Guides you through writing and registering Kubernetes webhooks and CEL admission validation rules in the Run:ai repo.
dsx-ai-factory/run-ai
Helps developers and admins work with the Run:ai Kubernetes platform, from writing docs and services to checking GPU health and quota.
- 15 skills
- 728 stars
The 5 most installed
Lists all buildable Run:ai services grouped by type, reading from CLAUDE.md without scanning the repo.
Audits GPU quota fairness across a Run:ai cluster, showing which projects are starved or borrowing too much.
Diagnoses why a Run:ai workload failed, crashed, or got stuck in errors like OOMKilled or ImagePullBackOff.
Generates a feature guide page for a Run:ai scheduling or infrastructure feature, covering concepts, setup, and API use.
The other 10 skills
Routes installation-doc changes to the right files and keeps the guided and manual install paths in sync.
Generates a quick start page for a Run:ai capability with step-by-step UI, CLI, and API instructions.
Generates documentation for a new Run:ai UI resource type, covering its table and create, edit, and delete flows.
Checks Run:ai cluster and node health, giving a per-node-pool verdict with the main blocker and GPU status.
Shows whether the GPUs held by projects and node pools are actually busy, flagging idle and underused ones.
Summarizes the health of Run:ai workloads in a scope, or of a single workload.
Detects changes to MCP server tools and agent skills, updates the reference docs, and opens a pull request.
Enforces using Run:ai wrapper components instead of raw Quasar components when writing Vue files in the frontend.
Checks a service README against its Makefile targets, for one service or all undocumented ones.
Generates or updates a detailed README.md and CLAUDE.md for a Run:ai monorepo service.