inference-serving-topology
LLM/model inference serving architecture: the engine → serving → orchestration layering (vLLM/SGLang/TensorRT-LLM, Triton, KServe/Ray Serve), KV-cache & continuous batching, prefill-decode disaggregation, and scaling. Architect-level topology, not model training. USE WHEN: designing model/LLM serving infra, "vLLM", "SGLang", "TensorRT-LLM", "Triton", "KServe", "Ray Serve", "continuous batching", "KV cache", "prefill decode", "TTFT", multi-GPU/multi-model serving, inference autoscaling. DO NOT USE FOR: on-device (use `edge-inference`); provider routing (use `model-gateway-routing`); RAG app logic (use rag/rag-frameworks skills).
What this skill does
# Inference Serving Topology ## The three layers (name which you're designing) 1. **Engine** — executes the model on accelerators: **vLLM**, **SGLang**, **TensorRT-LLM**. Owns paged **KV-cache**, **continuous (in-flight) batching**, quantization. This is where throughput/latency is won. 2. **Serving** — request routing, batching policy, API contract, metrics, rate limiting: **Triton** (production shell around an engine), **KServe**, **LiteLLM/Envoy AI Gateway**. 3. **Orchestration** — scaling, health, placement: **Kubernetes + KEDA**, **Ray Serve**, llm-d, GKE Inference Gateway. A common 2026 pairing: **vLLM as the token engine + Triton as the production shell**; **Ray Serve** when you need multi-GPU/multi-node distributed strategies. ## Levers that decide the topology - **KV-cache** is the memory bottleneck for LLMs → paged KV (vLLM), cache reuse, quantized KV. Drives max batch / context. - **Continuous batching** (vs static) is mandatory for throughput. - **Prefill–decode disaggregation**: split the compute-bound prefill from the memory-bound decode onto different pools → better utilization at scale. - **Parallelism**: tensor / pipeline / expert (MoE) / data-parallel attention — chosen by model size vs GPU memory. - **Targets**: state TTFT (time-to-first-token, low hundreds of ms) and inter-token latency (tens of ms) goals — they drive batching/parallelism. ## Scale ladder Single GPU + vLLM → multi-GPU one node → Ray Serve/KServe multi-node + KEDA autoscaling → disaggregated prefill/decode + multi-region. Don't jump tiers without a load/latency reason. ## When to recommend what - One model, moderate load → single-node vLLM (+ Triton for ops). - Many models / platform team → Triton + KServe on k8s. - Large model / high concurrency / SLO-driven → Ray Serve, disaggregation, tensor/expert parallelism.
Related in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.