v2.0.47: HTTPS Everywhere — TLS-only control plane, agents/CLI/MCP

Breaking change release. Plaintext HTTP listener removed. The certctl control plane now terminates TLS 1.3 on :8443 via http.Server.ListenAndServeTLS. No CERTCTL_TLS_ENABLED=false escape hatch. No dual-listener mode. One-step cutover per docs/upgrade-to-tls.md. Server - cmd/server/tls.go: certHolder with SIGHUP hot-reload + atomic cert swap, buildServerTLSConfig (TLS 1.3 min, GetCertificate callback), preflightServerTLS validation - cmd/server/main.go: ListenAndServeTLS in place of ListenAndServe, watchSIGHUP wiring, cert/key path config threading - tls_test.go: 418-line regression coverage of reload, preflight, callback behavior, SAN validation Config - CERTCTL_TLS_CERT_PATH / CERTCTL_TLS_KEY_PATH (required) - Plaintext rejection: agents/CLI/MCP pre-flight-fail on http:// URLs with a pointer to docs/upgrade-to-tls.md Agents, CLI, MCP - All three pre-flight-reject http:// URLs with fail-loud diagnostic - CERTCTL_SERVER_CA_BUNDLE_PATH for private-CA trust - CERTCTL_SERVER_TLS_INSECURE_SKIP_VERIFY for dev-only bypass (loud warning on startup) - install-agent.sh emits both vars as commented template lines docker-compose - certctl-tls-init sidecar generates SAN-valid self-signed cert into deploy/test/certs/ on first boot - All demo-stack curls pin against ca.crt with --cacert Helm chart - Three TLS provisioning modes, exactly one required: - server.tls.existingSecret (operator-supplied) - server.tls.certManager.enabled (cert-manager integration) - server.tls.selfSigned.enabled (eval only — not for production) - server-certificate.yaml template for cert-manager mode - helm install without a TLS source fails at template render with a pointer to docs/tls.md CI - .github/workflows/ci.yml Helm Chart Validation step renders the chart in both existingSecret and cert-manager modes, plus an inverse guard-regression test that asserts helm template MUST refuse to render when no TLS source is configured. Previously the single `helm template` invocation hit the certctl.tls.required fail-loud guard and exit-1'd CI. Four invocations now: lint (existingSecret), template (existingSecret), template (cert-manager), template (no args — must fail). Integration tests - deploy/test/integration_test.go stands up the Compose stack over HTTPS, extracts the CA bundle, and exercises every certctl API over https://localhost:8443 - All 34 integration subtests green (per Phase 8 local CI-parity) Documentation - New: docs/tls.md (provisioning patterns, rotation, SIGHUP reload) - New: docs/upgrade-to-tls.md (one-step cutover, no-downgrade warnings, fleet-roll sequencing) - CHANGELOG.md: v2.2.0 "HTTPS Everywhere — The Irony" entry (file heading unchanged; release tag is v2.0.47) - All curls in docs/, examples/, deploy/helm/ guides use https://localhost:8443 --cacert Verification - grep -rn "ListenAndServe[^T]" cmd/ internal/ → 0 hits - grep -rn "\"http://" cmd/ internal/ → 2 benign hits (Caddy admin API default, SSRF doc comment) — zero certctl endpoints - Tasks #197–#206 (Phases 0–8) all closed in the tracker Files: 65 changed, 3489 insertions, 372 deletions (pre-CI-fix).
2026-06-10 02:58:52 +00:00 · 2026-04-20 03:31:05 +00:00
parent 04c7eca615
commit 52248be717
66 changed files with 3518 additions and 375 deletions
@@ -7,6 +7,7 @@ import (
 	"crypto/elliptic"
 	"crypto/rand"
 	"crypto/rsa"
+	"crypto/tls"
 	"crypto/x509"
 	"crypto/x509/pkix"
 	"encoding/json"
@@ -72,7 +73,7 @@ func TestAgent_Heartbeat_Success(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Should not panic
 	agent.sendHeartbeat(context.Background())
@@ -93,7 +94,7 @@ func TestAgent_Heartbeat_ServerError(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Should increment consecutive failures
 	failureBefore := agent.consecutiveFailures
@@ -115,7 +116,7 @@ func TestAgent_Heartbeat_ConnectionError(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Should fail due to connection error
 	agent.sendHeartbeat(context.Background())
@@ -150,7 +151,7 @@ func TestAgent_PollWork_NoWork(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Should not panic
 	agent.pollForWork(context.Background())
@@ -195,7 +196,7 @@ func TestAgent_PollWork_Success(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Should not panic; work items are processed in separate gorines in real usage
 	agent.pollForWork(context.Background())
@@ -285,7 +286,7 @@ func TestParsePEMFile(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Parse the file
 	entries := agent.parsePEMFile(certPath)
@@ -336,7 +337,7 @@ func TestParsePEMFile_MultipleCerts(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	entries := agent.parsePEMFile(certPath)

@@ -362,7 +363,7 @@ func TestParseDERFile(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	entry, err := agent.parseDERFile(derPath)
 	if err != nil {
@@ -397,7 +398,7 @@ func TestParseDERFile_Invalid(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	_, err := agent.parseDERFile(derPath)
 	if err == nil {
@@ -439,7 +440,7 @@ func TestScanDirectory(t *testing.T) {
 		DiscoveryDirs: []string{tmpdir},
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Simulate directory walk manually (as runDiscoveryScan does)
 	var certs []discoveredCertEntry
@@ -474,7 +475,7 @@ func TestCreateTargetConnector_NGINX(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	configJSON := json.RawMessage(`{"cert_path":"/etc/nginx/cert.pem"}`)
 	connector, err := agent.createTargetConnector("NGINX", configJSON)
@@ -496,7 +497,7 @@ func TestCreateTargetConnector_Unsupported(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	_, err := agent.createTargetConnector("UnsupportedType", nil)

@@ -530,7 +531,7 @@ func TestFetchCertificate_Success(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	certPEM, err := agent.fetchCertificate(context.Background(), "mc-001")
 	if err != nil {
@@ -556,7 +557,7 @@ func TestFetchCertificate_NotFound(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	_, err := agent.fetchCertificate(context.Background(), "mc-nonexistent")
 	if err == nil {
@@ -592,7 +593,7 @@ func TestReportJobStatus_Success(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	err := agent.reportJobStatus(context.Background(), "j-001", "Completed", "")
 	if err != nil {
@@ -624,7 +625,7 @@ func TestReportJobStatus_WithError(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	err := agent.reportJobStatus(context.Background(), "j-001", "Failed", "deployment failed")
 	if err != nil {
@@ -658,7 +659,7 @@ func TestMakeRequest_Success(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	resp, err := agent.makeRequest(context.Background(), http.MethodPost, "/test", map[string]string{"key": "value"})
 	if err != nil {
@@ -680,7 +681,7 @@ func TestMakeRequest_InvalidURL(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	_, err := agent.makeRequest(context.Background(), http.MethodGet, "/test", nil)
 	if err == nil {
@@ -765,7 +766,7 @@ func TestNewAgent(t *testing.T) {
 	}

 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	if agent.config != cfg {
 		t.Error("config not set correctly")
@@ -791,7 +792,7 @@ func TestNewAgent_WithLogger(t *testing.T) {
 		Hostname:  "test-host",
 	}

-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	if agent.logger != logger {
 		t.Error("logger not set correctly")
@@ -954,7 +955,7 @@ func TestCreateTargetConnector_AllSupportedTypes(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
@@ -1007,7 +1008,7 @@ func TestCreateTargetConnector_InvalidJSON(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	invalidJSON := json.RawMessage("{invalid json}")

@@ -1031,7 +1032,7 @@ func TestCreateTargetConnector_UnknownType(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	_, err := agent.createTargetConnector("MagicBox", nil)

@@ -1061,7 +1062,7 @@ func TestCreateTargetConnector_EmptyConfig(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	for _, typeName := range tests {
 		t.Run(typeName, func(t *testing.T) {
@@ -1137,7 +1138,7 @@ func TestRunDiscoveryScan_ValidCerts(t *testing.T) {
 		DiscoveryDirs: []string{tmpDir},
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Run discovery scan
 	agent.runDiscoveryScan(context.Background())
@@ -1165,7 +1166,7 @@ func TestRunDiscoveryScan_NoCertificates(t *testing.T) {
 		DiscoveryDirs: []string{tmpDir},
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Run discovery scan - should complete without error even with empty directory
 	agent.runDiscoveryScan(context.Background())
@@ -1222,7 +1223,7 @@ func TestRunDiscoveryScan_MultipleCerts(t *testing.T) {
 		DiscoveryDirs: []string{tmpDir},
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Run discovery scan
 	agent.runDiscoveryScan(context.Background())
@@ -1273,7 +1274,7 @@ func TestRunDiscoveryScan_DERCertificate(t *testing.T) {
 		DiscoveryDirs: []string{tmpDir},
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Run discovery scan
 	agent.runDiscoveryScan(context.Background())
@@ -1331,7 +1332,7 @@ func TestRunDiscoveryScan_Subdirectories(t *testing.T) {
 		DiscoveryDirs: []string{tmpDir},
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Run discovery scan - should recursively find certs in subdirs
 	agent.runDiscoveryScan(context.Background())
@@ -1369,7 +1370,7 @@ func TestRunDiscoveryScan_ServerError(t *testing.T) {
 		DiscoveryDirs: []string{tmpDir},
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	// Should handle server error gracefully without panicking
 	agent.runDiscoveryScan(context.Background())
@@ -1396,7 +1397,7 @@ func TestDiscoveredCertEntry_ValidFields(t *testing.T) {
 		Hostname:  "test-host",
 	}
 	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
-	agent := NewAgent(cfg, logger)
+	agent, _ := NewAgent(cfg, logger)

 	entries := agent.parsePEMFile(certPath)

@@ -1447,3 +1448,244 @@ func TestDiscoveredCertEntry_ValidFields(t *testing.T) {
 		t.Error("PEMData should not be empty")
 	}
 }
+
+// ---------------------------------------------------------------------------
+// HTTPS-Everywhere milestone (v2.2, §3.2 / §7) — Phase 5 client-side tests.
+//
+// These tests pin the agent's pre-flight HTTPS-scheme guard and the TLS
+// configuration surface (CA bundle loading + TLS 1.3 round-trip) so that
+// regressions surface at unit-test time, not at the first heartbeat of a
+// production rollout. Matches the same contract asserted by the sibling
+// binaries cmd/cli/main_test.go and cmd/mcp-server/main_test.go — the three
+// must stay in lock-step because all three are HTTPS-only clients of the
+// same control plane.
+// ---------------------------------------------------------------------------
+
+// TestValidateHTTPSScheme pins the pre-flight URL-scheme guard that the
+// HTTPS-Everywhere milestone requires on the agent binary startup path. The
+// agent's diagnostic is distinct from the CLI/MCP variants because it names
+// CERTCTL_SERVER_URL (the only input channel — no --server flag on the
+// agent). Every case here mirrors the dispatch arms in cmd/agent/main.go:
+// validateHTTPSScheme; drifting the error-message substrings is what this
+// test is here to catch.
+func TestValidateHTTPSScheme(t *testing.T) {
+	tests := []struct {
+		name       string
+		serverURL  string
+		wantErr    bool
+		wantErrSub string
+	}{
+		{
+			name:      "https URL passes",
+			serverURL: "https://certctl-server:8443",
+			wantErr:   false,
+		},
+		{
+			name:      "https URL with path passes",
+			serverURL: "https://certctl.example.com/api/v1",
+			wantErr:   false,
+		},
+		{
+			name:      "uppercase HTTPS scheme passes (url.Parse lowercases)",
+			serverURL: "HTTPS://certctl-server:8443",
+			wantErr:   false,
+		},
+		{
+			name:       "empty URL rejected names CERTCTL_SERVER_URL",
+			serverURL:  "",
+			wantErr:    true,
+			wantErrSub: "CERTCTL_SERVER_URL is empty",
+		},
+		{
+			name:       "plaintext http rejected",
+			serverURL:  "http://certctl-server:8443",
+			wantErr:    true,
+			wantErrSub: "plaintext http://",
+		},
+		{
+			name:       "bare host missing scheme falls through to unsupported",
+			serverURL:  "localhost:8443",
+			wantErr:    true,
+			// url.Parse treats "localhost:8443" as scheme=localhost,
+			// opaque=8443 — exercises the default arm (unsupported scheme)
+			// rather than the empty-scheme arm. Both are fail-closed, which
+			// is what we care about.
+			wantErrSub: "unsupported scheme",
+		},
+		{
+			name:       "path-only URL rejected",
+			serverURL:  "//certctl-server:8443",
+			wantErr:    true,
+			wantErrSub: "missing a scheme",
+		},
+		{
+			name:       "unsupported scheme rejected",
+			serverURL:  "ftp://certctl-server:8443",
+			wantErr:    true,
+			wantErrSub: "unsupported scheme",
+		},
+		{
+			name:       "ws scheme rejected",
+			serverURL:  "ws://certctl-server:8443",
+			wantErr:    true,
+			wantErrSub: "unsupported scheme",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := validateHTTPSScheme(tt.serverURL)
+			if (err != nil) != tt.wantErr {
+				t.Fatalf("validateHTTPSScheme(%q) err=%v wantErr=%v", tt.serverURL, err, tt.wantErr)
+			}
+			if tt.wantErr && tt.wantErrSub != "" && !strings.Contains(err.Error(), tt.wantErrSub) {
+				t.Errorf("validateHTTPSScheme(%q) err=%q must contain %q so operators see the right diagnostic",
+					tt.serverURL, err.Error(), tt.wantErrSub)
+			}
+		})
+	}
+}
+
+// writeTestCABundle PEM-encodes a cert's DER bytes and writes the result to a
+// tmp file inside dir. Used by CA-bundle tests so each case owns a distinct
+// file path (matters for the "missing file" case which must point at a path
+// that provably does not exist). Returns the path.
+func writeTestCABundle(t *testing.T, dir string, certDER []byte, filename string) string {
+	t.Helper()
+	pemBytes := pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: certDER})
+	path := filepath.Join(dir, filename)
+	if err := os.WriteFile(path, pemBytes, 0644); err != nil {
+		t.Fatalf("writing CA bundle %q: %v", path, err)
+	}
+	return path
+}
+
+// TestNewAgent_CABundle_Success confirms that a well-formed PEM bundle gets
+// parsed into an x509.CertPool and wired onto the agent's HTTP client
+// transport. This is the happy path the docs/tls.md "Private CA signed
+// server cert" section depends on.
+func TestNewAgent_CABundle_Success(t *testing.T) {
+	cert, err := generateTestCertWithCN("test.certctl.local")
+	if err != nil {
+		t.Fatalf("generateTestCertWithCN: %v", err)
+	}
+	bundlePath := writeTestCABundle(t, t.TempDir(), cert.Raw, "ca-bundle.pem")
+
+	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
+	agent, err := NewAgent(&AgentConfig{
+		ServerURL:    "https://certctl-server:8443",
+		APIKey:       "test-key",
+		AgentID:      "a-test",
+		Hostname:     "test-host",
+		CABundlePath: bundlePath,
+	}, logger)
+	if err != nil {
+		t.Fatalf("NewAgent with valid CA bundle err=%v want nil", err)
+	}
+
+	transport, ok := agent.client.Transport.(*http.Transport)
+	if !ok {
+		t.Fatalf("agent.client.Transport is %T; want *http.Transport", agent.client.Transport)
+	}
+	if transport.TLSClientConfig == nil {
+		t.Fatal("TLSClientConfig is nil; HTTPS-everywhere milestone requires a non-nil TLS config")
+	}
+	if transport.TLSClientConfig.MinVersion != tls.VersionTLS13 {
+		t.Errorf("MinVersion=%x want TLS 1.3 (%x) per §2.3 of the milestone spec",
+			transport.TLSClientConfig.MinVersion, tls.VersionTLS13)
+	}
+	if transport.TLSClientConfig.RootCAs == nil {
+		t.Error("RootCAs is nil; the configured CA bundle was silently dropped")
+	}
+}
+
+// TestNewAgent_CABundle_MissingFile pins the fail-loud behavior when the
+// operator points CERTCTL_SERVER_CA_BUNDLE_PATH at a path that does not
+// exist. Falling back to system roots here would mask a misconfiguration as
+// a much harder-to-debug TLS handshake failure downstream.
+func TestNewAgent_CABundle_MissingFile(t *testing.T) {
+	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
+	missingPath := filepath.Join(t.TempDir(), "does-not-exist.pem")
+	_, err := NewAgent(&AgentConfig{
+		ServerURL:    "https://certctl-server:8443",
+		APIKey:       "test-key",
+		AgentID:      "a-test",
+		Hostname:     "test-host",
+		CABundlePath: missingPath,
+	}, logger)
+	if err == nil {
+		t.Fatal("NewAgent err=nil for missing CA bundle path; must fail loud at startup")
+	}
+	if !strings.Contains(err.Error(), "reading CA bundle") {
+		t.Errorf("err=%q must contain \"reading CA bundle\" so operators can trace the cause", err.Error())
+	}
+}
+
+// TestNewAgent_CABundle_EmptyPEM covers the "file exists but contains no
+// valid certs" case (garbage, wrong-format, stripped PEM). AppendCertsFromPEM
+// returns false in this case; NewAgent must translate that into a fail-loud
+// startup error rather than quietly carry on with an empty pool.
+func TestNewAgent_CABundle_EmptyPEM(t *testing.T) {
+	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
+	bundlePath := filepath.Join(t.TempDir(), "empty.pem")
+	if err := os.WriteFile(bundlePath, []byte("not a pem-encoded certificate, just garbage\n"), 0644); err != nil {
+		t.Fatalf("writing garbage bundle: %v", err)
+	}
+	_, err := NewAgent(&AgentConfig{
+		ServerURL:    "https://certctl-server:8443",
+		APIKey:       "test-key",
+		AgentID:      "a-test",
+		Hostname:     "test-host",
+		CABundlePath: bundlePath,
+	}, logger)
+	if err == nil {
+		t.Fatal("NewAgent err=nil for empty-PEM CA bundle; must fail loud at startup")
+	}
+	if !strings.Contains(err.Error(), "no valid PEM-encoded certificates") {
+		t.Errorf("err=%q must contain \"no valid PEM-encoded certificates\" so operators see why the bundle was rejected", err.Error())
+	}
+}
+
+// TestNewAgent_TLSRoundTrip is the end-to-end integration-style check: spin
+// up an httptest.NewTLSServer (which presents a self-signed cert over TLS
+// 1.3), feed that cert into the agent as a CA bundle, and confirm the agent
+// successfully completes a heartbeat round-trip over HTTPS. This proves that
+// (a) the CA pool is actually being consulted during verification and (b)
+// the TLS 1.3 MinVersion doesn't break against httptest's default
+// negotiation. Equivalent to the "TLS handshake succeeds against a
+// self-signed control plane" integration gate, but runs in-process with no
+// Docker dependency.
+func TestNewAgent_TLSRoundTrip(t *testing.T) {
+	var heartbeatHit int
+	server := httptest.NewTLSServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path == "/api/v1/agents/a-tls-test/heartbeat" && r.Method == http.MethodPost {
+			heartbeatHit++
+			w.WriteHeader(http.StatusOK)
+			return
+		}
+		w.WriteHeader(http.StatusNotFound)
+	}))
+	defer server.Close()
+
+	// server.Certificate() returns the *x509.Certificate httptest presents;
+	// PEM-encode its DER bytes so NewAgent's AppendCertsFromPEM can ingest it.
+	bundlePath := writeTestCABundle(t, t.TempDir(), server.Certificate().Raw, "httptest-ca.pem")
+
+	logger := slog.New(slog.NewTextHandler(io.Discard, nil))
+	agent, err := NewAgent(&AgentConfig{
+		ServerURL:    server.URL,
+		APIKey:       "test-key",
+		AgentID:      "a-tls-test",
+		Hostname:     "tls-test-host",
+		CABundlePath: bundlePath,
+	}, logger)
+	if err != nil {
+		t.Fatalf("NewAgent with httptest CA bundle err=%v want nil", err)
+	}
+
+	agent.sendHeartbeat(context.Background())
+
+	if heartbeatHit != 1 {
+		t.Fatalf("heartbeat handler hit %d times; want 1 — the TLS round-trip must actually complete", heartbeatHit)
+	}
+}
@@ -8,6 +8,7 @@ import (
 	"crypto/rand"
 	"crypto/rsa"
 	"crypto/sha256"
+	"crypto/tls"
 	"crypto/x509"
 	"crypto/x509/pkix"
 	"encoding/json"
@@ -19,6 +20,7 @@ import (
 	"log/slog"
 	"net"
 	"net/http"
+	"net/url"
 	"os"
 	"os/signal"
 	"path/filepath"
@@ -46,13 +48,15 @@ import (

 // AgentConfig represents the agent-side configuration.
 type AgentConfig struct {
-	ServerURL     string   // Control plane server URL (e.g., http://localhost:8443)
-	APIKey        string   // Agent API key for authentication
-	AgentName     string   // Agent name for identification
-	AgentID       string   // Agent ID for API calls (set after registration or from env)
-	Hostname      string   // Server hostname
-	KeyDir        string   // Directory for storing private keys (default: /var/lib/certctl/keys)
-	DiscoveryDirs []string // Directories to scan for certificates (comma-separated via env)
+	ServerURL          string   // Control plane server URL (e.g., https://localhost:8443) — must be https:// scheme
+	APIKey             string   // Agent API key for authentication
+	AgentName          string   // Agent name for identification
+	AgentID            string   // Agent ID for API calls (set after registration or from env)
+	Hostname           string   // Server hostname
+	KeyDir             string   // Directory for storing private keys (default: /var/lib/certctl/keys)
+	DiscoveryDirs      []string // Directories to scan for certificates (comma-separated via env)
+	CABundlePath       string   // Optional path to a PEM-encoded CA bundle that signed the server's cert (empty = system roots)
+	InsecureSkipVerify bool     // Dev-only: skip TLS certificate verification. Never enable in production. See docs/tls.md.
 }

 // ErrAgentRetired is the sentinel returned by [Agent.Run] when the control
@@ -113,16 +117,57 @@ type JobItem struct {
 }

 // NewAgent creates a new agent instance.
-func NewAgent(cfg *AgentConfig, logger *slog.Logger) *Agent {
+//
+// The returned HTTP client enforces HTTPS-only control-plane access per the
+// HTTPS-Everywhere milestone (see docs/tls.md). TLS 1.3 is required; the
+// optional CABundlePath loads a PEM bundle into RootCAs so the agent can
+// trust internal / self-signed server certs without touching system trust
+// stores. InsecureSkipVerify is a dev-only escape hatch — callers must log a
+// loud warning when it's set; never enable in production (see §2.4 of the
+// milestone spec and docs/upgrade-to-tls.md).
+//
+// Returns an error if CABundlePath is set but unreadable or malformed — fail
+// loud at startup rather than silently fall back to system roots, which would
+// turn a misconfigured bundle path into a cryptic "x509: certificate signed
+// by unknown authority" on the first heartbeat.
+func NewAgent(cfg *AgentConfig, logger *slog.Logger) (*Agent, error) {
+	tlsConfig := &tls.Config{
+		MinVersion:         tls.VersionTLS13,
+		InsecureSkipVerify: cfg.InsecureSkipVerify, //nolint:gosec // opt-in dev escape hatch, documented in docs/tls.md
+	}
+	if cfg.CABundlePath != "" {
+		pemBytes, err := os.ReadFile(cfg.CABundlePath)
+		if err != nil {
+			return nil, fmt.Errorf("reading CA bundle at %q: %w", cfg.CABundlePath, err)
+		}
+		pool := x509.NewCertPool()
+		if !pool.AppendCertsFromPEM(pemBytes) {
+			return nil, fmt.Errorf("CA bundle at %q contains no valid PEM-encoded certificates", cfg.CABundlePath)
+		}
+		tlsConfig.RootCAs = pool
+	}
+
+	httpClient := &http.Client{
+		Timeout: 30 * time.Second,
+		Transport: &http.Transport{
+			TLSClientConfig:       tlsConfig,
+			ForceAttemptHTTP2:     true,
+			MaxIdleConns:          10,
+			IdleConnTimeout:       90 * time.Second,
+			TLSHandshakeTimeout:   10 * time.Second,
+			ExpectContinueTimeout: 1 * time.Second,
+		},
+	}
+
 	return &Agent{
 		config:            cfg,
 		logger:            logger,
-		client:            &http.Client{Timeout: 30 * time.Second},
+		client:            httpClient,
 		heartbeatInterval: 60 * time.Second,
 		pollInterval:      30 * time.Second,
 		discoveryInterval: 6 * time.Hour, // scan for certs every 6 hours
 		retiredSignal:     make(chan struct{}),
-	}
+	}, nil
 }

 // markRetired records that the control plane has declared this agent retired
@@ -1118,12 +1163,14 @@ func certKeyInfo(cert *x509.Certificate) (string, int) {

 func main() {
 	// Parse command-line flags (with env var fallbacks for Docker deployment)
-	serverURL := flag.String("server", getEnvDefault("CERTCTL_SERVER_URL", "http://localhost:8443"), "Control plane server URL")
+	serverURL := flag.String("server", getEnvDefault("CERTCTL_SERVER_URL", "https://localhost:8443"), "Control plane server URL (must be https://)")
 	apiKey := flag.String("api-key", getEnvDefault("CERTCTL_API_KEY", ""), "Agent API key")
 	agentName := flag.String("name", getEnvDefault("CERTCTL_AGENT_NAME", "certctl-agent"), "Agent name")
 	agentID := flag.String("agent-id", getEnvDefault("CERTCTL_AGENT_ID", ""), "Agent ID (from registration)")
 	keyDir := flag.String("key-dir", getEnvDefault("CERTCTL_KEY_DIR", "/var/lib/certctl/keys"), "Directory for storing private keys")
 	discoveryDirsStr := flag.String("discovery-dirs", getEnvDefault("CERTCTL_DISCOVERY_DIRS", ""), "Comma-separated directories to scan for certificates")
+	caBundlePath := flag.String("ca-bundle", getEnvDefault("CERTCTL_SERVER_CA_BUNDLE_PATH", ""), "Path to a PEM-encoded CA bundle that signed the server's TLS cert (optional; falls back to system roots)")
+	insecureSkipVerify := flag.Bool("insecure-skip-verify", getEnvBoolDefault("CERTCTL_SERVER_TLS_INSECURE_SKIP_VERIFY", false), "Dev-only: skip TLS certificate verification. Never enable in production. See docs/tls.md.")
 	flag.Parse()

 	if *apiKey == "" {
@@ -1137,6 +1184,18 @@ func main() {
 		os.Exit(1)
 	}

+	// Pre-flight URL-scheme validation — reject plaintext http:// before any
+	// network call. The HTTPS-Everywhere milestone (§2.4, §7) mandates that
+	// mis-configured agents fail loudly at startup with a diagnostic pointing
+	// at the upgrade guide, rather than producing a TCP-refused or
+	// TLS-handshake-error that obscures the actual cause.
+	if err := validateHTTPSScheme(*serverURL); err != nil {
+		fmt.Fprintf(os.Stderr, "Error: %v\n", err)
+		fmt.Fprintf(os.Stderr, "\nThe certctl control plane is HTTPS-only as of v2.2.\n")
+		fmt.Fprintf(os.Stderr, "See docs/upgrade-to-tls.md for the cutover walkthrough.\n")
+		os.Exit(1)
+	}
+
 	// Set up structured logging
 	logLevel := slog.LevelInfo
 	if getEnvDefault("CERTCTL_LOG_LEVEL", "info") == "debug" {
@@ -1165,17 +1224,27 @@ func main() {

 	// Create agent configuration
 	agentCfg := &AgentConfig{
-		ServerURL:     *serverURL,
-		APIKey:        *apiKey,
-		AgentName:     *agentName,
-		AgentID:       *agentID,
-		Hostname:      hostname,
-		KeyDir:        *keyDir,
-		DiscoveryDirs: discoveryDirs,
+		ServerURL:          *serverURL,
+		APIKey:             *apiKey,
+		AgentName:          *agentName,
+		AgentID:            *agentID,
+		Hostname:           hostname,
+		KeyDir:             *keyDir,
+		DiscoveryDirs:      discoveryDirs,
+		CABundlePath:       *caBundlePath,
+		InsecureSkipVerify: *insecureSkipVerify,
+	}
+
+	if agentCfg.InsecureSkipVerify {
+		logger.Warn("TLS certificate verification is disabled (CERTCTL_SERVER_TLS_INSECURE_SKIP_VERIFY=true) — never enable this in production")
 	}

 	// Create and start agent
-	agent := NewAgent(agentCfg, logger)
+	agent, err := NewAgent(agentCfg, logger)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error: failed to initialize agent: %v\n", err)
+		os.Exit(1)
+	}

 	// Create context with cancellation for graceful shutdown
 	ctx, cancel := context.WithCancel(context.Background())
@@ -1233,3 +1302,49 @@ func getEnvDefault(key, defaultValue string) string {
 	}
 	return defaultValue
 }
+
+// getEnvBoolDefault parses an environment variable as a boolean. Accepts "1",
+// "t", "true", "T", "TRUE", "True" as true; anything else (including empty)
+// returns the provided default. Kept permissive on purpose so operators can
+// flip the dev-only TLS skip-verify toggle with any common truthy spelling
+// without having to remember exactly what we parse.
+func getEnvBoolDefault(key string, defaultValue bool) bool {
+	raw := os.Getenv(key)
+	if raw == "" {
+		return defaultValue
+	}
+	switch strings.ToLower(strings.TrimSpace(raw)) {
+	case "1", "t", "true", "yes", "on":
+		return true
+	case "0", "f", "false", "no", "off":
+		return false
+	default:
+		return defaultValue
+	}
+}
+
+// validateHTTPSScheme enforces the HTTPS-Everywhere milestone's §7 acceptance
+// criterion: "Agent with CERTCTL_SERVER_URL=http://... fails at startup with
+// a fail-loud diagnostic pointing at docs/upgrade-to-tls.md. Not TCP-refused,
+// not TLS-handshake-error — a pre-flight config validation failure before any
+// network call." Returns a descriptive error; the caller prints the upgrade
+// guide pointer and exits non-zero.
+func validateHTTPSScheme(serverURL string) error {
+	if serverURL == "" {
+		return fmt.Errorf("CERTCTL_SERVER_URL is empty — set it to an https:// URL (e.g., https://certctl-server:8443)")
+	}
+	u, err := url.Parse(serverURL)
+	if err != nil {
+		return fmt.Errorf("CERTCTL_SERVER_URL %q is not a valid URL: %w", serverURL, err)
+	}
+	switch strings.ToLower(u.Scheme) {
+	case "https":
+		return nil
+	case "http":
+		return fmt.Errorf("CERTCTL_SERVER_URL %q uses plaintext http:// — the certctl control plane is HTTPS-only", serverURL)
+	case "":
+		return fmt.Errorf("CERTCTL_SERVER_URL %q is missing a scheme — expected https://", serverURL)
+	default:
+		return fmt.Errorf("CERTCTL_SERVER_URL %q uses unsupported scheme %q — expected https://", serverURL, u.Scheme)
+	}
+}
@@ -228,7 +228,7 @@ func TestReportVerificationResult_Success(t *testing.T) {
 		ServerURL: server.URL,
 		APIKey:    "test-api-key",
 	}
-	agent := NewAgent(cfg, nil)
+	agent, _ := NewAgent(cfg, nil)

 	result := &VerificationResult{
 		ExpectedFingerprint: "abc123",
@@ -244,7 +244,7 @@ func TestReportVerificationResult_Success(t *testing.T) {
 }

 func TestReportVerificationResult_MissingFields(t *testing.T) {
-	agent := NewAgent(&AgentConfig{}, nil)
+	agent, _ := NewAgent(&AgentConfig{}, nil)

 	result := &VerificationResult{
 		Verified:   true,
@@ -343,7 +343,7 @@ func TestReportVerificationResult_ServerError(t *testing.T) {
 		ServerURL: server.URL,
 		APIKey:    "test-api-key",
 	}
-	agent := NewAgent(cfg, nil)
+	agent, _ := NewAgent(cfg, nil)

 	result := &VerificationResult{
 		ExpectedFingerprint: "abc123",