Coding standards, frameworks, refactoring, and code review (16.8K skills)
Use this skill to turn a Git diff or supplied change context into a clear Conventional Commits message. It selects a change type and optional scope, explains the reason and impact, and can include issue or breaking change references.
Use this skill to understand LoongCollector’s architecture, terminology, key code locations, and C++ and Go conventions. It gives agents project context for investigating code and considering changes across the collector and its plugins.
Review LoongCollector code changes for concrete defects, security risks, and architectural mismatches. It tracks review history to focus on new changes and produces evidence-backed findings, lifecycle checks, and practical fix and validation plans.
Prepare a local pull request draft for Windows node changes using the current template, scoped validation results, and review evidence. It identifies readiness and blockers while keeping publication separate from preparation.
Use the Windows app’s local MCP server and `winnode` CLI to inspect available tools, invoke commands, and investigate development issues. The skill helps distinguish local app checks from tests that require the OpenClaw gateway.
Create A2UI v0.8 interfaces for the OpenClaw Windows tray node using its native WinUI renderer. Build valid JSONL surfaces from the supported component catalog, bind values to data, and follow action-context rules so only permitted data is sent back.
Use the shared autoreview skill to review a selected code scope with explicit context and P0–P3 priorities. Check the fresh report, process outcome, and coverage together, and describe any fallback review with its actual limitations.
Triage the full OpenClaw Windows Node issue and pull request backlog. Classify landing readiness, required proof, ownership, review status, and dependencies, then produce an evidence-based report and an ordered work and release plan.
Use the `winnode` CLI to discover and invoke OpenClaw commands on a local Windows app through its MCP endpoint. The reference explains command arguments, outputs, and A2UI JSONL input so agents can prepare calls and troubleshoot errors.
Run independent code reviews with two AI models, then compare their findings and verify potential issues. The skill organizes results by consensus and severity, adds fix confidence ratings, and reports clearly when either review is incomplete.
Reviews a Skill’s `SKILL.md` description for accuracy, clarity, and useful task triggers, then can recommend a revision. It also guides copying a complete Agent Skill into a user-level or project-level `.agents/skills` location.
Helps plan, write, update, and validate Agent Skills for AI agents. It guides authors through scope, folder structure, YAML frontmatter, clear activation descriptions, examples, supporting files, and troubleshooting when a skill is not discovered or used as expected.
Review code changes across many languages and frameworks for correctness, security, performance, architecture, and maintainability. The skill provides checklists and language-specific guidance, and helps produce constructive feedback that explains priorities and supports team learning.
Create and maintain agent skills by shaping their instructions, organizing supporting files, and checking required metadata. The workflow helps produce focused `SKILL.md` files with clear triggers, practical procedures, and validated directory structure.
Use the `gh` command line tool to review GitHub issues, pull requests, and CI workflow runs. Retrieve additional repository data through the API and filter structured JSON output for focused queries.
Design or update browser-visible interfaces, from forms and dashboards to landing pages and prototypes. It considers layout, visual hierarchy, interaction states, responsive behavior, and accessibility, then checks the rendered result in a browser when browser tools are available.
Creates a native SwiftUI iOS app, runs it in the Simulator, and guides command-line archiving and TestFlight upload. Includes scripts for generating icons, launching a simulator build, and uploading a signed app through the App Store Connect API.
Build MCP servers in TypeScript with the Leanmcp SDK and decorators. It guides tool, resource, and prompt definitions, automatic input schema generation, authentication, structured user input collection, and injection of user-specific environment secrets.
Build and troubleshoot 3D city games in Godot 4, from procedural streets and buildings to playable scenes. It also guides asset imports, project checks, browser exports, and cloud GPU streaming so you can prepare a game for others to play.
Create TypeScript MCP servers with the official SDK to expose APIs and application features to AI agents. The guidance covers tools for actions, read-only resources, reusable prompts, Streamable HTTP transport, and session handling for connected clients.
Build interactive React interfaces for MCP tools with the Leanmcp UI SDK. The skill covers connecting tool methods to UI components, reading tools and resources with hooks, creating forms and data grids, and supporting ChatGPT Apps.
Guides work on the Claude Code codebase by explaining its project structure, coding conventions, and patterns. Use it when adding or changing tools, commands, utilities, components, types, hooks, or services, fixing bugs, or learning how the code is organized.
Adjusts a Next.js app so PostHog session recordings and dead-click capture can load when browser blockers filter three specific script filenames. It routes those files through same-origin aliases while leaving other PostHog traffic on its existing host.
Refresh website and README screenshots from the actual QualityLayer app after interface changes. It captures curated example scenes in light and dark themes, checks the output and manifest, and updates local image assets for the developer gallery.