Skill Claude CodeCodex
Manage LLM context windows efficiently. Use when implementing context compaction, conversation summarization, token budget management, sliding window strategies, or optimizing prompt length for cost and quality.
Skill Claude CodeCodex
Design strategies to detect, monitor, and remediate data drift in GenAI applications and ML pipelines. Use when monitoring LLM input patterns, detecting query distribution shifts, tracking embedding drift, building RAG retrieval quality monitoring, or establishing data governance for model inputs.
Skill Claude CodeCodex
Design and implement feedback loops for continuous AI/ML improvement. Use when building RLHF/RLAIF pipelines, user feedback collection systems, reward modeling, iterative model refinement workflows, or online learning strategies.
Skill Claude CodeCodex
Implement agentic applications with the Microsoft Foundry SDKs. Use when coding against Microsoft Foundry project clients, agent operations, evaluations, datasets, indexes, tracing, or SDK-driven tool wiring rather than only high-level architecture guidance.
Skill Claude CodeCodex
Operate generative AI systems safely in production. Use when designing GenAIOps workflows for prompt and model release management, evaluation gates, observability, drift detection, rollback, governance, and continuous improvement.
Skill Claude CodeCodex
Progressive context refinement for subagents and tool-using LLMs. Use when a parent agent must delegate research, code reading, or document Q&A to a subagent with a tight context budget. Replaces one-shot context dumps with a query -> retrieve -> filter -> requery loop that returns only the evidence the parent needs.
Skill Claude CodeCodex
Build stateful, durable agent workflows with LangGraph. Use when implementing graph-based orchestration, long-running agent state, human-in-the-loop checkpoints, subgraphs, memory, or LangSmith-observed agent execution.
Skill Claude CodeCodex
Front LLM calls behind a gateway: model routing, fallbacks, semantic caching, rate limits, key vaulting, cost controls, and unified APIs. Covers LiteLLM, Portkey, Azure AI Gateway / APIM, OpenRouter, AWS Bedrock proxy, and self-built gateways. Includes routing policies (cost/latency/capability/RBAC) and caching (exact…
Skill Claude CodeCodex
Build MCP Apps (ext-apps) that render interactive UI inside conversational AI clients. Use when creating visual tool outputs, interactive dashboards, form-based tools, or rich media experiences in MCP-compatible hosts like Claude Desktop, VS Code Copilot Chat, or other MCP clients that support the Apps specification.
Skill Claude CodeCodex needs its repo
Build Model Context Protocol (MCP) servers that expose tools, resources, and prompts to AI agents. Use when creating MCP servers in TypeScript or Python, defining MCP tools, implementing resource providers, or integrating MCP servers with AI agent workflows.
Skill Claude CodeCodex
Detect, monitor, and manage model drift in production GenAI and ML systems. Use when monitoring LLM output quality, detecting prompt regression, managing model version changes, implementing drift detection for traditional ML, or establishing model governance policies.
Skill Claude CodeCodex
Fine-tune foundation models for domain-specific tasks. Use when adapting LLMs with LoRA/QLoRA/PEFT, full fine-tuning, knowledge distillation, or designing training data pipelines for model customization.
Skill Claude CodeCodex
Design and operate multi-agent systems where several specialized LLM agents collaborate. Use when choosing between supervisor/worker, swarm/handoff, hierarchical, or graph patterns; selecting frameworks (AutoGen, CrewAI, OpenAI Swarm/Agents SDK, LangGraph, Microsoft Agent Framework, Google A2A); designing handoff…
Skill Claude Code
Use when designing coding-agent prompts, tool contracts, structured outputs and model-adaptive context, or evaluating prompt changes.
Skill Claude CodeCodex
Version, test, promote, and roll back prompts like production artifacts. Use when managing prompt lifecycle, prompt variants, few-shot changes, rubric alignment, baseline comparisons, and safe prompt release workflows across models or environments.
Skill Claude CodeCodex
Design and build production RAG (Retrieval-Augmented Generation) pipelines. Use when implementing document ingestion, chunking strategies, embedding selection, vector search, hybrid retrieval, reranking, or generation with grounding.
Skill Claude CodeCodex
Use reasoning / thinking models (GPT-6 Astra, Claude Opus 5.5, DeepSeek R1, Gemini Thinking) effectively. Covers when to choose reasoning vs fast models, prompt patterns for reasoners, effort / thinking controls, structured outputs with reasoning, cost/latency trade-offs, and combining reasoners with fast models.
Skill Claude CodeCodex
Generate synthetic data for fine-tuning, eval-set bootstrapping, RAG corpus augmentation, and rare-case coverage. Covers Self-Instruct, Evol-Instruct, persona-based generation, distillation from larger models, dataset curation (filtering, dedup, decontamination), and provenance / dataset cards.
Skill Claude CodeCodex
Design robust tool-use and function-calling for LLM agents. Use when defining tool schemas (JSON Schema, OpenAPI), enabling parallel tool calls, structured outputs (JSON mode / responseformat), tool error handling, retries, idempotency, and tool selection prompts. Covers OpenAI tools, Anthropic tooluse, Gemini…
Skill Claude CodeCodex
Choose, configure, and operate vector databases for embeddings and hybrid search. Covers Azure AI Search, Pinecone, Qdrant, Weaviate, Milvus, pgvector, LanceDB, MongoDB Atlas Vector, Elasticsearch / OpenSearch kNN. Selection criteria, index types (HNSW, IVF, DiskANN), filters, hybrid (BM25 + vector), embedding model…
Skill Claude CodeCodex
Build low-latency voice agents using speech-to-speech (S2S) realtime APIs and cascaded STT->LLM->TTS pipelines. Covers OpenAI Realtime / GPT Realtime, Azure Voice Live, Gemini Live, Deepgram Voice Agent, ElevenLabs Conversational AI; turn-taking, barge-in, latency budgets, tool use during speech, and telephony…
Skill Claude CodeCodex
Design robust REST APIs with proper versioning, pagination, error handling, rate limiting, and documentation. Use when creating new API endpoints, designing resource naming conventions, implementing pagination or filtering, adding rate limiting, or documenting APIs with OpenAPI/Swagger.
Skill Claude Code
Apply fundamental coding principles (SOLID, DRY, KISS) and structure projects for maintainability with clean architecture and separation of concerns. Use when refactoring, reviewing patterns, setting up project structure, or organizing modules.
Skill Claude CodeCodex
Estimate and attribute cloud running cost for a proposed architecture before it is built. Use when writing an ADR options table, sizing a prototype, comparing managed versus self-hosted services, or answering "what will this cost to leave running". Produces a load-enveloped estimate with an idle-versus-active split…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: