Edge AI
Use when the user asks about edge AI, on-device inference, TensorRT on Jetson, ONNX Runtime, LiteRT (TFLite), TinyML, int8 quantization, Coral, NPUs or ExecuTorch. Produces export, quantization and on-device benchmarking plans. Not for model training (use ml-deep-learning) or cloud LLM serving (use ml-inference-optimization).
Install
npx skills add j4flmao/agent-skills --skill edge-aigit clone --depth 1 https://github.com/j4flmao/agent-skills.git /tmp/agent-skills
mkdir -p ~/.claude/skills
cp -r /tmp/agent-skills/skills/embedded-systems/edge-ai ~/.claude/skills/edge-aiPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# Edge AI Deployment
## Purpose
A model that works in a notebook still has to fit the device's memory, latency, power and operator support. This skill picks the runtime for the target hardware, converts and quantizes the model without silently losing accuracy, and measures it on the device.
## Agent Protocol
### Trigger
"edge AI", "on-device inference", "TensorRT", "ONNX Runtime", "TFLite", "LiteRT", "TinyML", "int8 quantization", "Jetson", "Coral", "NPU", "ExecuTorch".
### Input Context
- Model framework, architecture, input shape, accuracy metric and baseline
- Target hardware (MCU, Cortex-A CPU, mobile NPU/GPU, Jetson, Hailo, Coral) and OS
- Latency, throughput, memory and power budgets
### Output Artifact
Export/convert/quantize scripts, runtime integration code, benchmark plan. No file unless requested.
### Response Format
```
Target: {hardware + OS}
Path: {framework -> format -> runtime/EP}
Precision: {FP32 | FP16 | INT8 (PTQ/QAT) | INT4}
Calibration: {dataset, size}
Result: {latency p50/p99, memory, accuracy delta vs FP32}
Risks: {unsupported ops, fallback to CPU, accuracy drop}
```
## 1. Choose the Runtime by Hardware
| Hardware | Typical runtime | Notes |
|----------|-----------------|-------|
| NVIDIA Jetson / dGPU | TensorRT (or ONNX Runtime with TensorRT/CUDA EP) | Engines are built per GPU + TensorRT version; rebuild on upgrade |Details
- Skill name
edge-ai- Repository
- j4flmao/agent-skills
- Path
skills/embedded-systems/edge-ai/SKILL.md- License
- MIT
- Repo updated
- 17 November 2026
FAQ
What does the Edge AI skill do?
Use when the user asks about edge AI, on-device inference, TensorRT on Jetson, ONNX Runtime, LiteRT (TFLite), TinyML, int8 quantization, Coral, NPUs or ExecuTorch. Produces export, quantization and on-device benchmarking plans. Not for model training (use ml-deep-learning) or cloud LLM serving (use ml-inference-optimization).
How do I install the Edge AI skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add j4flmao/agent-skills --skill edge-ai. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Edge AI popular?
It lives in j4flmao/agent-skills, which has 45 GitHub stars and 6 forks.
Related skills
Agent Observability
45j4flmao/agent-skills
Use when the user asks about agent observability, tracing agent reasoning, debugging an agent loop, OpenTelemetry GenAI spans for LLM calls, LLM cost tracking per agent, agent audit logs or hallucination/eval monitoring in production. Produces instrumentation config, trace schema, dashboards, alerts and cost attribution. Not for generic service tracing without LLMs (use devops-opentelemetry).
AI Agent Architectures
45j4flmao/agent-skills
Use when the user asks to build an AI agent, design an agent that calls tools and keeps memory, write an agent system prompt or pick an agent framework (Claude Agent SDK, OpenAI Agents SDK, LangGraph). Produces an architecture brief covering loop, tools, memory, guardrails and evals, routing depth to child skills. Not for workflow patterns (use ai-agentic-workflows).
AI Agent Core Loop
45j4flmao/agent-skills
Use when the user asks about the agent loop, ReAct vs plan-and-solve, re-planning, Reflexion, an agent stuck in a loop, max iterations or termination conditions. Produces a loop spec or patch with state, success predicate, caps and no-progress detection. Not for overall agent design (use ai-agent-architectures) or tool schemas (use tool-calling-principles).
Extreme Quantization Mastery
45j4flmao/agent-skills
Use when the user asks about FP8, INT4, NVFP4, MXFP4, GPTQ, AWQ, SmoothQuant, W4A16, W8A8, NF4 or bitsandbytes for LLMs on NVIDIA GPUs. Produces a quantization recipe, code and accuracy-eval plan matched to the GPU and serving engine. Not for format/GGUF basics (use llm-model-quantization) or TensorRT engines (use tensorrt-optimization).
Data Vector Databases
45j4flmao/agent-skills
Use when the user asks to pick an index type for vector similarity search, pgvector, Qdrant, Milvus, Pinecone, Weaviate, HNSW vs IVF vs PQ, metadata filtering or storing embeddings. Produces vector store choice, collection schema and ANN index settings for recall, latency and memory. Not for full RAG pipeline design (use llm-rag-retrieval-internals).
Genai Vision
45j4flmao/agent-skills
Use when the user asks to generate images with Stable Diffusion, SDXL, FLUX or a diffusers pipeline, img2img, inpainting, ControlNet, LoRA styles, or transcribe audio with Whisper. Produces model choice with license note, pipeline code, memory settings and serving outline. Not for VLM image understanding (use multimodal-pipelines) or detection models (use ml-computer-vision).
Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.