发现 Skills
为真实工作流程挑选经过整理的 Skill,连接、安装并开始使用。
deepspeed
Expert guidance for distributed training with DeepSpeed - ZeRO optimization stages, pipeline parallelism, FP16/BF16/FP8, 1-bit Adam, sparse attention
distributed-llm-pretraining-torchtitan
Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing.
railway-docs
Fetch up-to-date Railway documentation to answer questions accurately. Use when user asks about Railway features, how Railway works, or shares a docs.railway.com URL.
beadwork-for-decisions
Interactive deep-dive on using beadwork (bw — github.com/jallum/beadwork) as a decision log / ADR / audit-trail substrate. Covers how decisions map onto bw's ticket shape (title = decision, description = options, comments = discussion, close = verdict + rationale), and concrete persona scenarios: engineer ADR for architecture choices, lawyer privilege log preserved court-defensibly, solo entrepreneur tax-time defense. Triggers: bw decision log, beadwork ADR, audit trail, decision log, why did we decide, rationale, defensible decisions.
uw-code-review
Pre-commit code quality review for Unity C# code. Checks against Rules, CODING_STANDARDS.md, NAMING_CONVENTIONS.md, and GDD Gherkin test coverage. Use when code has been written and tests are passing but before committing. Runs automatically at Step 8.5 of /uw-cmd-implement-feature. Triggers on "review this code", "check my implementation", "pre-commit review", "code review", "check before commit", "is this ready to commit", "review my changes", or any request to validate code quality before committing.
docker-architect
Generate hardened Dockerfiles, docker-compose stacks, and Swarm deploy configs. Containerize any app with multi-stage builds, non-root users, healthchecks, secrets management, .dockerignore, resource limits, and security hardening. Supports Python, Node.js, Rust, Go, Nginx, and custom stacks. Includes CI/CD pipeline scanning with hadolint, trivy, and docker scout. Produces deterministic, production-ready container architecture with digest-pinned images, OCI labels, SBOM, and provenance controls.
skill-reinforcement
Always and Automatically improve skills after each use by capturing learnings and anti-patterns
OpenTelemetry Instrumentation Extension
Extend OpenTelemetry instrumentation when new functionality is added to the MCP Gateway. Use when (1) new operations/functions are added, (2) reviewing code for missing instrumentation, (3) user requests otel/telemetry additions, or (4) working with state-changing operations. Analyzes git diff, suggests instrumentation points following project standards in docs/telemetry/README.md, implements with approval, writes tests, updates documentation, and verifies with debug logging and docker logs.
package-triage
Triage a suspicious or newly encountered software package end-to-end using Spectra Assure Community reports, OSV vulnerability data, OpenSSF Scorecard, and static inspection. Use when the user asks to "triage a package", "investigate this library", "is this npm/pypi/gem safe", or supplies a purl or package name.
Obsidian Vault Management
A skill for interacting with an Obsidian vault using the local obsidian MCP server. It provides capabilities to read, write, search, and manage markdown notes.
react-components
Use this skill when developing React components. Covers functional component patterns, hooks, naming conventions, state management with Zustand and TanStack Query, and TypeScript best practices.
javascript-testing-expert
Expert-level JavaScript testing skill focused on writing high-quality tests that find bugs, serve as documentation, and prevent regressions. Advocates for property-based testing with fast-check and protects against indeterministic code in tests. Does not cover black-box e2e testing.
markdown-to-pdf
[Document Processing] Convert markdown files to PDF with syntax highlighting and custom CSS support. Cross-platform (Windows, macOS, Linux).
EvoWeb.ai AI Website Builder
Create a Website in 4 Minutes Designed to Bring Clients from ChatGPT, Gemini & Modern Search
emotional-disconnect
Use when animation feels wrong, creates unintended emotional response, or mismatches context
creating-an-agent
Use when creating specialized subagents for Claude Code plugins or the Task tool - covers description writing for auto-delegation, tool selection, prompt structure, and testing agents
dqmc-parameter-scans
Set up systematic DQMC parameter studies across temperature, interaction strength U, or chemical potential mu. Use when doing temperature sweeps, phase diagram calculations, or any grid of simulations.
git-workflow
Use when running claudikins-kernel:execute, decomposing plans into tasks, setting up two-stage review, deciding batch sizes, or handling stuck agents — enforces isolation, verification, and human checkpoints; prevents runaway parallelization and context death
design-system-creator
Builds comprehensive design systems and design bibles with production-ready CSS. Expert in design tokens, component libraries, CSS architecture. Use for design system creation, token architecture, component documentation, style guide generation. Activate on "design system", "design tokens", "CSS architecture", "component library", "style guide", "design bible". NOT for typography deep-dives (use typography-expert), color theory mathematics (use color-theory-palette-harmony-expert), brand identity strategy (use web-design-expert), or actual UI implementation (use web-design-expert or native-app-designer).
langgraph-architecture
Guides architectural decisions for LangGraph applications. Use when deciding between LangGraph vs alternatives, choosing state management strategies, designing multi-agent systems, or selecting persistence and streaming approaches.
netdata-monitor
Install and configure Netdata for real-time system monitoring with web dashboards, alerting, and health checks.
buck2-rule-basics
Guide users through writing their first Buck2 rule to learn fundamental concepts including rules, actions, targets, configurations, analysis, and select(). Use this skill when users want to learn Buck2 basics hands-on or need help understanding rule writing.
prompt-refiner-claude
Refine prompts for Claude models (Opus, Sonnet, Haiku) using Anthropic's best practices. Use when preparing complex tasks for Claude.
git-commit-pro
Generate professional git commit messages following Conventional Commits specification. Use when user asks to commit changes or write a commit message.