Sglang Model Gateway | Agent Skills Atlas

About this skill

SGLang Model Gateway (sgl-model-gateway, formerly sgl-router) — Rust router fronting vLLM/SGLang inference workers on Kubernetes. Trigger on "sgl-model-gateway", "sgl-router", "sglang router", "smg", "amg", "model gateway", "inference gateway", "load balance vllm replicas", "fan out same model", "kubernetes vllm router", "cache-aware routing", "prefixhash policy", "PD disaggregation router", "--worker-urls", "--service-discovery", "--enable-mesh", "smg metrics". Covers: first-class vLLM gRPC backend (RuntimeType::Vllm) plus HTTP transparent-proxy for vanilla vLLM; eight policies; air-gapped recipe (gateway ignores HFENDPOINT, mount tokenizer on PVC); K8s manifests with modelid labels + per-model RBAC; three HA mitigations (single+PDB / sessionAffinity / --enable-mesh CRDT sync); pitfalls (vLLM HTTP discovery registers empty labels, gRPC probes need numeric ports, sglrouter → smg rename Dec 2025).

Quick start

Copy the install prompt. Use the green button above this README.
Send it to your agent. Paste it into your coding agent’s chat and send.
Review the setup. Your agent checks the skill’s instructions, requirements, and installation scope.

Installation guide →

Review, license & listing details

Review the source before using this skill

This is an imported listing. We haven’t certified its safety, license, or compatibility. Skills may include executable scripts.

Automated security scan: Partial

Only SKILL.md was scanned. Supporting files and scripts were not checked. No safety grade has been assigned.

0 potential findings · Scanned 11/4/2026 · Scanner 0.2.1

Content SHA-256: 01365e956a9e1ad1c9111a3adb1d87180769d783474fb3ec7d2aea7797e92be5

About scan coverage and limitations

Original listing metadata

Original name

sglang-model-gateway

Original description

SGLang Model Gateway (`sgl-model-gateway`, formerly `sgl-router`) — Rust router fronting vLLM/SGLang inference workers on Kubernetes. Trigger on "sgl-model-gateway", "sgl-router", "sglang router", "smg", "amg", "model gateway", "inference gateway", "load balance vllm replicas", "fan out same model", "kubernetes vllm router", "cache-aware routing", "prefix_hash policy", "PD disaggregation router", "--worker-urls", "--service-discovery", "--enable-mesh", "smg_* metrics". Covers: first-class vLLM gRPC backend (`RuntimeType::Vllm`) plus HTTP transparent-proxy for vanilla vLLM; eight policies; air-gapped recipe (gateway ignores `HF_ENDPOINT`, mount tokenizer on PVC); K8s manifests with `model_id` labels + per-model RBAC; three HA mitigations (single+PDB / `sessionAffinity` / `--enable-mesh` CRDT sync); pitfalls (vLLM HTTP discovery registers empty labels, gRPC probes need numeric ports, `sgl_router_*` → `smg_*` rename Dec 2025).

About this listing

Category: Engineering
Reported license: Not specified
Listing language (estimated): English
Review status: Not individually reviewed
Agent compatibility: Atlas installation supports Claude Code, Codex, and Cursor. Skill-specific dependencies and features are checked during setup.