release: promote Slice 0 foundation

.github/workflows/ci.yml

@@ -0,0 +1,69 @@
+name: ci
+
+on:
+  push:
+    branches: [dev]
+  pull_request:
+    branches: [dev]
+
+permissions:
+  contents: read
+
+concurrency:
+  group: ci-${{ github.ref }}
+  cancel-in-progress: true
+
+env:
+  GO_VERSION: "1.25.x"
+  GOLANGCI_VERSION: "v2.12.2"
+
+jobs:
+  build-test-lint:
+    name: build · vet · gofmt · lint · test · e2e
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@9f698171ed81b15d1823a05fc7211befd50c8ae0 # v6.0.3
+        with:
+          persist-credentials: false
+
+      - uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        with:
+          go-version: ${{ env.GO_VERSION }}
+          check-latest: true
+          cache: true
+          cache-dependency-path: go.sum
+
+      - name: Download and verify modules
+        run: |
+          go mod download
+          go mod verify
+
+      - name: gofmt check
+        run: |
+          diff=$(gofmt -l .)
+          if [ -n "$diff" ]; then
+            echo "::error::gofmt needed on:"
+            echo "$diff"
+            exit 1
+          fi
+
+      - name: go vet
+        run: go vet ./...
+
+      - name: golangci-lint
+        uses: golangci/golangci-lint-action@d583c34f0599d37dbac4a198b9c83201be380893 # v9.3.0
+        with:
+          version: ${{ env.GOLANGCI_VERSION }}
+          args: ./...
+
+      - name: Format and imports check
+        run: golangci-lint fmt --diff ./...
+
+      - name: Build
+        run: go build -trimpath ./...
+
+      - name: Unit tests with race detector
+        run: go test -race -count=1 -coverprofile=coverage.out ./...
+
+      - name: Binary integration smoke test
+        run: go test -race -count=1 -tags=e2e ./tests/e2e

.gitignore

@@ -32,6 +32,7 @@ bin/
 vendor/
 *.test
 *.out
+coverage.*
 # Node / UI
 node_modules/
 dist/

.golangci.yml

@@ -0,0 +1,43 @@
+version: "2"
+
+run:
+  timeout: 5m
+  tests: true
+
+linters:
+  default: none
+  enable:
+    - bodyclose
+    - errcheck
+    - gosec
+    - govet
+    - ineffassign
+    - misspell
+    - revive
+    - staticcheck
+    - unconvert
+    - unused
+  settings:
+    revive:
+      rules:
+        - name: exported
+        - name: package-comments
+        - name: error-strings
+        - name: context-as-argument
+        - name: unreachable-code
+    misspell:
+      locale: US
+  exclusions:
+    generated: lax
+    rules:
+      - path: _test\.go
+        linters: [errcheck, gosec]
+
+formatters:
+  enable:
+    - gofmt
+    - goimports
+  settings:
+    goimports:
+      local-prefixes:
+        - github.com/ArdurAI/sith

COMPETITIVE.md

@@ -0,0 +1,124 @@
+# Sith — Competitive Landscape
+
+**Status:** planning · **Date:** 2026-07-08 · **All facts web-verified July 2026** (sources at end)
+
+This is an honest map, including where the space is **crowded**. The conclusion is *not*
+"nobody does this" — several strong efforts converge on parts of it. The conclusion is that
+**vendor-neutral, OSS, governed *action* federation as an adoptable primitive** — with the
+AI as a client of the governance rather than the product — is a narrow but real gap, and
+that Sith is defensible only if it stays narrow and gets the governance right.
+
+---
+
+## 1. The substrate we adopt (not compete with)
+
+| Project | What it is | Verified status / version | Role for Sith |
+|---|---|---|---|
+| **Open Cluster Management (OCM)** | CNCF framework for multi-cluster management | **CNCF Sandbox** (accepted 2021-11-09); core `ocm` **v1.3.1** (2026-05-19) | The substrate — Sith builds on it |
+| **OCM `cluster-proxy`** | Reverse tunnels managed→hub; hub reaches cluster-local services across isolated VPCs; automates konnectivity | **v0.10.0** (2026-02-02); "Service Proxy" added v0.9.0 | Adopted transport |
+| **OCM `managed-serviceaccount`** | Syncs SAs to spokes, projects scoped tokens back to hub | **v0.10.0** (2026-02-02) | Adopted scoped identity |
+| **Kubernetes SIG `apiserver-network-proxy` (Konnectivity)** | The underlying reverse-tunnel technique | Kubernetes SIG project | The pattern cluster-proxy automates |
+| **Rancher `remotedialer`** | Outbound reverse tunnel (K3s/Rancher scale) | **v0.6.1** | Proves the pattern is commodity |
+
+**Takeaway:** the "outbound-only agent + hub reaches cluster-local services" mechanism is
+**commodity, hardened, CNCF/vendor-maintained plumbing** — building it bespoke would be
+re-implementing security-sensitive infrastructure worse than three existing efforts. So it
+is **out of scope** ([ADR-0001](docs/adr/0001-adopt-ocm-vs-bespoke-tunnel.md)).
+
+## 2. Adjacent but different (clear category lines)
+
+| Project | Category | Verified | Why it is *not* Sith |
+|---|---|---|---|
+| **Karmada** | Multi-cluster **scheduling / workload placement** | CNCF; **v1.18.1** (2026-06-30) | Places workloads; Sith governs *operations*, not placement |
+| **Clusterpedia** | Multi-cluster **resource inventory / search** | CNCF Sandbox; **v0.9.1** (2026-04-16) | Read/search only; no governed action federation |
+| **Headlamp** | Single-/multi-cluster **UI** | Kubernetes-sigs; **v0.43.0** (2026-06-16) | A console, not a governed action control plane |
+| **Devtron** | OSS K8s **platform** (CI/CD + GitOps + obs + security) | **v2.1.1** (2026-03-24) | Batteries-included platform, not a narrow federation-governance layer |
+| **argocd-agent** | Agent-based **Argo CD** multi-cluster | argoproj-labs; **v0.9.0** (2026-06-04) | Federates Argo CD's *own* surface; Sith is tool-agnostic + governed |
+| **kagent** | CNCF K8s-native **agent framework** | **v0.10.0-beta4** (2026-07-06) | A framework to *build* agents; a potential **MCP client** of Sith, not a competitor |
+
+## 3. The genuinely competitive zone (be honest — it is crowded)
+
+The product space *above* the plumbing is occupied and converging on "federate existing
+tools + govern + put AI on top":
+
+- **Komodor** — Autonomous AI SRE. Shipped an **extensible multi-agent architecture**
+  (2026-03-18) letting orgs bring their own tools/services/agents via **MCP or OpenAPI**,
+  orchestrated alongside **50+ specialized agents**, with unified fleet visibility. This is
+  the closest thing to "federate existing cluster-local tools + AI", **shipping and
+  funded**.
+- **SUSE Rancher Prime** — At **KubeCon EU 2026**, positioned as an **agentic AI
+  ecosystem**: the "Liz" assistant became a **crew** of agents (Security/Observability/
+  Platform/SLE/App), with **external MCP server** plug-in so third-party tools become
+  "crew members", plus K3k virtual-cluster GPU multi-tenancy. Hyperscaler-scale, vendor
+  platform.
+- **Devtron** — OSS, multi-cluster, AI-native platform at scale.
+
+**Honest read:** the *visibility + AI-assistant* layer is well-served and competitive. If
+Sith tried to be "an AI SRE that sees your fleet", it would be entering Komodor's and
+Rancher's strongest 2026 investment with none of their funding, data, or vendor weight.
+
+## 4. Where the wedge is defensible
+
+Sith does **not** win on visibility or on "most agents". It is defensible only on a narrow
+axis the incumbents under-serve as an *adoptable, neutral primitive*:
+
+1. **Governed *action* federation, not just visibility.** The incumbents excel at *see* and
+   at *investigate*; safe, policy-gated, **typed-intent fan-out to N clusters** — with wave
+   gates, multi-approver prod, partial-failure/rollback, and **federation-specific
+   abstention** — is under-served as a clean primitive rather than one vendor's feature.
+2. **AI as a *client of the governance*, not the product.** Komodor and Rancher put agents
+   *on top*. Sith inverts it: it is a **governed MCP server** where the org's own agent and
+   *any* external agent (Claude Code, Codex, kagent) inherit the **same** PDP, closed verb
+   vocabulary, scoped identity, and audit. "A governed MCP gateway to your whole fleet" is a
+   platform position, not a chatbot.
+3. **Vendor-neutral, OSS, and deliberately narrow.** Not tied to one vendor's stack (SUSE)
+   or one SaaS (Komodor). Built on a CNCF substrate (OCM), tool-agnostic, and scoped so
+   tightly (no portal, no scheduler, no telemetry store, **no shell ever**) that it can be
+   *correct* where a broad platform can only be *featureful*.
+4. **Correctness of isolation + action safety as the product.** The hard, valuable,
+   defensible thing is getting multi-tenant isolation, signed intents, per-spoke local
+   enforcement, and least-privilege identity **right** — precisely where broad platforms
+   carry more surface and more risk.
+
+## 5. Risks to the thesis (stated plainly)
+
+- **The gap is narrowing.** As OCM's addon ecosystem, argocd-agent, and vendor MCP surfaces
+  mature, the space compresses. Milestone-0 + Phase-1 must validate demand fast.
+- **Governance-of-action is contested at the edges** (AI gateways, vendor MCP governance).
+  Sith's differentiation is the *fan-out policy* + *neutral OSS* + *action-federation*
+  intersection — thinner than a whole product, which is exactly why it must stay narrow.
+- **Execution risk is the isolation problem itself.** The single hardest part (correct
+  multi-tenant isolation + safe action) is the part predecessors got wrong. This plan makes
+  those day-one, structural requirements rather than later hardening.
+
+## 6. One-line positioning
+
+> **Komodor/Rancher put AI *on top* of the fleet. Sith puts *governance* under the fleet —
+> and lets any AI be a governed client of it.**
+
+---
+
+## Sources (verified July 2026)
+
+- OCM — CNCF status: <https://www.cncf.io/projects/open-cluster-management/> · project:
+  <https://open-cluster-management.io/> · core repo: <https://github.com/open-cluster-management-io/ocm>
+- OCM `cluster-proxy` (v0.10.0, reverse tunnels / cross-VPC / konnectivity):
+  <https://github.com/open-cluster-management-io/cluster-proxy> · Service Proxy:
+  <https://open-cluster-management.io/blog/2025/cluster-proxy-now-supports-service-proxy-an-easy-way-to-access-services-in-managed-clusters/>
+- OCM `managed-serviceaccount` (v0.10.0):
+  <https://github.com/open-cluster-management-io/managed-serviceaccount> ·
+  <https://open-cluster-management.io/docs/getting-started/integration/managed-serviceaccount/>
+- Konnectivity / `apiserver-network-proxy`: <https://github.com/kubernetes-sigs/apiserver-network-proxy>
+- Rancher `remotedialer`: <https://github.com/rancher/remotedialer>
+- Karmada (v1.18.1): <https://karmada.io/> · <https://github.com/karmada-io/karmada>
+- Clusterpedia (v0.9.1): <https://github.com/clusterpedia-io/clusterpedia>
+- Headlamp (v0.43.0): <https://github.com/kubernetes-sigs/headlamp>
+- Devtron (v2.1.1): <https://github.com/devtron-labs/devtron>
+- argocd-agent (v0.9.0): <https://github.com/argoproj-labs/argocd-agent>
+- kagent (v0.10.0-beta4): <https://kagent.dev/> · <https://github.com/kagent-dev/kagent>
+- Komodor extensible multi-agent + MCP/OpenAPI (2026-03-18):
+  <https://komodor.com/blog/komodor-introduces-extensible-autonomous-multi-agent-architecture-for-ai-driven-site-reliability-engineering/>
+- SUSE Rancher Prime agentic AI + MCP (KubeCon EU 2026):
+  <https://www.suse.com/c/kubecon-eu-2026-prime-mcp-plug-and-play/> · <https://www.suse.com/c/kubecon-eu-2026-prime-ai-crews/>
+- MCP tool annotations (hints, enforce server-side): <https://blog.modelcontextprotocol.io/posts/2026-03-16-tool-annotations/>
+- MCP Elicitation (2025-06-18): <https://modelcontextprotocol.io/specification/2025-06-18/client/elicitation>

Makefile

@@ -0,0 +1,60 @@
+# Sith — Makefile
+SHELL := /usr/bin/env bash
+
+BINARY   := sith
+PKG      := github.com/ArdurAI/sith
+CMD      := ./cmd/sith
+BIN_DIR  := bin
+GOLANGCI ?= golangci-lint
+
+VERSION ?= $(shell git describe --tags --always --dirty 2>/dev/null || echo dev)
+COMMIT  ?= $(shell git rev-parse --short HEAD 2>/dev/null || echo none)
+DATE    ?= $(shell date -u +%Y-%m-%dT%H:%M:%SZ)
+
+LDFLAGS := -s -w \
+	-X $(PKG)/internal/buildinfo.Version=$(VERSION) \
+	-X $(PKG)/internal/buildinfo.Commit=$(COMMIT) \
+	-X $(PKG)/internal/buildinfo.Date=$(DATE)
+
+.PHONY: all build test e2e lint fmt fmt-check vet tidy clean run ci help
+
+all: build
+
+build: ## Build the sith binary into bin/
+	@mkdir -p $(BIN_DIR)
+	go build -trimpath -ldflags '$(LDFLAGS)' -o $(BIN_DIR)/$(BINARY) $(CMD)
+
+test: ## Run unit tests with the race detector and report coverage
+	go test -race -count=1 -coverprofile=coverage.out ./...
+
+e2e: ## Build and exercise the real binary as a subprocess
+	go test -race -count=1 -tags=e2e ./tests/e2e
+
+lint: ## Run golangci-lint (v2)
+	$(GOLANGCI) run ./...
+
+fmt: ## Format code (gofmt + goimports via golangci-lint v2 formatters)
+	$(GOLANGCI) fmt ./...
+
+fmt-check: ## Fail if formatting/imports would change anything
+	gofmt -l . | tee /dev/stderr | (! read)
+	$(GOLANGCI) fmt --diff ./...
+
+vet: ## Run go vet
+	go vet ./...
+
+tidy: ## Tidy and verify modules
+	go mod tidy
+	go mod verify
+
+clean: ## Remove build and coverage artifacts
+	rm -rf $(BIN_DIR) coverage.out
+
+run: build ## Build then run sith version
+	$(BIN_DIR)/$(BINARY) version
+
+ci: fmt-check vet lint test e2e build ## Run the full CI gate locally
+
+help: ## List targets
+	@grep -E '^[a-zA-Z_-]+:.*?## ' $(MAKEFILE_LIST) | \
+		awk 'BEGIN{FS=":.*?## "}{printf "  \033[36m%-10s\033[0m %s\n",$$1,$$2}'

README.md

@@ -1,13 +1,30 @@
 # Sith
 
-**Status: planning.** No product code yet.
+**Status: Slice 0 foundation.** The local-first CLI walking skeleton is runnable; Kubernetes
+context discovery arrives in Slice 1.
 
-Sith is a governed, multi-tenant control plane for **cross-cluster Kubernetes fleet
-operations** — one place to safely *see* and *act* across many clusters, while deep
-cluster access stays local. It is built on [Open Cluster Management (OCM)](https://open-cluster-management.io/).
+Sith is ArdurAI's single-binary, local-first Kubernetes fleet tool: **k9s for your whole fleet**.
+It is designed to aggregate every kubeconfig context without an account, telemetry, or cluster
+data leaving the machine. The same source-abstract fleet model will later power an optional
+governed hub.
 
-The full charter, architecture, ADRs, threat model, roadmap, and competitive analysis
-land via the initial planning pull request. The owner reviews the plan before any
-implementation begins.
+## Build and run
 
-See the open PR and `docs/` for the plan.
+Sith requires a supported Go 1.25 toolchain.
+
+```bash
+make build
+./bin/sith version
+./bin/sith version --output json
+./bin/sith clusters
+```
+
+Slice 0 intentionally returns a typed empty fleet through the stubbed `fleet.Source` seam. Run the
+full local quality gate with a pinned golangci-lint v2.12.2 on `PATH`:
+
+```bash
+make ci
+```
+
+The architecture, threat model, ADRs, and roadmap live under [`docs/`](docs/). Build-session
+checkpoints are recorded under [`sessions/`](sessions/).

cmd/sith/main.go

@@ -0,0 +1,14 @@
+// SPDX-License-Identifier: Apache-2.0
+
+// Package main is the process entrypoint for the Sith binary.
+package main
+
+import (
+	"os"
+
+	"github.com/ArdurAI/sith/internal/cli"
+)
+
+func main() {
+	os.Exit(cli.Execute())
+}