aruing · docs GitHub ↗

aruing · v0.1.4

Evidence, not vibes.

A Kubernetes ops agent that answers with tools and evidence. Ask in natural language; get root causes with a traceable evidence chain. The immersive cluster stage → runs on the home page.

how it works

Innate and adaptive —
an immune system for ops.

Two layers of response, exactly like biology: a fast always-on patrol for routine questions, and a forced evidence-adjudication chain when you need the root cause.

Always-on patrol

Observe → think → call tools → answer. Routine questions — “what’s installed?”, “how is this configured?” — get answers grounded in live cluster data, never stale docs.

≈ innate immunity

Forced adjudication

Ask for a root cause and aruing escalates into a formal chain: hypothesis → tasks → Evidence → Verdict → Report. No “I think it’s X” without a tool trail.

≈ adaptive immunity

Memory that survives

Sessions, messages and diagnostic ledgers persist to disk. Come back tomorrow and ask “why was redis slow last night?” — the evidence is still there, still citable.

≈ immune memory

The metaphor is load-bearing, not decoration: containers are cells, faults are pathogens and mutations, reports are antibodies, and the run ledger is immune memory.

demo

Ask anything.
Get receipts.

A real session shape: a baseline patrol escalates into a formal diagnosis, files the report to the ledger, and answers follow-ups by re-reading its own evidence.

aruing chat — demo

install

One line to a
guarded cluster.

curl -fsSL https://raw.githubusercontent.com/Aruing/Aruing/main/scripts/install.sh | bash

Then run aruing connect to point it at any OpenAI-compatible LLM. Upgrades: aruing update.

macOS / Linux kubectl any OpenAI-compatible LLM Go 1.26+ (from source)