AI & Machine Learning Bundle

Prompt engineering, LLM integration, agent design, model fine-tuning, and AI workflow skills for ML practitioners.

Browse all skill bundles

Included Skills (90)

Langchain — LangChain enables developers to construct advanced LLM‑powered applications, such as RAG pipelines and conversational agents, using Python and TypeScript tools. It provides agents, structured output, prompt templates, memory‑augmented chatbots, and vector store integration to streamline workflow creation. Data‑centric teams and AI engineers deploy and scale production‑ready language model solutions with this framework.
PydanticAI Agent Framework — This skill enables developers to build robust, type-safe AI agents in Python using PydanticAI for structured outputs, tool integration, and reliable model testing.
MCP Server Architect — Develop and deploy robust Model Context Protocol servers using TypeScript or Python to extend AI agent capabilities with custom tools and resources.
Huggingface Llm Trainer — The Huggingface Llm Trainer equips data scientists and ML engineers to train or fine‑tune language and vision models on cloud GPU infrastructure, using TRL or Unsloth for SFT, DPO, GRPO, reward modeling, and GGUF conversion for local deployment. It guides users through dataset preparation, hardware selection, cost estimation, and monitoring with Trackio, while managing Hub authentication, model selection, and persistence. This skill streamlines efficient, cloud‑based training workflows for professionals lacking local GPU resources.
Pydantic AI — Pydantic AI enables developers to build type‑safe AI agents in Python using the Pydantic agent framework, validating structured output and injecting tools via dependency injection. The skill streams responses and allows seamless switching between OpenAI, Anthropic, Gemini, or Ollama models. It supports model‑agnostic testing, benefiting data‑science teams and developers who require robust, reliable agent pipelines.
Neon AI Gateway — Neon AI Gateway provides a single API endpoint and credential that unifies access to frontier and open‑source LLMs and embedding models, including OpenAI, Anthropic, Google Gemini, Meta, Alibaba, and more. It integrates directly into a Neon branch and is powered by Databricks, allowing developers to call LLMs, generate embeddings, add AI chat or agents, and route between providers without juggling separate keys or accounts. The gateway works with the OpenAI SDK, Anthropic SDK, google‑genai, Vercel AI SDK, and Mastra by simply changing the base URL, streamlining AI requests for projects already using Neon.
RouterBase Model Gateway — This skill provides an OpenAI-compatible gateway for developers to route diverse AI model requests, including text, media, and embeddings, through RouterBase.
Langsmith — Langsmith monitors, traces, and evaluates LLM applications, providing real‑time observability of agent performance, token usage, and latency. It enables developers to debug chain executions, run prompt experiments, compare model outputs, and build testing pipelines. The platform supports evaluation datasets, annotation queues, a prompt hub, and production monitoring for continuous AI quality assurance.
Gke Inference — The Gke Inference skill deploys and optimizes AI and ML inference workloads on Google Kubernetes Engine, leveraging GPUs, TPUs, and model servers to deliver high‑performance inference. It enables teams to configure GKE GPU resources and deploy large language models, ensuring efficient scaling and resource utilization for real‑time inference scenarios. The skill is tailored for inference workloads and does not support migration, GKE RAG with Cloud SQL or AlloyDB, or batch/HPC use cases.
Llm Application Dev Langchain Agent — You are an expert LangChain agent developer specializing in production-grade AI systems using LangChain 0.1+ and LangGraph.
AI Development Workflow — This comprehensive workflow assists developers in building, integrating, and scaling LLM applications, RAG systems, and intelligent agent architectures for production environments.
Claude API — The Claude API skill enables developers to integrate Anthropic’s Claude models into applications using the Anthropic SDK or Claude Agent SDK. It activates automatically when code imports `anthropic`, `@anthropic-ai/sdk`, or `claude_agent_sdk`, or when a user explicitly requests Claude API usage. The skill is tailored for building AI-powered features and excludes unrelated AI or general programming libraries.
AI Agent Tool Routing — Use when designing or reviewing an AI agent tool registry or MCP router: document parameter schemas, caller authorization, execution limits, replay-safe retries and verification evidence.
AI Agent Evaluation Benchmarking — Autonomous AI agent benchmark evaluation register: task completion rates, planning accuracy, tool invocation precision, and cost benchmarks.
Cerebras — Cerebras Inference delivers a hosted LLM API that runs open models on Cerebras wafer‑scale chips, providing thousands of tokens per second through an OpenAI‑compatible endpoint. Developers benefit from the lowest‑latency chat, code completion, and agent loops, with support for streaming, tool calling, strict JSON schema output, and reasoning control. The service guarantees reliable production access by eliminating “model not found” errors from retired Cerebras model IDs.
AI Eval Ci — AI Eval CI embeds AI agent and LLM evaluations into CI/CD pipelines, automatically failing builds when output quality falls below defined thresholds. It supports Cobalt, Promptfoo, and Braintrust frameworks, providing LLM-as-judge scoring, threshold-based assertions, and seamless GitHub Actions integration. Developers use it to detect prompt regressions, compare models, and benchmark agents before deployment.
RAG System Architect — This workflow guides developers through building and optimizing RAG systems by streamlining embedding selection, vector database configuration, and retrieval performance evaluation.
AI Prompt Regression Testing — Use when comparing prompt changes against a fixed evaluation set: record actual model settings, required checks, missing results, judge limitations and case-level regression evidence.
System Prompt Lookup — Checks what a shipped AI product's system prompt and tool schema actually say, by reading a dated archive of captured prompts instead of recalling them. Use before asserting or accepting any claim about an agent's instructions.
ML Engineer — Build production ML systems with PyTorch 2.x, TensorFlow, and modern ML frameworks. Implements model serving, feature engineering, A/B testing, and monitoring.
Tactical Voice Assistant — This skill enables developers to build a low-latency, Iron Man-inspired conversational voice agent using the Pipecat framework, Gemini, and OpenAI services.
Llm Application Dev Prompt Optimize — Improve LLM prompts using task-specific evaluation, structured instructions, and regression checks.
RAG Security Auditor — Probes Retrieval-Augmented Generation pipelines for indirect prompt injection vulnerabilities to help security engineers validate guardrails and secure document-based AI systems.
Using Agent Skills — The skill guides users through locating, selecting, installing, and executing agent skills that adhere to the Agent Skills specification across Claude Code, Codex, Gemini CLI, and Cursor. It provides clear, step‑by‑step instructions for common tasks such as listing available skills, ensuring compatibility across agents, troubleshooting activation issues, and verifying safety before installation. This empowers developers and power users to efficiently integrate and manage skills within their AI workflows.
AI Tool Discovery — This tool enables AI agents to dynamically discover, verify, and connect to external APIs and MCP endpoints for enhanced workflow automation.
Agent Platform Prompt Management — Manages and orchestrates prompts in Agent Platform. Use when you need to create, list, retrieve, version, or delete managed prompts in Agent Platform. Don't use for model training, model deployment to endpoints, or managing non-Agent Platform prompts.
Hermes Agent — Hermes Agent is an open‑source AI agent that functions as a terminal and messaging gateway, recording memories, writing and patching its own skills, and retrieving past sessions. Developers embed it in Python, select a model provider, and build adaptive assistants that grow more capable with each use. Teams benefit from a continuously evolving, self‑optimizing tool that enhances workflow automation and intelligent interaction.
AI Agent Security — Secure AI agents against prompt injection, tool abuse, and data exfiltration
Structured Output — Structured Output directs large language models to produce validated JSON, ensuring machine‑readable data integrity. Developers and data engineers use it to extract and validate structured information from OpenAI, Anthropic, or Claude APIs, integrating seamlessly with Zod schemas, Pydantic models, and the Instructor library. Its retry strategies for malformed responses guarantee robust data extraction.
MCP Server Builder — The MCP Server Builder constructs Model Context Protocol servers that link AI agents to external APIs, databases, and services. It generates TypeScript or Python code that includes tools, resources, and prompts in compliance with the official MCP specification. This skill benefits developers who need to expose data or services to AI agents in a standardized, secure manner.
MCP Governance Controller — This skill enables developers to secure AI agent tool calls using Cedar policies, shadow-mode testing, and cryptographic Ed25519 audit receipt verification.
Haystack — Haystack is an open‑source Python framework that lets developers build and debug retrieval‑augmented generation pipelines, search systems, and LLM agents from modular components such as document stores, retrievers, prompt builders, chat generators, and agents. It supports custom component development, document indexing for retrieval, agent creation with tool integration, and migration from Haystack 2.x to 3.x. Data scientists, AI engineers, and software developers benefit from its flexible, component‑based approach to advanced search and conversational AI solutions.
Airunway Aks Setup — The Airunway Aks Setup skill provisions an AI Runway environment on an AKS cluster, verifying cluster readiness, installing the controller, assessing GPU resources, configuring the provider, and deploying the first model. It enables data scientists and ML engineers to run large language models, vLLM, and GPU‑accelerated inference on AKS with minimal manual effort. The service streamlines onboarding, deployment, and model serving for teams seeking scalable AI workloads on Kubernetes.
Agent Memory Systems — Design tenant-scoped memory, retrieval and retention for agents. Use when planning RAG, conversation history, vector stores or remembering across sessions.
Agent Platform Model Registry — The Agent Platform Model Registry centralizes machine learning model management, enabling users to upload, list, describe, update, or delete models while enforcing strict version control and detailed metadata. By streamlining inventory organization, it frees data scientists and ML engineers to focus on training and deployment, accelerating development cycles and improving deployment reliability.
Agent Platform Tuning — The Agent Platform Tuning skill fine‑tunes open and Gemini models within the Agent Platform infrastructure. It streamlines the adjustment of model parameters for developers who need precise control over performance. The skill is not intended for external training, endpoint deployment, or serving endpoint management.
Agent Platform Rag Engine Management — The Agent Platform Rag Engine Management skill enables professionals to list, inspect, and query RAG corpora and files within the Agent Platform, retrieving grounded contexts via the Google GenAI SDK. Data scientists and AI developers use it to generate content anchored in a specific RAG corpus, ensuring precise, contextually relevant information. It is not intended for general database queries or other RAG products such as Google Workspace RAG or gRAG.
LLM Prompt Engineer — This skill helps AI developers design effective prompts for LLMs, improving output quality and reliability across various applications.
Context Engineering — The Context Engineering skill audits and reorganises an AI coding agent’s context window, managing instruction files, path‑scoped rules, skills, subagents, hooks, and the current conversation. It ensures the agent follows the correct CLAUDE.md, AGENTS.md, and repository‑specific guidelines, preventing quality degradation in long sessions and avoiding unnecessary API invention. Developers and teams using Claude Code, Codex, Gemini CLI, or Cursor benefit from optimal context management and clear separation of responsibilities between skills and rules.
Peft Fine Tuning — The Peft Fine Tuning skill performs parameter‑efficient fine‑tuning of large language models with LoRA, QLoRA, and over 25 additional techniques. It enables users to adapt 7B‑70B models such as Llama or Mistral to their own data or custom tasks, even on consumer GPUs. This skill is ideal for developers and data scientists who need efficient, scalable model customization without extensive computational resources.
Fireworks AI Assistant — Provides expert guidance for Fireworks AI, helping developers integrate the inference API, fine-tune models, and deploy custom endpoints.
RAG Systems Architect — This expert assists developers in building high-quality Retrieval-Augmented Generation systems by optimizing embedding models, vector database architecture, and advanced retrieval strategies for LLMs.
Embedding Model Trainer — This skill provides specialized templates and guidance for developers to fine-tune SentenceTransformer and CrossEncoder models for advanced retrieval and semantic similarity tasks.
Together AI Platform — Together AI provides a cloud platform for developers to run and fine-tune open-source AI models via inference APIs.
Azure AI — The Azure AI skill combines search, speech, OpenAI, and document intelligence services to deliver powerful vector, hybrid, and semantic search capabilities. It also converts audio and scanned documents into searchable text through speech‑to‑text, text‑to‑speech, transcription, and OCR functions. Developers and data professionals use it to streamline end‑to‑end AI‑powered content retrieval and processing.
AI Tool Architect — This skill guides developers in designing robust, efficient AI agent tools through optimized schemas and descriptions to ensure reliable model performance.
Prompt Refinement Specialist — This tool refines AI prompt files by applying proven structural patterns and linguistic improvements to ensure optimal performance for end users.
Gemini Enterprise Integrator — This skill assists developers in integrating Gemini models into enterprise applications using the Google Gen AI SDK on the Agent Platform.
LLM Prompt Optimizer — This tool refines LLM prompts using production trace data and performance metrics to help developers improve AI output quality through data-driven optimization.
Google Cloud Solution Agentic AI Data Science Workflow — The Google Cloud Solution Agentic AI Data Science Workflow designs a customized, multi-product agentic data science architecture on Google Cloud, embedding opinionated best practices. It is used by architects building agent-based data analytics or machine learning solutions across multiple products. The skill is not intended for simple queries, non-agentic pipelines, general cloud reviews, or writing agent code.
Huggingface Spaces — Build, deploy, and maintain applications on Hugging Face Spaces — Gradio / Docker / Static SDKs, ZeroGPU and dedicated hardware, model loading, debugging, buckets, inference providers, community grants.
Transformer Model Trainer — This skill enables developers to fine-tune and align transformer language models using advanced TRL techniques like SFT, DPO, GRPO, and reward model training.
Agent Tool Security — Implements defense-in-depth controls for AI agents to prevent unauthorized tool execution and mitigate risks from prompt injection or malicious tool usage.
Jev Use — Route enumerable judgment steps - did it work, which option, how risky, is this safe to run - to the Jev judgment model through the jev_judge and jev_gate MCP tools, batched into one call per state.
Microsoft Foundry — The Microsoft Foundry skill orchestrates agents, models, and resources from build to deployment, providing tools for prompt optimization, continuous monitoring, CI/CD workflows, dataset curation, and fine‑grained RBAC and quota controls. Data scientists, AI engineers, and operations teams use it to automate and streamline AI solution deployment and maintenance on Microsoft Foundry.
LLM Evaluation Pipeline — This tool helps AI developers improve application quality by establishing automated evaluation pipelines, creating golden datasets, and generating actionable insights for performance optimization.
TAO AutoML Optimizer — Automates hyperparameter optimization and experiment tracking for NVIDIA TAO networks, assisting developers in efficiently tuning model training, evaluation, and inference workflows.
Ruby MCP Generator — This tool generates a production-ready Model Context Protocol server project in Ruby, helping developers quickly scaffold robust AI-integrated server applications.
Copilot Agent SDK — This SDK empowers developers to embed production-tested AI agent workflows into applications, streamlining custom tool creation, session management, and seamless MCP server integration.
Multi-Agent System Architect — This skill empowers developers to design, build, and scale production-grade multi-agent workflows using LangGraph, LangChain, and advanced autonomous agent orchestration frameworks.
Azure Openai — Azure OpenAI delivers Microsoft’s hosted OpenAI models—including GPT, GPT‑image, Whisper, and embeddings—on Azure infrastructure, providing secure, compliant access via Managed Identity or Entra ID. It supports content filtering, regional data residency, and private networking, enabling enterprises to deploy advanced AI models within a regulated Azure environment. This service benefits organizations that require robust, compliant AI deployment across their Azure ecosystem.
Mistral AI Interface — Provides access to Mistral AI's language models for code generation, multilingual tasks, and GDPR-compliant AI inference.
Hugging Face Toolkit — Enables users to leverage Hugging Face's libraries for tasks like model inference, fine-tuning, and publishing, benefiting ML practitioners.
LLM Security Guardrails — This skill implements robust input and output validation guardrails for LLM applications to ensure safety, compliance, and protection against malicious prompt injections.
Agentic Evaluation Framework — This framework provides developers with systematic patterns for implementing self-critique and iterative refinement loops to significantly enhance the quality of AI agent outputs.
Gemini Media Generator — This tool enables developers to generate and edit high-quality images and videos using Google Gemini models with integrated cost-approval and audit logging.
Mem0 — Mem0 equips LLM applications and agents with personalized long‑term memory by extracting and storing conversational facts in a vector database, enabling continuity across sessions. Developers use this capability to build chatbots and agents that remember user preferences, delivering a more coherent and user‑centric experience for AI‑powered services.
Yann LeCun Persona — This agent simulates Yann LeCun to provide expert insights on deep learning, neural networks, and AI research for developers and researchers.
System Prompt Optimizer — This tool converts verbose human-written instructions into efficient, AI-native structured formats to improve model compliance and reduce token consumption for developers.
Arize Evaluation Manager — This skill automates LLM-as-judge evaluation workflows on the Arize platform, helping developers manage evaluators, execute scoring tasks, and monitor model performance metrics.
Advanced Evaluation — Design LLM-as-judge rubrics, direct scores and pairwise comparisons. Use for judge models, evals, position bias and calibration against human labels.
Agent Framework Azure AI Py — Build persistent agents on Azure AI Foundry using the Microsoft Agent Framework Python SDK.
Agent Swarm Orchestration — The skill orchestrates multiple AI agents to collaborate on complex tasks, managing routing, handoffs, consensus, memory sharing, and quality gates. It enables teams building multi‑agent systems, specialist pipelines, or autonomous agent swarms to implement hierarchical, mesh, or pipeline topologies with supervision and quality control, ensuring seamless agent-to-agent communication and robust orchestration.
Llm Evaluation — Master comprehensive evaluation strategies for LLM applications, from automated metrics to human evaluation and A/B testing.
Vertex AI Gemini — This skill helps enterprises deploy and fine-tune Gemini models on Google Cloud Vertex AI with enterprise-grade features.
Ragas — Ragas is an open‑source Python framework that evaluates retrieval‑augmented generation pipelines and LLM applications using metrics such as faithfulness, response relevancy, context precision, and context recall. It allows developers to score RAG systems, generate synthetic test sets from documents, embed quality checks into CI pipelines, and create custom LLM‑judged metrics. Data scientists, ML engineers, and product teams benefit from its rigorous, automated assessment of retrieval‑augmented generation performance.
Embedding Optimization Guide — This guide assists developers in selecting, fine-tuning, and optimizing embedding models to improve the performance and accuracy of vector search applications.
Sentence Transformer Trainer — This skill facilitates training and fine-tuning diverse embedding models, helping developers implement efficient retrieval, reranking, and classification pipelines using standardized production templates.
AI Model Recommender — This tool identifies optimal AI models based on specific task requirements and hardware constraints, helping developers and data scientists select the best-performing solutions.
Retrieving Developer Knowledge — The skill searches, retrieves, and synthesizes official Google developer documentation across Cloud, AI/Gemini, Android, Chrome, Web, Flutter, Go, Firebase, and related platforms. It integrates with the Developer Knowledge MCP server or REST API to provide accurate gcloud CLI commands, API syntax, IAM permissions, and architectural comparisons. Developers and architects use it to quickly locate authoritative documentation and make informed product choices.
Serverless AI Inference — Deploy and run AI models instantly with fal.ai, enabling developers to generate media and run ML inference at scale.
Langchain Architecture — Master the LangChain framework for building sophisticated LLM applications with agents, chains, memory, and tool integration.
Prompt Safety Auditor — This tool analyzes AI prompts for security vulnerabilities, bias, and safety risks while providing actionable improvements for developers building responsible and effective AI applications.
Groq API Assistant — Provides expert guidance for integrating Groq's ultra-fast LLM inference API into real-time AI applications, assisting developers.
Autonomous Agent Architect — This workflow assists developers in designing, implementing, and orchestrating autonomous AI agents, multi-agent systems, and complex tool integrations for scalable applications.
Prompt Engineering Expert — This guide provides advanced prompt engineering patterns and optimization techniques to help developers improve LLM performance, reliability, and output quality for complex tasks.
E2B — Sandboxed Code Execution for AI — E2B provides AI agents with secure, isolated sandboxes—ephemeral Linux VMs equipped with a filesystem, package managers, and internet access—to execute code safely. It allows users to run Python or JavaScript, install packages, read and write files, or perform long processes without compromising the host system. Developers integrating code execution into LangChain, CrewAI, or custom agent workflows benefit from this skill.
Crawl4AI Web Crawler — Assists developers in extracting clean, structured website data for AI applications like LLM training and RAG pipelines using Crawl4AI.
Pinecone Vector Database — This skill helps AI developers use Pinecone, a managed vector database, to build semantic search and RAG applications.
AI Product Architect — This skill assists developers in building robust, scalable, and cost-effective AI-powered products by implementing reliable LLM integration patterns and trusted user experiences.