发现 Skills

Expert DevOps architecture including CI/CD, Infrastructure as Code, containerization, and observability

Master REST API design principles to build intuitive, scalable, and maintainable APIs. Applicable for designing new APIs, reviewing API specifications, or establishing API design standards in Java Spring projects.

Use this skill when creating, modifying, or managing Claude Code plugins and plugin marketplaces. Trigger it when working with plugin manifests (plugin.json, marketplace.json), creating plugin directory structures, adding plugin components (commands, skills, agents, hooks), bumping versions, or discussing plugin development workflows.

Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, or extract information from web pages.

Delegate complex, long-running tasks to the Manus AI agent for autonomous execution. Use when a user says 'use manus', 'delegate to manus', 'send to manus', 'have manus do', 'ask manus', 'check manus sessions', or when tasks require deep web research, market analysis, product comparisons, stock analysis, competitive research, document generation, data analysis, or multi-step workflows that benefit from autonomous agent execution with parallel processing.

Core Effect v4 patterns including error handling (catchReason, catchReasons, catchCause, catchTag, catchIf, unwrapReason, catchNoSuchElement), tap overloads, asEffect() and the Yieldable protocol, Effect.fn, forkChild, Schema.TaggedErrorClass, Data.TaggedError, tagDefaultOmit, Latch, Deferred, Ref, Queue, filterOrFail, acquireRelease, andThen. Use when working with Effect v4 code. Triggers on catchReason, catchReasons, catchCause, catchTag, catchIf, unwrapReason, catchNoSuchElement, Effect.fn, Effect.tap, asEffect, Yieldable, forkChild, forkDetach, Schema.TaggedErrorClass, Data.TaggedError, Schema.tagDefaultOmit, Latch, Deferred, Ref, Queue, filterOrFail, acquireRelease, addFinalizer, andThen, Effect.result.

Generates and updates README files for projects and packages. Scaffolds content from package.json, directory structure, and tsconfig; validates code examples; and syncs badges. Use when creating a README for a new project or package, updating one after significant changes, or fixing stale examples and badges.

Develops firmware for microcontrollers, implements RTOS applications, optimizes power consumption for STM32, ESP32, FreeRTOS, and bare-metal systems.

Used when initializing a frontend UI project. Project initialization is integrated with visual definition: it derives the visual direction from brand guidelines, selects a component library and customizes the theme, and simultaneously generates the project scaffold and design-context files. Keywords: project initialization, design system, visual direction, theme customization, project scaffold, project setup, design guidelines, theme colors.

Scrape LinkedIn profiles, job listings, and company pages. Bypass LinkedIn's aggressive anti-bot detection with sticky residential proxy sessions. Extract names, titles, experience, skills, and job data.

Creates comprehensive Technical Design Documents (TDDs) with mandatory and optional sections through interactive discovery. Use when the user asks to "write a design doc", "create a TDD", "technical spec", "architecture document", "RFC", "design proposal", or needs to document a technical decision before implementation. Do NOT use for README files, API docs, or general documentation (use docs-writer instead).

Enhances decision-making by integrating system design thinking, product impact analysis, and advanced debugging heuristics. Works with the massa-th0th memory operating system to persist and recall decision patterns.

Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable.

Allows to interact with web pages by performing actions such as clicking buttons, filling out forms, and navigating links. It works by remote controlling Google Chrome or Chromium browsers using the Chrome DevTools Protocol (CDP). When Claude needs to browse the web, it can use this skill to do so.

Build accessible web interfaces that work for people with motor, cognitive, visual, and communication disabilities. Covers WCAG 2.2 AA compliance, motor accessibility patterns from AAC (Augmentative and Alternative Communication) development, keyboard navigation, screen reader compatibility, and inclusive testing methodology.

Build browser automation and web scraping with Playwright on your local machine. Prevents 10 documented errors including CI timeout hangs, extension testing failures, and Ubuntu compatibility issues. Includes stealth mode for anti-bot bypass, authenticated sessions, infinite scroll handling, screenshot/PDF generation, and v1.57 Speedboard performance analysis. Use when: automating browsers, scraping protected sites, testing with real IPs, bypassing bot detection, generating screenshots/PDFs, or troubleshooting "target closed", "page.pause() hangs CI", "permission prompts block tests", or "Ubuntu 25.10 installation" errors.

Titanium Alloy MVC official framework reference. Use when working with, reviewing, analyzing, or examining Alloy models, views, controllers, Backbone.js data binding, TSS styling, widgets, Alloy CLI, sync adapters, migrations, or MVC compilation. Explains how Backbone.js models and collections work in Alloy.

Titanium PurgeTSS utility-first styling toolkit. Use when styling, reviewing, analyzing, or examining Titanium UI with utility classes, configuring config.cjs, creating dynamic components with $.UI.create(), building animations, using grid layouts, setting up icon fonts, or working with TSS styles. Never suggest other CSS framework classes - verify in class-index.md first.

When a request to create a PR is received, execute verification, `$push`, and PR creation in a single end-to-end flow. If there is an Issue, prioritize that Issue when determining the PR body and base branch; if the Issue cannot be identified, continue without confirmation and operate as 'no Issue'. If there is an existing open PR, confirm whether to update it or create a new one.

Launch multiple sub-agents to iteratively review a change diff or a specified range from multiple perspectives: assess the validity of findings, fix findings that can be safely applied, re-review, and validate. Use this for deep code reviews, review-with-fixes, vulnerability reviews, cohesion reviews, responsibility-boundary reviews, directory-responsibility-boundary reviews, and iterative improvement until there are no remaining actionable findings — or when an ultra-review is explicitly requested. For ordinary review requests (where there is no explicit read-only or edit-prohibited instruction), it will safely fix adoptable findings.

The persona skill. Use when the user asks who Sensei is, what Sensei does, or any meta question. Also use when the user asks about a topic and Sensei has no skill for it — Sensei should suggest installing a relevant pack instead of answering from training data.

Use when the user wants to learn about METRICS_INFO or TS_INFO in ES|QL (Elastic 9.4 feature), wants to explore metric metadata in a time-series data stream, or when PromQL is unavailable on the cluster and you need a fallback 9.4 drill.

Ultra-compressed communication mode designed to reduce token usage by employing a terse style while preserving full technical accuracy. Intensity levels: lite, medium (default), max. Use when the user says “brief mode”, “terse style”, “terse”, “speak briefly”, “fewer tokens”, “be concise” or explicitly invokes /briefmode. Also triggers on requests to save tokens.

Shared intelligence memory plane for AI agents. Use when an agent (Claude Code, Claude Desktop, Codex, ChatGPT) needs workspace-scoped, evidence-grounded memory — search entities/claims/evidence, inspect one entity in detail, ground a claim, profile an entity, or discover what changed recently. Triggers / keywords: nerve, mnerve, workspace memory, shared agent memory, procurement memory, compliance memory, tender review, knowledge layer, shared memory, workspace memory. Do NOT use for: Claude Code's built-in auto-memory (different scope), /recall (past sessions), or raw file reading.