks, reviews, and executes a tailored execution plan per user request. Use when no other specific workflow matches the request.

Run the validation pipeline and produce a short markdown report.

Run Frontend/UI QA: Storybook coverage and visual checks, responsive behaviour, component consistency. Use when the user runs a full QA, /qa-frontend, or asks for frontend-only QA. Output structured findings in the QA report format.

Implements Geoffrey Huntley's Ralph Wiggum autonomous iteration technique for managing LLM context. Use when working on long-running tasks, when context is getting polluted, or when you need autonomous development with deliberate context rotation. Treats LLM context like memory - rotates to fresh context before pollution builds up, with state persisting in files and git.

Run all linting, type checking, tests, Markdown lint, and documentation build for the project. Check for errors and warnings, then fix any problems found. Use when the user asks to run checks, verify the build, run CI locally, or fix lint/type/test errors.

*nd*You** (the lead) do the commits. Do not dispatch sub-agents to commit.

Guidance for `effect/Console` focused on APIs like dir, log, and info. Load after `effect-skill-router` when this module is the primary owner.