{"slug": "break-a-kind-cluster-on-purpose-then-watch-an-observe-agent", "title": "Break a kind cluster on purpose, then watch an Observe agent", "summary": "Kprompt v0.5 ships an optional Observe agent that continuously watches Kubernetes clusters for failures, correlates incidents, and gates notifications via Slack or webhook, with Autopilot remaining propose-only. The kprompt-examples repository provides a fixture set with seven failure scenarios (CrashLoop, ImagePull, OOM, stalled rollout, unbound PVC, failing CronJob, missing Redis hostname) to demo the agent offline in heuristic mode without spending LLM tokens. The agent never applies, patches, or deletes by default, and Autopilot proposals require explicit approval.", "body_md": "One command from kprompt-examples: kind up, seven failure scenarios, verify they actually broke, then run the Observe agent offline in heuristic mode — $0, no silent auto-heal.\n\n*Originally published at https://kprompt.ai/blog/observe-agent-kind-demo.*\n\nv0.5 shipped the optional [Observe agent](https://kprompt.ai/docs/agent): always-on watch → correlated Incident → gated Slack/webhook, with Autopilot still propose-only. The missing piece for a live walkthrough was a `payments`\n\nnamespace that actually misbehaves — not a slide claiming CrashLoopBackOff.\n\n[kprompt-examples](https://github.com/kprompt/kprompt-examples) is that fixture set. Break the cluster on purpose, then watch the agent react without spending an LLM token.\n\nNeeds Docker, kind, kubectl, and kprompt with the `agent`\n\nsubcommand. Heuristic mode — no API key, no spend. `DEMO_SECONDS=60`\n\nstretches the agent window for recordings.\n\n```\ngit clone https://github.com/kprompt/kprompt-examples.git\ncd kprompt-examples\nmake walkthrough   # up → break-all → verify → agent-full (~45s)\n```\n\nOr step by step when you want to film each failure:\n\n```\nmake up\nmake break SCENARIO=01-crashloop\nmake verify\nkprompt agent run -n payments \\\n  --emit-initial --analyze --fetch-logs --health --heuristic \\\n  --memory --patterns --autopilot-propose\n```\n\n`--emit-initial`\n\nmatters: a live watch stays quiet until new Pod/Event traffic. Already-broken workloads may not re-emit until the next `BackOff`\n\n. Emitting current state first makes demos and CI deterministic.\n\nSeven scenarios in one namespace, with a healthy baseline so the health score has something Ready to weigh against:\n\n| Scenario | Rough signal |\n|---|---|\n| CrashLoop | Restarting container / BackOff |\n| ImagePull | ImagePullBackOff / ErrImagePull |\n| OOM | OOMKilled / Exit 137 |\n| Stalled rollout | Progress deadline / unavailable replicas |\n| Unbound PVC | Pending Pod waiting on volume |\n| Failing CronJob | Job failures / backoff |\n| Missing Redis hostname | DNS / connection errors to a stub Service name |\n\nExpect:\n\n`web`\n\nstays Ready`--autopilot-propose`\n\n: a propose-only rollback suggestion on the stalled checkout rollout (`Applied`\n\nstays That last point is the product claim: Observe never applies, patches, or deletes by default ([ADR-0013](https://github.com/kprompt/kprompt-architecture/blob/main/decisions/ADR-0013-in-cluster-agent.md)). Autopilot emits PlanResult-shaped proposals ([ADR-0015](https://github.com/kprompt/kprompt-architecture/blob/main/decisions/ADR-0015-autopilot-mode.md)); apply stays gated.\n\n| Surface | Job |\n|---|---|\n| Laptop CLI | Reactive: NL → PlanResult → approve → apply |\n| Observe agent | Always-on: watch → Incident → severity/confidence gate → notify |\n| Autopilot | Opt-in propose-only; apply needs policy + explicit approve |\n\nK8sGPT-style tools are great at on-demand scan → explain. Observe is a different job: continuous watch with threaded alerts from live Events/Pods. Do not expect fleet-scanner or multi-agent-framework parity — we ship one kprompt-native pipeline.\n\n`brew install kprompt/tap/kprompt`\n\nor `curl -fsSL https://kprompt.ai/install | bash`\n\n**Try:** [kprompt.ai](https://kprompt.ai) · [GitHub](https://github.com/kprompt/kprompt) · [kprompt-examples](https://github.com/kprompt/kprompt-examples)\n\nMuhtalip", "url": "https://wpnews.pro/news/break-a-kind-cluster-on-purpose-then-watch-an-observe-agent", "canonical_source": "https://dev.to/muhtalipdede/break-a-kind-cluster-on-purpose-then-watch-an-observe-agent-2ceg", "published_at": "2026-08-16 11:47:42+00:00", "updated_at": "2026-08-16 12:13:22.557153+00:00", "lang": "en", "topics": ["ai-agents", "ai-products", "developer-tools", "mlops"], "entities": ["kprompt", "Observe agent", "Autopilot", "kprompt-examples", "Kubernetes", "Slack", "webhook", "Docker"], "alternates": {"html": "https://wpnews.pro/news/break-a-kind-cluster-on-purpose-then-watch-an-observe-agent", "markdown": "https://wpnews.pro/news/break-a-kind-cluster-on-purpose-then-watch-an-observe-agent.md", "text": "https://wpnews.pro/news/break-a-kind-cluster-on-purpose-then-watch-an-observe-agent.txt", "jsonld": "https://wpnews.pro/news/break-a-kind-cluster-on-purpose-then-watch-an-observe-agent.jsonld"}}