Platform

kubehero-io2stars

Kubernetes cost allocation, logs, continuous profiling and eBPF network cost in one install. OpenCost- and Loki-compatible APIs, rightsizing that applies through guarded CRDs, and read-only agents wit

Cloud Infrastructure Devops

Install Run on PlugMCP

GitHub Website

Run on PlugMCP

1# Platform

README.mdMCP server

Transport: stdio
Status: Local

Power your agent

Add this server

Install from the publisher. Not a PlugMCP endpoint.

Run on PlugMCP

Host a copy with our keys and logs.

No packaged release in the registry yet. Follow the repository README.

RepositoryPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Server facts

Local

Transport: stdio
License: NOASSERTION
Last checked: 2026-10-28

Stars: 2
Tools: —
Uptime: —

Overview

About Platform

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

How do I install Platform?

Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does Platform need an API key?

No required environment variable is listed on this record.

What transport does Platform use?

Platform runs locally over stdio.

When was Platform last checked?

Last checked 2026-10-28. Kubernetes cost allocation, logs, continuous profiling and eBPF network cost in one install. OpenCost- and Loki-compatible APIs, rightsizing that applies through guarded CRDs, and read-only agents with an MCP server.

From the publisher

Documentation

KubeHero

Open-source Kubernetes observability and FinOps in one install — cost allocation, logs, continuous profiling, eBPF network cost, alerting, guarded rightsizing and read-only AI agents for AKS, GKE and EKS. Self-hosted; nothing leaves your cluster.

helm install kubehero oci://ghcr.io/kubehero-io/charts/kubehero \
  -n kubehero-system --create-namespace

One DaemonSet and one control plane replace the stack most teams stitch together:

You'd otherwise run	KubeHero gives you
OpenCost / Kubecost	Cost allocation by any dimension with idle + shared cost, CPU/RAM/GPU/network split, an OpenCost-compatible API, FinOps FOCUS export, chargeback, forecasts
Loki + Promtail	Container log collection, LogQL (filters, parsers, metric queries), live tail, pattern mining, log $ per team — Loki push + query API, OTLP/HTTP logs
Pyroscope + profiling agents	eBPF whole-node CPU profiling with no code changes, pprof scraping, Pyroscope ingest, flamegraphs, diff flamegraphs, $/month per function
Network observability	eBPF flow map with cross-zone and internet egress $, TCP retransmits
VPA / rightsizing tools	Percentile rightsizing from measured usage, applied only through guarded, human-armed `RightsizingPolicy` CRDs — bounded, OOM-aware, reversible
Alertmanager rules sprawl	One alert engine over logs, spend, budget burn, anomalies, network $ and cluster events → Slack, PagerDuty, Opsgenie, Teams, webhooks, Alertmanager
—	Agents: daily briefings (text + voice), "Ask KubeHero" investigations that cite evidence, and an MCP server so Claude can query your fleet — read-only, BYO Anthropic key

It keeps what you already run: Prometheus scrapes it, Grafana reads its logs as a Loki datasource, OpenCost consumers read its allocation API.

Quickstart

# 1. install (embedded Postgres + ClickHouse, auth on)
helm install kubehero oci://ghcr.io/kubehero-io/charts/kubehero -n kubehero-system --create-namespace

# 2. your admin token
kubectl -n kubehero-system get secret kubehero-control-plane \
  -o jsonpath='{.data.admin-token}' | base64 -d; echo

# 3. the dashboard (sign in with the token)
kubectl -n kubehero-system port-forward svc/kubehero-dashboard 3001:3001

No cluster handy? docker compose up --build -d runs the whole stack locally with a synthetic three-cluster fleet streamed through the real ingest APIs, then open http://localhost:3001.

Layout

apps/        dashboard · docs
services/    control-plane · collector · operator · advisor · pricing-engine
cli/         kubehero (CLI + MCP server)
packages/    proto · cost-model · tsconfig
deploy/      helm · terraform
infra/       demo (compose, kind e2e, telemetry generator) · dagger

Architecture, guardrails and honest limits: ARCHITECTURE.md. Docs: https://kubehero.io/docs.

Development

task build              # every Go module + TS workspace
task test               # go test ./... everywhere
task proto              # buf lint + generate (plugins pinned)
task helm:lint
docker compose up --build -d            # full stack + demo fleet
./infra/demo/e2e-kind.sh                # real chart on kind, every signal asserted

See LOCAL_DEV.md and CONTRIBUTING.md.

Licensing

Open source and free to self-host.

Apache 2.0 — CLI, collector, cost-model, proto
BSL 1.1 — control plane, operator, advisor, pricing engine, dashboard (source-available, free to run yourself)
GPL-2.0-only OR Apache-2.0 — the eBPF programs in services/collector/internal/ebpf/bpf/ (the kernel only lets GPL-compatible programs call the helpers they use); the Go loader and the rest of the collector are Apache 2.0. Texts: LICENSE-APACHE-2.0, LICENSE-BSL-1.1, LICENSE-GPL-2.0.

Platform

kubehero-io2stars

Cloud Infrastructure Devops

Install Run on PlugMCP

GitHub Website

Run on PlugMCP

1# Platform

README.mdMCP server

Transport: stdio
Status: Local

Power your agent

Add this server

Install from the publisher. Not a PlugMCP endpoint.

Run on PlugMCP

Host a copy with our keys and logs.

No packaged release in the registry yet. Follow the repository README.

RepositoryPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Server facts

Local

Transport: stdio
License: NOASSERTION
Last checked: 2026-10-28

Stars: 2
Tools: —
Uptime: —

Overview

About Platform

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

How do I install Platform?

Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does Platform need an API key?

No required environment variable is listed on this record.

What transport does Platform use?

Platform runs locally over stdio.

When was Platform last checked?

From the publisher

Documentation

KubeHero

helm install kubehero oci://ghcr.io/kubehero-io/charts/kubehero \
  -n kubehero-system --create-namespace

One DaemonSet and one control plane replace the stack most teams stitch together:

You'd otherwise run	KubeHero gives you
OpenCost / Kubecost	Cost allocation by any dimension with idle + shared cost, CPU/RAM/GPU/network split, an OpenCost-compatible API, FinOps FOCUS export, chargeback, forecasts
Loki + Promtail	Container log collection, LogQL (filters, parsers, metric queries), live tail, pattern mining, log $ per team — Loki push + query API, OTLP/HTTP logs
Pyroscope + profiling agents	eBPF whole-node CPU profiling with no code changes, pprof scraping, Pyroscope ingest, flamegraphs, diff flamegraphs, $/month per function
Network observability	eBPF flow map with cross-zone and internet egress $, TCP retransmits
VPA / rightsizing tools	Percentile rightsizing from measured usage, applied only through guarded, human-armed `RightsizingPolicy` CRDs — bounded, OOM-aware, reversible
Alertmanager rules sprawl	One alert engine over logs, spend, budget burn, anomalies, network $ and cluster events → Slack, PagerDuty, Opsgenie, Teams, webhooks, Alertmanager
—	Agents: daily briefings (text + voice), "Ask KubeHero" investigations that cite evidence, and an MCP server so Claude can query your fleet — read-only, BYO Anthropic key

It keeps what you already run: Prometheus scrapes it, Grafana reads its logs as a Loki datasource, OpenCost consumers read its allocation API.

Quickstart

# 1. install (embedded Postgres + ClickHouse, auth on)
helm install kubehero oci://ghcr.io/kubehero-io/charts/kubehero -n kubehero-system --create-namespace

# 2. your admin token
kubectl -n kubehero-system get secret kubehero-control-plane \
  -o jsonpath='{.data.admin-token}' | base64 -d; echo

# 3. the dashboard (sign in with the token)
kubectl -n kubehero-system port-forward svc/kubehero-dashboard 3001:3001

No cluster handy? docker compose up --build -d runs the whole stack locally with a synthetic three-cluster fleet streamed through the real ingest APIs, then open http://localhost:3001.

Layout

apps/        dashboard · docs
services/    control-plane · collector · operator · advisor · pricing-engine
cli/         kubehero (CLI + MCP server)
packages/    proto · cost-model · tsconfig
deploy/      helm · terraform
infra/       demo (compose, kind e2e, telemetry generator) · dagger

Architecture, guardrails and honest limits: ARCHITECTURE.md. Docs: https://kubehero.io/docs.

Development

task build              # every Go module + TS workspace
task test               # go test ./... everywhere
task proto              # buf lint + generate (plugins pinned)
task helm:lint
docker compose up --build -d            # full stack + demo fleet
./infra/demo/e2e-kind.sh                # real chart on kind, every signal asserted

See LOCAL_DEV.md and CONTRIBUTING.md.

Licensing

Open source and free to self-host.

Apache 2.0 — CLI, collector, cost-model, proto
BSL 1.1 — control plane, operator, advisor, pricing engine, dashboard (source-available, free to run yourself)
GPL-2.0-only OR Apache-2.0 — the eBPF programs in services/collector/internal/ebpf/bpf/ (the kernel only lets GPL-compatible programs call the helpers they use); the Go loader and the rest of the collector are Apache 2.0. Texts: LICENSE-APACHE-2.0, LICENSE-BSL-1.1, LICENSE-GPL-2.0.