Agent Skills

Author custom MCP servers: tool schema design, resource exposure, prompt templates, transport selection, and security review.

Checks

Metadata only

Evidence

1 source

Route Foundry development work to current model, hosted-agent, toolbox, knowledge, memory, observability, and governance guidance from Microsoft.

Checks

Metadata only

Evidence

1 source

Design end-to-end ML pipelines: data ingestion, feature engineering, model training, evaluation, experiment tracking, and MLflow/W&B integration.

Checks

Metadata only

Evidence

1 source

Build multimodal AI pipelines: vision-language models, image-to-text, text-to-image, video understanding, and audio transcription.

Checks

Metadata only

Evidence

1 source

Run routed chat and asynchronous deep-research jobs against an approved local or self-hosted NVIDIA AI-Q Blueprint backend.

Checks

Recorded checks passed

Evidence

1 source

Formulate and solve routing, scheduling, linear, and quadratic optimization problems with NVIDIA cuOpt APIs and explicit result validation.

Checks

Recorded checks passed

Evidence

1 source

Deploy, configure, troubleshoot, and operate NVIDIA's RAG Blueprint through documented Docker Compose and Helm workflows.

Checks

Recorded checks passed

Evidence

1 source

Ground OpenAI product and API implementation work in current official documentation, with source-aware browsing and citations.

Checks

Metadata only

Evidence

1 source

Design type-safe Prisma schemas, write efficient queries, manage migrations, and configure connection pooling for production.

Checks

Metadata only

Evidence

1 source

Apply chain-of-thought, few-shot, structured output, tool-use, and system-prompt patterns to improve LLM reliability.

Checks

Metadata only

Evidence

1 source

Optimize RAG systems: semantic chunking strategies, embedding model selection, hybrid search, reranking, and context window management.

Checks

Metadata only

Evidence

1 source

Evaluate RAG pipelines with RAGAS metrics: faithfulness, context precision/recall, answer relevance, and golden dataset curation.

Checks

Metadata only

Evidence

1 source

Design and build retrieval-augmented generation pipelines: chunking, embedding, vector stores, hybrid search, and re-ranking.

Checks

Metadata only

Evidence

1 source

PostgreSQL query, schema, and configuration optimization guidance from Supabase.

Checks

Metadata only

Evidence

1 source

Apply Tailwind CSS v4 design tokens, responsive utilities, dark-mode strategies, and component composition patterns.

Checks

Metadata only

Evidence

1 source

Full browser automation for agents — navigate, click, fill forms, extract structured data, and take screenshots.

Checks

Metadata only

Evidence

1 source

Apply compound component, render props, and context patterns for scalable, composable React component APIs.

Checks

Metadata only

Evidence

1 source

Deploy Next.js and other web apps to Vercel with correct vercel.json configuration, environment variables, and domain setup.

Checks

Metadata only

Evidence

1 source

Versioned React and Next.js performance guidance maintained by Vercel Engineering.

Checks

Metadata only

Evidence

1 source