fix: remove kubectl from follower state helpers

.agents/skills/unidesk-cicd/SKILL.md

@@ -44,7 +44,9 @@ bun scripts/cli.ts cicd branch-follower debug-step --follower web-probe-sentinel
 - 触发或验收 rollout 时必须绑定 lane、source commit、PipelineRun/GitOps revision、runtime ready 和 `/health` 端点验证结果；web-probe/Playwright 结果只能作为单独的 post-deploy 证据。
 - CI/CD 状态、日志和事件查询必须减少 trans/SSH 传输：能在目标 NODE/k8s 内解析、聚合、裁剪的内容，必须在目标侧计算成短 JSON/table 摘要后再回传；禁止为了本地解析而把完整 ConfigMap、大对象、长日志或原始 API payload 透传回来。
 - branch-follower 排障必须优先使用 `debug-step` 做单步调试：`state-read`、`status-read`、`decide`、`state-write` 分别定位状态读取、K8s native status、决策和 ConfigMap patch，不要通过反复推小提交触发整条自动跟随大回环来定位同一问题。
+- CI/CD 排障中再次踩到已经暴露过的运行面坑、工具误用、镜像假设或状态可见性缺口时，必须先把长期规则写入本 skill/reference，再继续单步调试；branch-follower 必须先让相关 `debug-step` 单步全部通过，再做 `run-once`、自动 loop 或小 PR/小提交联调。
 - CI/CD 验证、测试和性能度量必须在目标 NODE/k8s 内执行，尤其是 branch-follower、Tekton/Argo、runtime reuse/env reuse、git mirror 和 runtime-ready 相关改动；不要在 master/local host 跑 test 或用本地验证结果替代目标运行面证据。本机只用于源码阅读、编辑和必要静态语法检查，正式收敛结论必须来自目标 NODE 计算出的短摘要。
+- in-cluster/controller/native helper 不能假设镜像内存在 `kubectl` binary。目标 Pod/Job 内读取或写入 Kubernetes 对象必须走 serviceaccount token + Kubernetes HTTPS API 或已封装的 native helper；`kubectl` 只允许在 operator 侧受控 CLI/trans 边界作为 transport/debug 包装，不得进入正式 controller 状态读写链路。
 - 一旦发现 CI/CD CLI 被误用且可能写入错误状态、产生伪证据或绕过目标运行面，必须立刻先把用法改成更符合直觉的公开入口并更新本 skill/reference，再继续验证或交付；不要只靠口头记忆、隐藏 flag、手动约定或后续小心来避免复发。内部 in-cluster 模式必须只由目标 k8s Job/Pod 调用，操作者从本机只能用公开入口提交目标侧 Job 或读取目标侧摘要。
 - Secret 只通过 YAML sourceRef/targetKey 和受控 CLI 下发；输出只披露 presence/fingerprint。
 - 长命令用异步 job 或短轮询；不要长时间挂住 trans/ssh。

.agents/skills/unidesk-cicd/references/branch-follower.md

@@ -30,6 +30,8 @@ bun scripts/cli.ts cicd branch-follower logs --follower <id>
 
 Do not debug the same state/read/write problem by repeatedly pushing empty or tiny source commits to drive the full automatic follower loop.
 
+When a repeated runtime pitfall or visibility defect is found during branch-follower work, update this reference or the skill entry first, then continue with the narrow debug step. Do not proceed to `run-once`, controller loop observation, automatic follower validation, or source-commit-driven integration until the relevant `state-read`, `status-read`, `decide`, and `state-write` debug steps pass for the affected follower.
+
 ## Source Authority
 
 - Follower decisions must not read host source worktrees, target dev directories, `.worktree/*`, local git state, or direct GitHub branch refs.
@@ -107,6 +109,8 @@ Status and decision inputs are Kubernetes-native:
 
 The branch follower must not parse downstream CLI stdout/stderr, `kubectl` human tables, `argo` text, `tkn` text, or curl output to infer observed sha, target sha, readiness or closeout. `kubectl -o json` may be used inside the controller/Job as a structured Kubernetes API transport only.
 
+In-cluster controller and native helper scripts must not require a `kubectl` binary in the image. Native helpers that read or write ConfigMaps, Jobs, PipelineRuns, Argo Applications, Pods or logs must use the serviceaccount token and Kubernetes HTTPS API directly, or a shared native helper that does the same. A missing `kubectl` binary is a product defect in the helper, not a node problem. Operator-side `kubectl` through the controlled CLI/trans boundary remains acceptable only as a transport/debug wrapper.
+
 The controller automatic loop submits trigger work without a blocking wait; later loops close out via the native state objects above. Failed state must not dedupe a source commit forever: retries may reuse deterministic native objects for the same source commit, and a new compact observation should be able to move the follower back into triggering or closeout.
 
 State ConfigMaps must stay bounded and human-queryable. Store compact summaries, stage refs, conditions, short messages, and drill-down object names; do not store full API payloads or long log dumps. Cleanup is an explicit operator operation for stale/broken state and must not be required for normal convergence.

scripts/native/cicd/patch-configmap-data.mjs

@@ -0,0 +1,50 @@
+import { readFileSync } from "node:fs";
+import https from "node:https";
+
+const namespace = process.env.NAMESPACE || "";
+const configMap = process.env.CONFIGMAP || "";
+const patch = JSON.parse(Buffer.from(process.env.PATCH_B64 || "", "base64").toString("utf8"));
+const host = process.env.KUBERNETES_SERVICE_HOST;
+const port = Number(process.env.KUBERNETES_SERVICE_PORT || "443");
+const token = readFileSync("/var/run/secrets/kubernetes.io/serviceaccount/token", "utf8").trim();
+const ca = readFileSync("/var/run/secrets/kubernetes.io/serviceaccount/ca.crt");
+
+function request(method, path, body, contentType = "application/json") {
+  return new Promise((resolve, reject) => {
+    const headers = { authorization: `Bearer ${token}` };
+    const payload = body === undefined ? null : typeof body === "string" ? body : JSON.stringify(body);
+    if (payload !== null) {
+      headers["content-type"] = contentType;
+      headers["content-length"] = Buffer.byteLength(payload);
+    }
+    const req = https.request({ host, port, path, method, ca, headers }, (res) => {
+      let text = "";
+      res.setEncoding("utf8");
+      res.on("data", (chunk) => { text += chunk; });
+      res.on("end", () => resolve({ status: res.statusCode || 0, text }));
+    });
+    req.on("error", reject);
+    if (payload !== null) req.write(payload);
+    req.end();
+  });
+}
+
+const path = `/api/v1/namespaces/${encodeURIComponent(namespace)}/configmaps/${encodeURIComponent(configMap)}`;
+const before = await request("GET", path);
+if (before.status === 404) {
+  process.stdout.write(JSON.stringify({ ok: true, present: false, patched: false, reason: "state-configmap-not-found", parsedDownstreamCliOutput: false }));
+  process.exit(0);
+}
+if (before.status < 200 || before.status >= 300) throw new Error(before.text || `kube api GET configmap status ${before.status}`);
+const beforeObject = JSON.parse(before.text);
+const result = await request("PATCH", path, patch, "application/merge-patch+json");
+if (result.status < 200 || result.status >= 300) throw new Error(result.text || `kube api PATCH configmap status ${result.status}`);
+const afterObject = JSON.parse(result.text);
+process.stdout.write(JSON.stringify({
+  ok: true,
+  present: true,
+  patched: true,
+  beforeResourceVersion: beforeObject?.metadata?.resourceVersion || null,
+  afterResourceVersion: afterObject?.metadata?.resourceVersion || null,
+  parsedDownstreamCliOutput: false,
+}));

scripts/native/cicd/patch-follower-state.mjs

@@ -1,38 +1,54 @@
-import { execFileSync } from "node:child_process";
+import { readFileSync } from "node:fs";
+import https from "node:https";
 
 const namespace = process.env.NAMESPACE || "";
 const configMap = process.env.CONFIGMAP || "";
 const followerId = process.env.FOLLOWER_ID || "";
 const specRef = process.env.SPEC_REF || "";
 const stateJson = Buffer.from(process.env.STATE_B64 || "", "base64").toString("utf8");
+const host = process.env.KUBERNETES_SERVICE_HOST;
+const port = Number(process.env.KUBERNETES_SERVICE_PORT || "443");
+const token = readFileSync("/var/run/secrets/kubernetes.io/serviceaccount/token", "utf8").trim();
+const ca = readFileSync("/var/run/secrets/kubernetes.io/serviceaccount/ca.crt");
 
-function kubectl(args, input) {
-  return execFileSync("kubectl", ["-n", namespace, ...args], {
-    input,
-    encoding: "utf8",
-    stdio: ["pipe", "pipe", "pipe"],
+function request(method, path, body, contentType = "application/json") {
+  return new Promise((resolve, reject) => {
+    const headers = { authorization: `Bearer ${token}` };
+    const payload = body === undefined ? null : typeof body === "string" ? body : JSON.stringify(body);
+    if (payload !== null) {
+      headers["content-type"] = contentType;
+      headers["content-length"] = Buffer.byteLength(payload);
+    }
+    const req = https.request({ host, port, path, method, ca, headers }, (res) => {
+      let text = "";
+      res.setEncoding("utf8");
+      res.on("data", (chunk) => { text += chunk; });
+      res.on("end", () => resolve({ status: res.statusCode || 0, text }));
+    });
+    req.on("error", reject);
+    if (payload !== null) req.write(payload);
+    req.end();
   });
 }
 
-function readConfigMap() {
-  try {
-    return JSON.parse(kubectl(["get", "configmap", configMap, "-o", "json"]));
-  } catch (error) {
-    const stderr = String(error?.stderr || error?.message || "");
-    if (/not found/i.test(stderr)) return null;
-    throw error;
-  }
+async function readConfigMap() {
+  const result = await request("GET", `/api/v1/namespaces/${encodeURIComponent(namespace)}/configmaps/${encodeURIComponent(configMap)}`);
+  if (result.status === 404) return null;
+  if (result.status < 200 || result.status >= 300) throw new Error(result.text || `kube api GET configmap status ${result.status}`);
+  return JSON.parse(result.text);
 }
 
-function ensureConfigMap() {
-  if (readConfigMap() !== null) return;
+async function ensureConfigMap() {
+  if (await readConfigMap() !== null) return;
   const object = {
     apiVersion: "v1",
     kind: "ConfigMap",
     metadata: { name: configMap, namespace },
     data: { _createdAt: new Date().toISOString(), _specRef: specRef },
   };
-  kubectl(["apply", "-f", "-"], JSON.stringify(object));
+  const result = await request("POST", `/api/v1/namespaces/${encodeURIComponent(namespace)}/configmaps`, object);
+  if (result.status === 409) return;
+  if (result.status < 200 || result.status >= 300) throw new Error(result.text || `kube api POST configmap status ${result.status}`);
 }
 
 function stringOrNull(value) {
@@ -91,8 +107,8 @@ function preserveExistingTiming(state, existing) {
   };
 }
 
-ensureConfigMap();
-const current = readConfigMap();
+await ensureConfigMap();
+const current = await readConfigMap();
 const beforeResourceVersion = stringOrNull(current?.metadata?.resourceVersion);
 const beforeUpdatedAt = stringOrNull(current?.data?._updatedAt);
 const currentText = current?.data?.[followerId];
@@ -106,8 +122,9 @@ const patch = {
     _specRef: specRef,
   },
 };
-kubectl(["patch", "configmap", configMap, "--type", "merge", "-p", JSON.stringify(patch)]);
-const updated = readConfigMap();
+const patchResult = await request("PATCH", `/api/v1/namespaces/${encodeURIComponent(namespace)}/configmaps/${encodeURIComponent(configMap)}`, patch, "application/merge-patch+json");
+if (patchResult.status < 200 || patchResult.status >= 300) throw new Error(patchResult.text || `kube api PATCH configmap status ${patchResult.status}`);
+const updated = await readConfigMap();
 process.stdout.write(JSON.stringify({
   ok: true,
   followerId,

scripts/native/cicd/read-state-summary.mjs

@@ -1,9 +1,14 @@
-import { execFileSync } from "node:child_process";
+import { readFileSync } from "node:fs";
+import https from "node:https";
 
 const namespace = process.env.NAMESPACE || "";
 const configMap = process.env.CONFIGMAP || "";
 const followerIds = parseFollowerIds(process.env.FOLLOWERS_JSON || "[]");
 const maxTimingStages = Number(process.env.MAX_TIMING_STAGES || "24");
+const host = process.env.KUBERNETES_SERVICE_HOST;
+const port = Number(process.env.KUBERNETES_SERVICE_PORT || "443");
+const token = readFileSync("/var/run/secrets/kubernetes.io/serviceaccount/token", "utf8").trim();
+const ca = readFileSync("/var/run/secrets/kubernetes.io/serviceaccount/ca.crt");
 
 function parseFollowerIds(text) {
   try {
@@ -14,18 +19,34 @@ function parseFollowerIds(text) {
   }
 }
 
-function kubectlConfigMap() {
-  try {
-    const stdout = execFileSync("kubectl", ["-n", namespace, "get", "configmap", configMap, "-o", "json"], {
-      encoding: "utf8",
-      maxBuffer: 16 * 1024 * 1024,
-      stdio: ["ignore", "pipe", "pipe"],
+function request(method, path, body, contentType = "application/json") {
+  return new Promise((resolve, reject) => {
+    const headers = { authorization: `Bearer ${token}` };
+    const payload = body === undefined ? null : typeof body === "string" ? body : JSON.stringify(body);
+    if (payload !== null) {
+      headers["content-type"] = contentType;
+      headers["content-length"] = Buffer.byteLength(payload);
+    }
+    const req = https.request({ host, port, path, method, ca, headers }, (res) => {
+      let text = "";
+      res.setEncoding("utf8");
+      res.on("data", (chunk) => { text += chunk; });
+      res.on("end", () => resolve({ status: res.statusCode || 0, text }));
     });
-    return { ok: true, present: true, object: JSON.parse(stdout), error: "" };
+    req.on("error", reject);
+    if (payload !== null) req.write(payload);
+    req.end();
+  });
+}
+
+async function readConfigMap() {
+  try {
+    const result = await request("GET", `/api/v1/namespaces/${encodeURIComponent(namespace)}/configmaps/${encodeURIComponent(configMap)}`);
+    if (result.status === 404) return { ok: true, present: false, object: null, error: result.text };
+    if (result.status < 200 || result.status >= 300) return { ok: false, present: false, object: null, error: result.text || `kube api GET configmap status ${result.status}` };
+    return { ok: true, present: true, object: JSON.parse(result.text), error: "" };
   } catch (error) {
-    const stderr = String(error?.stderr || error?.message || "");
-    if (/not found/i.test(stderr)) return { ok: true, present: false, object: null, error: stderr };
-    return { ok: false, present: false, object: null, error: stderr || "kubectl configmap read failed" };
+    return { ok: false, present: false, object: null, error: error?.message || String(error) };
   }
 }
 
@@ -139,7 +160,7 @@ function numberOrNull(value) {
   return typeof value === "number" && Number.isFinite(value) ? value : null;
 }
 
-const result = kubectlConfigMap();
+const result = await readConfigMap();
 const errors = [];
 const stateByFollower = {};
 const valueBytes = {};

scripts/src/cicd.ts

@@ -1857,8 +1857,9 @@ function mergeFollowerStatus(
 function readK8sState(registry: BranchFollowerRegistry, options: ParsedOptions): K8sStateRead {
   const errors: string[] = [];
   const stateResult = kubeConfigMapFollowerState(registry, options);
-  const deploymentResult = kubeJson(registry, options, `kubectl -n ${shQuote(registry.controller.namespace)} get deploy ${shQuote(registry.controller.deploymentName)} -o json`, 10_000);
-  const leaseResult = kubeJson(registry, options, `kubectl -n ${shQuote(registry.controller.namespace)} get lease ${shQuote(registry.controller.leaseName)} -o json`, 10_000);
+  const namespace = registry.controller.namespace;
+  const deploymentResult = kubeJson(registry, options, `kubectl -n ${shQuote(namespace)} get deploy ${shQuote(registry.controller.deploymentName)} -o json`, 10_000, `/apis/apps/v1/namespaces/${encodeURIComponent(namespace)}/deployments/${encodeURIComponent(registry.controller.deploymentName)}`);
+  const leaseResult = kubeJson(registry, options, `kubectl -n ${shQuote(namespace)} get lease ${shQuote(registry.controller.leaseName)} -o json`, 10_000, `/apis/coordination.k8s.io/v1/namespaces/${encodeURIComponent(namespace)}/leases/${encodeURIComponent(registry.controller.leaseName)}`);
   const podSelector = labelSelector(registry.controller.labels);
   const podsResult = kubePodList(registry, options, podSelector);
   if (!stateResult.ok) errors.push(`state configmap: ${stateResult.error}`);
@@ -1937,7 +1938,20 @@ function stateWriteSummary(followerId: string, result: CommandResult): Record<st
 
 function removeFollowerStateKeys(registry: BranchFollowerRegistry, options: ParsedOptions, ids: string[]): CommandResult {
   const patch = JSON.stringify({ data: Object.fromEntries(ids.map((id) => [id, null])) });
-  const script = [
+  const script = options.inCluster
+    ? [
+        "set -eu",
+        "tmpdir=$(mktemp -d)",
+        "cleanup() { rm -rf \"$tmpdir\"; }",
+        "trap cleanup EXIT INT TERM",
+        nativeCicdScriptLoadShell(["patch-configmap-data.mjs"]),
+        `NAMESPACE=${shQuote(registry.controller.namespace)}`,
+        `CONFIGMAP=${shQuote(registry.controller.stateConfigMapName)}`,
+        `PATCH_B64=${shQuote(Buffer.from(patch, "utf8").toString("base64"))}`,
+        "export NAMESPACE CONFIGMAP PATCH_B64",
+        "node \"$tmpdir/patch-configmap-data.mjs\"",
+      ].join("\n")
+    : [
         "set -eu",
         "tmpdir=$(mktemp -d)",
         "cleanup() { rm -rf \"$tmpdir\"; }",
@@ -1960,8 +1974,18 @@ function removeFollowerStateKeys(registry: BranchFollowerRegistry, options: Pars
   return runKubeScript(registry, options, script, "", 10_000);
 }
 
-function kubeJson(registry: BranchFollowerRegistry, options: ParsedOptions, command: string, timeoutMs: number): { ok: boolean; value: Record<string, unknown> | null; error: string } {
-  const result = runKubeScript(registry, options, `set -eu\n${command}`, "", timeoutMs);
+function kubeJson(registry: BranchFollowerRegistry, options: ParsedOptions, command: string, timeoutMs: number, inClusterPath?: string): { ok: boolean; value: Record<string, unknown> | null; error: string } {
+  const script = options.inCluster && inClusterPath !== undefined
+    ? [
+        "set -eu",
+        "tmpdir=$(mktemp -d)",
+        "cleanup() { rm -rf \"$tmpdir\"; }",
+        "trap cleanup EXIT INT TERM",
+        nativeCicdScriptLoadShell(["kube-get.mjs"]),
+        `node "$tmpdir/kube-get.mjs" ${shQuote(inClusterPath)}`,
+      ].join("\n")
+    : `set -eu\n${command}`;
+  const result = runKubeScript(registry, options, script, "", timeoutMs);
   const value = result.exitCode === 0 ? parseJsonObject(result.stdout) : null;
   return {
     ok: result.exitCode === 0 && value !== null,
@@ -1972,12 +1996,21 @@ function kubeJson(registry: BranchFollowerRegistry, options: ParsedOptions, comm
 
 function kubePodList(registry: BranchFollowerRegistry, options: ParsedOptions, selector: string): { ok: boolean; value: Record<string, unknown> | null; error: string } {
   const command = `kubectl -n ${shQuote(registry.controller.namespace)} get pods -l ${shQuote(selector)} -o name`;
-  const result = runKubeScript(registry, options, `set -eu\n${command}`, "", 10_000);
-  const names = result.stdout
-    .split(/\r?\n/u)
-    .map((line) => line.trim())
-    .filter((line) => line.length > 0)
-    .map((line) => line.replace(/^pod\//u, ""));
+  const script = options.inCluster
+    ? [
+        "set -eu",
+        "tmpdir=$(mktemp -d)",
+        "cleanup() { rm -rf \"$tmpdir\"; }",
+        "trap cleanup EXIT INT TERM",
+        nativeCicdScriptLoadShell(["kube-get.mjs"]),
+        `node "$tmpdir/kube-get.mjs" ${shQuote(`/api/v1/namespaces/${encodeURIComponent(registry.controller.namespace)}/pods?labelSelector=${encodeURIComponent(selector)}`)}`,
+      ].join("\n")
+    : `set -eu\n${command}`;
+  const result = runKubeScript(registry, options, script, "", 10_000);
+  const parsed = options.inCluster && result.exitCode === 0 ? parseJsonObject(result.stdout) : null;
+  const names = options.inCluster
+    ? arrayRecords(parsed?.items).map((item) => stringOrNull(asOptionalRecord(item.metadata)?.name)).filter((name): name is string => name !== null)
+    : result.stdout.split(/\r?\n/u).map((line) => line.trim()).filter((line) => line.length > 0).map((line) => line.replace(/^pod\//u, ""));
   return {
     ok: result.exitCode === 0,
     value: result.exitCode === 0 ? { items: names.map((name) => ({ metadata: { name } })) } : null,