@integrity-labs/agt-cli 0.28.981 → 0.28.982

package/dist/bin/agt.js

@@ -40,7 +40,7 @@ import {
   success,
   table,
   warn
-} from "../chunk-AT643QAO.js";
+} from "../chunk-CD42ZALM.js";
 import {
   getProjectDir,
   isSessionResumeDisabled,
@@ -5472,7 +5472,7 @@ import { execFileSync, execSync } from "child_process";
 import { existsSync as existsSync11, realpathSync as realpathSync2 } from "fs";
 import chalk18 from "chalk";
 import ora16 from "ora";
-var cliVersion = true ? "0.28.981" : "dev";
+var cliVersion = true ? "0.28.982" : "dev";
 async function fetchLatestVersion() {
   const host2 = getHost();
   if (!host2) return null;
@@ -6732,7 +6732,7 @@ function handleError(err) {
 }
 
 // src/bin/agt.ts
-var cliVersion2 = true ? "0.28.981" : "dev";
+var cliVersion2 = true ? "0.28.982" : "dev";
 var program = new Command();
 program.name("agt").description("Augmented CLI \u2014 agent provisioning and management").version(cliVersion2).option("--json", "Emit machine-readable JSON output (suppress spinners and colors)").option("--skip-update-check", "Skip the automatic update check on startup");
 program.hook("preAction", async (thisCommand, actionCommand) => {

package/dist/{chunk-AT643QAO.js → chunk-CD42ZALM.js}

@@ -6967,7 +6967,7 @@ function exchangeFailureKind(err) {
 }
 
 // src/lib/api-client.ts
-var agtCliVersion = true ? "0.28.981" : "dev";
+var agtCliVersion = true ? "0.28.982" : "dev";
 var lastConfigHash = null;
 function setConfigHash(hash) {
   lastConfigHash = hash && hash.length > 0 ? hash : null;
@@ -11709,4 +11709,4 @@ export {
   managerInstallSystemUnitCommand,
   managerUninstallSystemUnitCommand
 };
-//# sourceMappingURL=chunk-AT643QAO.js.map
+//# sourceMappingURL=chunk-CD42ZALM.js.map

package/dist/lib/manager-worker.js

@@ -63,7 +63,7 @@ import {
   safeWriteJsonAtomic,
   setConfigHash,
   tripClass
-} from "../chunk-AT643QAO.js";
+} from "../chunk-CD42ZALM.js";
 import {
   getProjectDir as getProjectDir2,
   getReadyTasks,
@@ -6461,6 +6461,16 @@ import { join as join30 } from "path";
 function claudeDeviceStatePath(env = process.env) {
   return join30(env.HOME?.trim() || homedir13(), ".claude.json");
 }
+function probeClaudeDeviceStatePresence(path = claudeDeviceStatePath()) {
+  try {
+    return statSync9(path).isFile() ? { present: true } : { present: false, reason: "not-a-file" };
+  } catch (err) {
+    return {
+      present: false,
+      reason: err?.code === "ENOENT" ? "missing" : "unreadable"
+    };
+  }
+}
 function checkClaudeDeviceState(path = claudeDeviceStatePath()) {
   try {
     if (!statSync9(path).isFile()) return { usable: false, reason: "not-a-file" };
@@ -6489,10 +6499,18 @@ function checkClaudeDeviceState(path = claudeDeviceStatePath()) {
   return { usable: true };
 }
 var DeviceStateRotationGate = class {
-  constructor(check = () => checkClaudeDeviceState()) {
+  constructor(check = () => checkClaudeDeviceState(), probe = () => probeClaudeDeviceStatePresence()) {
     this.check = check;
+    this.probe = probe;
   }
   held = /* @__PURE__ */ new Map();
+  /**
+   * ENG-10730: when the CHEAP probe first saw the file gone with nothing held.
+   * Separate from `held`, which records when each agent's rotation was stopped -
+   * in this state no rotation has been stopped at all. Cleared whenever the file
+   * is seen present again, so a later episode gets its own clock.
+   */
+  unusableSinceMs = null;
   /**
    * Ask before tearing `codeName`'s session down for a planned rotation.
    * Returns `ready: true` when the respawn can start; otherwise the rotation must
@@ -6510,22 +6528,62 @@ var DeviceStateRotationGate = class {
     return { ready: false, reason: verdict.reason, firstHold };
   }
   /**
-   * The value for the heartbeat. Re-checks the file while anything is held, so
-   * the alert closes on the next heartbeat after the operator pairs the host,
-   * not only when some agent's rollover next asks. Reads nothing when nothing is
-   * held, so a healthy host pays no file read per heartbeat.
+   * The value for the heartbeat.
+   *
+   * ENG-10730. This used to open with `if (this.held.size === 0) return null;` -
+   * it returned the affirmative all-clear WITHOUT LOOKING, and the docblock
+   * defended that as "a healthy host pays no file read per heartbeat". The
+   * saving was real and so was the hole: a host whose `~/.claude.json` had been
+   * unlinked, and which had not happened to attempt a rotation since, reported
+   * `null` every 30 seconds while already doomed. Nothing asked until some
+   * agent's rollover did, which on ENG-10724 was the maintenance window - i.e.
+   * the moment the agents were lost. The host was discovered one whole host at a
+   * time, at whatever hour its window fell.
+   *
+   * So the heartbeat now always looks, and pays for it with ONE `statSync` -
+   * `probeClaudeDeviceStatePresence`, no read, no parse. The original perf
+   * objection is answered rather than overridden: what was rejected in PR #6250
+   * was a multi-MB synchronous `JSON.parse` on every healthy host's event loop,
+   * and that is still not happening here.
+   *
+   * THE TWO PATHS ARE NOT SYMMETRIC, AND THAT IS DELIBERATE.
+   * - Something held: the FULL check, exactly as before. A hold must only clear
+   *   on a genuinely usable file, so a present-but-corrupt one cannot read as a
+   *   recovery and release a rotation onto a host that still cannot respawn.
+   * - Nothing held: the CHEAP probe. It can only report that the file has GONE,
+   *   never that one is well - so it can raise this warning and it can clear
+   *   this warning, but it is never allowed to clear a hold.
+   *
+   * `deferred_agents: []` with a reason is therefore a THIRD state, distinct
+   * from both `null` and a hold: the file is unusable and nothing has been
+   * stopped yet. The API must not describe that as a hold - see
+   * `summariseDeviceStateBlock`, which separates the two so the alert does not
+   * claim the manager is holding work it is not.
    */
-  heartbeatField() {
-    if (this.held.size === 0) return null;
-    const verdict = this.check();
-    if (verdict.usable) {
-      this.held.clear();
+  heartbeatField(nowMs = Date.now()) {
+    if (this.held.size > 0) {
+      const verdict = this.check();
+      if (verdict.usable) {
+        this.held.clear();
+        this.unusableSinceMs = null;
+        return null;
+      }
+      return {
+        reason: verdict.reason,
+        deferred_agents: [...this.held.keys()].sort(),
+        blocked_since: new Date(Math.min(...this.held.values())).toISOString()
+      };
+    }
+    const presence = this.probe();
+    if (presence.present) {
+      this.unusableSinceMs = null;
       return null;
     }
+    if (this.unusableSinceMs === null) this.unusableSinceMs = nowMs;
     return {
-      reason: verdict.reason,
-      deferred_agents: [...this.held.keys()].sort(),
-      blocked_since: new Date(Math.min(...this.held.values())).toISOString()
+      reason: presence.reason,
+      deferred_agents: [],
+      blocked_since: new Date(this.unusableSinceMs).toISOString()
     };
   }
   /** For tests and logs. */
@@ -13955,7 +14013,7 @@ var agentRestartTimezoneInputs = /* @__PURE__ */ new Map();
 var lastVersionCheckAt = 0;
 var VERSION_CHECK_INTERVAL_MS = 5 * 60 * 1e3;
 var lastResponsivenessProbeAt = 0;
-var agtCliVersion = true ? "0.28.981" : "dev";
+var agtCliVersion = true ? "0.28.982" : "dev";
 function resolveBrewPath(execFileSync2) {
   try {
     const out = execFileSync2("which", ["brew"], { timeout: 5e3 }).toString().trim();