Claude Skills by anubhavg-icpl

|. Use when you need help with magazine poster.

Discover, list, create, edit, toggle, copy, move, and delete AI agent skills across 11 tools (Cursor, Claude, Agents, Windsurf, Copilot, Codex, Cline, Aider, Continue, Roo Code, Augment). Use when you need help with manage skills.

Pick a managed LLM platform (Bedrock, Azure OpenAI, Vertex AI) and a second for redundancy, given workload, SLA, and compliance requirements — then produce a FinOps instrumentation plan. Use when you need help with managed platform picker.

markdown-formatter. Use when you need help with markdown formatter.

Pick the right multi-agent RL regime (IPPO, CTDE, self-play, league) for a given task. Use when you need help with marl architect.

Choose a MARL algorithm (MADDPG, QMIX, MAPPO, IQL, or extensions) for a given multi-agent task. Consider cooperative vs competitive, action-space type, heterogeneity, reward structure, and scale. Use when you need help with marl picker.

Deep expertise in Marqo — end-to-end vector search with embedding inference baked in, ONNX/GPU acceleration, and ecommerce-specialized models. Use when implementing vector search, embeddings storage, or similarity queries with marqo.

Martin Kleppmann — Designing Data-Intensive Applications, CRDTs, local-first software, Automerge. Use when you want code review, architecture advice, or opinions in the style of martin kleppmann.

Run a MAST-style failure-mode audit on a multi-agent system. Categorize execution-trace failures into Specification / Coordination / Verification and the Groupthink families; rank mitigations by expected failure reduction. Use when you need help with mast auditor.

Expert in the Mastercard design system - Global payments network. Warm cream canvas, orbital pill shapes, editorial warmth. Use when building UI components, applying design tokens, or implementing visual styles for mastercard.

Build TypeScript AI agents and workflows with Mastra's modern stack from the Gatsby team. Use when building AI applications with mastra.

Expert in the Material design system - Google's Material Design with layered surfaces, dynamic theming, built-in motion, and responsive cross-platform patterns. Use when building UI components, applying design tokens, or implementing visual styles for material.

Expert in Materialize streaming SQL, sources, sinks, and incremental views. Use when working with materialize for data processing, streaming, or analytics.

Expert in the mathematical foundations behind every AI algorithm from the AI Engineering from Scratch curriculum. Use when you need help with math foundations.

Evaluate a policy via Monte Carlo rollouts and produce a convergence report with DP-comparison if available. Use when you need help with mc evaluator.

Produce the full MCP Apps contract for a tool that needs an interactive UI resource. Use when you need help with mcp apps spec.

Wire production MCP authorization (RFC 8414, 7591, 8707, 7636 PKCE, 9728) onto iii primitives — registerTrigger for HTTP/cron, registerFunction for validation, state::* for JWKS cache. Use when you need help with mcp auth iii wiring.

Given a declarative list of MCP servers (name, command, args), scaffold a multi-server client with handshake, namespace merge, and routing. Use when you need help with mcp client harness.

MCP server development including tool design, resource endpoints, prompt templates, and transport configuration. Use when you need help with mcp development.

Given a pcap-style transcript of an MCP client-server conversation, annotate every message with its primitive, lifecycle phase, and capability dependency. Use when you need help with mcp handshake tracer.

Design and scaffold an MCP server with tools, resources, and safety defaults. Use when you need help with mcp server designer.

Deploy a production MCP server with StreamableHTTP, OAuth 2.1 scopes, OPA policy, human-approval gate for destructive tools, and a registry for discovery. Use when you need help with mcp server platform.

Scaffold a domain-specific MCP server with the right tools/resources/prompts split and SDK graduation path. Use when you need help with mcp server scaffolder.

Produce a threat model for an MCP deployment naming the applicable attack classes, defenses in place, and Rule-of-Two violations. Use when you need help with mcp threat model.

Produce a migration plan from legacy HTTP+SSE to Streamable HTTP with session id continuity and Origin validation. Use when you need help with mcp transport migrator.

Given a task description, produce a Markov Decision Process spec and flag formulation risks before training. Use when you need help with mdp modeler.

|. Use when you need help with meeting notes.

Build long-term, scalable memory layers for AI agents with Mem0. Use when building AI applications with mem0.

Audit a multi-agent system's shared-memory design for provenance, versioning, verifier separation, and projection schema. Flag memory-poisoning exposure before production. Use when you need help with memory auditor.

Generate a Letta-shaped three-tier memory system (core blocks, recall, archival) with a sleep-time consolidation agent off the critical path. Use when you need help with memory blocks.

Expert in memory profiling, leak detection, and memory optimization across languages. Use when diagnosing, troubleshooting, or fixing bugs with memory profiler.

MergeKit recipes — SLERP, TIES, DARE, model soups, task arithmetic, MoE merging. Use when fine-tuning, training, or adapting language models with merge experts techniques.

Classify an observed safety failure as outer-alignment, proxy-inner, or deceptive-inner. Use when you need help with mesa diagnostic.

Expert in the Meta (Store) design system - Tech retail store. Photography-first, binary light/dark surfaces, Meta Blue CTAs. Use when building UI components, applying design tokens, or implementing visual styles for meta.

Recommend head count, KV-head count, and projection strategy (MHA / MQA / GQA / MLA) for a new transformer. Use when you need help with mha configurator.

Microservices design patterns including service mesh, event-driven architecture, saga pattern, and API gateway. Use when building UI with microservices design language and visual style.

Build a repo-level code migration agent that combines deterministic recipes with an agent fallback loop, passes MigrationBench, and publishes a failure taxonomy. Use when you need help with migration agent.

Expert in Million.js block-DOM compiler that speeds up React rendering up to 70%. Use when building web applications with million.

Deep expertise in Milvus 2.5+ — index zoo (HNSW/DiskANN/IVF/SCANN/CAGRA), partitions, multi-vector hybrid search, GPU indexes, and Milvus Lite. Use when implementing vector search, embeddings storage, or similarity queries with milvus.

Expert in the Minimal design system - Stripped-back design emphasizing whitespace, clean typography, and restrained color for maximum clarity and focus. Use when building UI components, applying design tokens, or implementing visual styles for minimal.

Expert in the MiniMax design system - AI model provider. Bold dark interface with neon accents. Use when building UI components, applying design tokens, or implementing visual styles for minimax.

Expert in the Mintlify design system - Documentation platform. Clean, green-accented, reading-optimized. Use when building UI components, applying design tokens, or implementing visual styles for mintlify.

Expert in the Miro design system - Visual collaboration. Bright yellow accent, infinite canvas aesthetic. Use when building UI components, applying design tokens, or implementing visual styles for miro.

misc. Use when you need help with misc.

Expert in the Mistral AI design system - Open-weight LLM provider. French-engineered minimalism, purple-toned. Use when building UI components, applying design tokens, or implementing visual styles for mistral ai.

Mitchell Hashimoto — HashiCorp founder, infra-as-code, Go, Zig, terminal craft, ship-the-thing. Use when you want code review, architecture advice, or opinions in the style of mitchell hashimoto.

Expert in classical machine learning, the backbone of most production AI, from the AI Engineering from Scratch curriculum. Use when you need help with ml fundamentals.

MLflow Tracing for LLMs, Prompt Engineering UI, mlflow.evaluate(), prompt registry. Use when evaluating, monitoring, or observing LLM performance with mlflow llm.

Expert in MLOps practices, ML pipelines, model deployment, and monitoring. Use when you need deep expertise in mlops.

Run, quantize, fine-tune (LoRA/QLoRA), and serve LLMs and VLMs natively on Apple Silicon with MLX and mlx-lm. Use when deploying, running, or configuring local LLM inference with mlx apple silicon.