llm-inference
Use when "LLM inference", "serving LLM", "vLLM", "llama.cpp", "GGUF", "text generation", "model serving", "inference optimization", "KV cache", "continuous batching", "speculative decoding", "local LLM", "CPU inference"
What this skill does
# LLM Inference High-performance inference engines for serving large language models. --- ## Engine Comparison | Engine | Best For | Hardware | Throughput | Setup | |--------|----------|----------|------------|-------| | **vLLM** | Production serving | GPU | Highest | Medium | | **llama.cpp** | Local/edge, CPU | CPU/GPU | Good | Easy | | **TGI** | HuggingFace models | GPU | High | Easy | | **Ollama** | Local desktop | CPU/GPU | Good | Easiest | | **TensorRT-LLM** | NVIDIA production | NVIDIA GPU | Highest | Complex | --- ## Decision Guide | Scenario | Recommendation | |----------|----------------| | Production API server | vLLM or TGI | | Maximum throughput | vLLM | | Local development | Ollama or llama.cpp | | CPU-only deployment | llama.cpp | | Edge/embedded | llama.cpp | | Apple Silicon | llama.cpp with Metal | | Quick experimentation | Ollama | | Privacy-sensitive (no cloud) | llama.cpp | --- ## vLLM Production-grade serving with PagedAttention for optimal GPU memory usage. ### Key Innovations | Feature | What It Does | |---------|--------------| | **PagedAttention** | Non-contiguous KV cache, better memory utilization | | **Continuous batching** | Dynamic request grouping for throughput | | **Speculative decoding** | Small model drafts, large model verifies | **Strengths**: Highest throughput, OpenAI-compatible API, multi-GPU **Limitations**: GPU required, more complex setup **Key concept**: Serves OpenAI-compatible endpoints—drop-in replacement for OpenAI API. --- ## llama.cpp C++ inference for running models anywhere—laptops, phones, Raspberry Pi. ### Quantization Formats (GGUF) | Format | Size (7B) | Quality | Use Case | |--------|-----------|---------|----------| | **Q8_0** | ~7 GB | Highest | When you have RAM | | **Q6_K** | ~6 GB | High | Good balance | | **Q5_K_M** | ~5 GB | Good | Balanced | | **Q4_K_M** | ~4 GB | OK | Memory constrained | | **Q2_K** | ~2.5 GB | Low | Minimum viable | **Recommendation**: Q4_K_M for best quality/size balance. ### Memory Requirements | Model Size | Q4_K_M | RAM Needed | |------------|--------|------------| | 7B | ~4 GB | 8 GB | | 13B | ~7 GB | 16 GB | | 30B | ~17 GB | 32 GB | | 70B | ~38 GB | 64 GB | ### Platform Optimization | Platform | Key Setting | |----------|-------------| | **Apple Silicon** | `n_gpu_layers=-1` (Metal offload) | | **CUDA GPU** | `n_gpu_layers=-1` + `offload_kqv=True` | | **CPU only** | `n_gpu_layers=0` + set `n_threads` to core count | **Strengths**: Runs anywhere, GGUF format, Metal/CUDA support **Limitations**: Lower throughput than vLLM, single-user focused **Key concept**: GGUF format + quantization = run large models on consumer hardware. --- ## Key Optimization Concepts | Technique | What It Does | When to Use | |-----------|--------------|-------------| | **KV Cache** | Reuse attention computations | Always (automatic) | | **Continuous Batching** | Group requests dynamically | High-throughput serving | | **Tensor Parallelism** | Split model across GPUs | Large models | | **Quantization** | Reduce precision (fp16→int4) | Memory constrained | | **Speculative Decoding** | Small model drafts, large verifies | Latency sensitive | | **GPU Offloading** | Move layers to GPU | When GPU available | --- ## Common Parameters | Parameter | Purpose | Typical Value | |-----------|---------|---------------| | **n_ctx** | Context window size | 2048-8192 | | **n_gpu_layers** | Layers to offload | -1 (all) or 0 (none) | | **temperature** | Randomness | 0.0-1.0 | | **max_tokens** | Output limit | 100-2000 | | **n_threads** | CPU threads | Match core count | --- ## Troubleshooting | Issue | Solution | |-------|----------| | Out of memory | Reduce n_ctx, use smaller quant | | Slow inference | Enable GPU offload, use faster quant | | Model won't load | Check GGUF integrity, check RAM | | Metal not working | Reinstall with `-DLLAMA_METAL=on` | | Poor quality | Use higher quant (Q5_K_M, Q6_K) | ## Resources - vLLM: <https://docs.vllm.ai> - llama.cpp: <https://github.com/ggerganov/llama.cpp> - TGI: <https://huggingface.co/docs/text-generation-inference> - Ollama: <https://ollama.ai> - GGUF Models: <https://huggingface.co/TheBloke>
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.