Shubh Shah
Community@Shubh2310-developer
Shubh Shah curates 50 skills spanning LLM observability, RAG pipelines, agent orchestration, model serving, and full-stack engineering guidance.
Agent Skills by Shubh Shah
Showing 50 vetted skills indexed across 1 GitHub repositories.
Shubh2310-developerlangfuse
Instrument LLM applications with Langfuse tracing, prompt management, and evaluation.
Shubh2310-developerui-ux-pro-max
Generates UI/UX design systems and searches style, color, typography, and stack guidelines via BM25.
Shubh2310-developerlanggraph
Build stateful multi-actor AI agents using LangGraph graphs, reducers, and checkpointers.
Shubh2310-developerclean-code
Enforces pragmatic coding standards for concise, direct, and maintainable code generation.
Shubh2310-developerbackend-dev-guidelines
Enforces layered architecture patterns for Node.js Express TypeScript microservices.
Shubh2310-developergoogle-analytics
Analyze Google Analytics 4 metrics and generate website performance recommendations.
Shubh2310-developerlangsmith-observability
Trace, evaluate, and monitor LLM applications using the LangSmith platform.
Shubh2310-developerrag-implementation
Implements retrieval-augmented generation pipelines with chunking, embeddings, vector stores, and reranking.
Shubh2310-developerlangchain
Build LLM applications with agents, chains, RAG pipelines, and vector store retrieval.
Shubh2310-developercontext7-auto-research
Fetches up-to-date library and framework documentation via the Context7 API.
Shubh2310-developergguf-quantization
Convert and quantize HuggingFace models to GGUF format for llama.cpp inference.
Shubh2310-developerbackend-patterns
Implements backend architecture patterns for Node.js, Express, and Next.js API routes.
Shubh2310-developerprompt-engineer
Designs and evaluates system prompts, few-shot examples, and chain-of-thought structures for LLM applications.
Shubh2310-developere2e-page-validator
Automates end-to-end testing of web pages by exercising inputs and verifying backend service responses.
Shubh2310-developerserving-llms-vllm
Serve LLMs with high throughput using vLLM's PagedAttention and continuous batching.
Shubh2310-developersenior-prompt-engineer
Designs and optimizes LLM prompts, RAG pipelines, and agentic system architectures.
Shubh2310-developeragent-manager-skill
Manage multiple local CLI agents in tmux sessions with start, stop, monitor, and assign commands.
Shubh2310-developerrag-engineer
Designs retrieval-augmented generation pipelines with chunking, embeddings, and hybrid search strategies.
Shubh2310-developersenior-architect
Generates architecture diagrams, analyzes project structure, and audits dependencies for system design decisions.
Shubh2310-developerdocumentation-templates
Provides templates and structure guidelines for READMEs, API docs, changelogs, and ADRs.
Shubh2310-developersegment-anything-model
Segment objects in images using point, box, or mask prompts with zero-shot transfer.
Shubh2310-developeragent-memory-mcp
Provides persistent searchable memory storage for AI agents via an MCP server.
Shubh2310-developerresearch-engineer
Implements rigorously verified algorithms and systems code with formal complexity analysis and critique.
Shubh2310-developersenior-fullstack
Scaffold fullstack projects and analyze code quality for React, Next.js, Node.js, and PostgreSQL applications.
Frequently Asked Questions About Shubh Shah
FAQPage SchemaWhat tasks can I accomplish with Shubh Shah's skill collection?▼
You can trace and evaluate LLM apps with Langfuse and LangSmith, build RAG pipelines with FAISS and Sentence Transformers, orchestrate stateful agents via LangGraph, serve models with vLLM or GGUF quantization, and generate full-stack code with React, Next.js, Node.js, and Tailwind guidance.
Who is the target audience for these skills?▼
ML engineers, backend and full-stack developers, prompt engineers, UX designers, and QA engineers. Persona skills cover senior architect, research engineer, and UX researcher roles, while framework skills target practitioners deploying LangChain, LlamaIndex, and multimodal models like LLaVA and Segment Anything.
How do I install and run these skills?▼
Skills are folder-based with native frontmatter declaring dependencies. Python-based skills specify packages such as vllm, torch, sentence-transformers, faiss-cpu, or llama-cpp-python; install via pip, then invoke the skill by name or trigger phrase within a compatible Claude Code environment.
Are these skills open source and what do they cost?▼
Most skills carry MIT licenses from Orchestra Research or Apache 2.0 licenses from vibeship-spawner-skills, making them free to use and modify. Runtime costs depend only on underlying model providers or GPU infrastructure you choose, such as Modal serverless GPUs.
What prerequisites and dependencies are required?▼
Dependencies vary per skill: vLLM serving needs torch and transformers; GGUF quantization needs llama-cpp-python; LangSmith needs langsmith>=0.2.0; multimodal skills need torch and Pillow. Loki Mode additionally requires the --dangerously-skip-permissions flag for autonomous multi-agent operation.