skills.rest
moreh-dev
moreh-devOFFICIAL
Explore all skills by this author →
WebsiteGitHub

guide-heimdall

Configure Heimdall routing and scheduling for MoAI Inference Framework on Kubernetes.

Updated Oct 23, 2025
Search Tags:#kubernetes#helm#scheduling#inference#gateway#heimdall
One-click install
npx skills add https://github.com/moreh-dev/mif --skill guide-heimdall
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: guide-heimdall
Source: https://github.com/moreh-dev/mif/tree/main/skills/guide-heimdall
Command: npx skills add https://github.com/moreh-dev/mif --skill guide-heimdall

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Heimdall handles routing and scheduling for MoAI Inference Framework, ensuring requests are directed to the most suitable inference pods based on configured plugins and profiles.

Core Features & Use Cases

  • Plugin-driven decision pipeline with filters, scorers, and pickers
  • Management of EndpointPickerConfig and InferencePool resources for Kubernetes deployments
  • Helm-based deployment, observability, and troubleshooting guidance

Quick Start

Create a Heimdall values file (heimdall-values.yaml) with a basic EndpointPickerConfig and a default scheduling profile, then install using Helm.

Frequently Asked Questions about guide-heimdall

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
What is the best way to route inference requests across pods using Gateway API?▼

The best way to route inference requests using Gateway API is to configure Heimdall via the Gateway API Inference Extension CRDs. This integrates scheduling profiles and prefill/decode filters to optimize request distribution across your deployment.

Can I use Helm to install and manage Heimdall scheduling profiles?▼

Yes, you can use Helm to install and manage Heimdall scheduling profiles. You create a heimdall-values.yaml file containing your desired EndpointPickerConfig, image pull secrets, and gateway class selection, then deploy it using Helm.

Does Heimdall support different scheduling modes for prefill and decode operations?▼

Yes, Heimdall supports different scheduling modes including PD and aggregate modes. You can configure specific prefill and decode filters within your scheduling profiles to optimize how requests are distributed across inference pods.

Why does my Heimdall deployment require image pull secrets in the values file?▼

Heimdall deployment requires image pull secrets in the heimdall-values.yaml file to authenticate and access container images from private registries. Without proper image pull secrets and gateway class selection, the Kubernetes deployment will fail to provision pods correctly.

Related Skills

Explore More Software Engineering Skills→
dennisonbertramdennisonbertram
1

helm-deploy

Automate Helm chart installation, upgrade, and rollback for Kubernetes deployments.

Community
Intermediate
DevelopersGlobalDevelopersGlobal
65

observability

Implement structured logging, distributed tracing, and alerting with OpenTelemetry.

Official
Advanced
totallyGregtotallyGreg
2

kgateway

Configure Kubernetes Gateway API resources with kgateway for Envoy-based routing.

Community
Advanced
finomfinom
52

decorators

Build and stack Vovk.ts decorators for Next.js route handlers.

Community
Intermediate