@bastani/atomic 0.9.28-alpha.4 → 0.9.28-alpha.5

package/dist/builtin/intercom/index.bundle.mjs

@@ -1406,13 +1406,43 @@ var init_client = __esm(() => {
   };
 });
 
+// dist/builtin/intercom/broker/socket-liveness.ts
+import net2 from "net";
+function isSocketAnswering(socketPath, timeoutMs = SOCKET_PROBE_TIMEOUT_MS) {
+  return new Promise((resolve) => {
+    const socket = net2.connect(socketPath);
+    const finish = (isConnected) => {
+      clearTimeout(timeout);
+      socket.off("connect", onConnect);
+      socket.off("error", onError);
+      resolve(isConnected);
+    };
+    const onConnect = () => {
+      socket.on("error", () => {});
+      socket.end();
+      finish(true);
+    };
+    const onError = () => {
+      socket.destroy();
+      finish(false);
+    };
+    socket.on("connect", onConnect);
+    socket.on("error", onError);
+    const timeout = setTimeout(() => {
+      socket.destroy();
+      finish(false);
+    }, timeoutMs);
+  });
+}
+var SOCKET_PROBE_TIMEOUT_MS = 1000;
+var init_socket_liveness = () => {};
+
 // dist/builtin/intercom/broker/spawn.ts
 import { spawn } from "child_process";
-import { chmodSync, closeSync, existsSync, mkdirSync, openSync, readFileSync, readSync, statSync, unlinkSync, writeFileSync } from "fs";
+import { appendFileSync, chmodSync, closeSync, existsSync, mkdirSync, openSync, readFileSync, readSync, statSync, unlinkSync, writeFileSync } from "fs";
 import { join as join2, dirname } from "path";
 import { fileURLToPath } from "url";
 import { createRequire } from "module";
-import net2 from "net";
 import { createChildProcessEnvironment, isBunBinary } from "@bastani/atomic";
 function isIntercomPackageRoot(dir) {
   return existsSync(join2(dir, "package.json")) && existsSync(join2(dir, "broker"));
@@ -1537,7 +1567,15 @@ function getBrokerSpawnOptions(extensionDir = EXTENSION_DIR, stderr = "ignore")
 }
 function resetBrokerLog(logPath = BROKER_LOG) {
   ensurePrivateDirectory(dirname(logPath));
-  closeSync(openSync(logPath, "w"));
+  try {
+    closeSync(openSync(logPath, "w"));
+    return true;
+  } catch (error) {
+    const code = error.code;
+    if (process.platform === "win32" && (code === "EBUSY" || code === "EPERM"))
+      return false;
+    throw error;
+  }
 }
 function readBrokerLogTail(logPath = BROKER_LOG, maxBytes = BROKER_LOG_TAIL_BYTES) {
   let fd;
@@ -1571,16 +1609,17 @@ ${tail}`;
 function toError2(error) {
   return error instanceof Error ? error : new Error(String(error));
 }
-async function spawnBrokerIfNeeded(brokerCommand, brokerArgs) {
+async function spawnBrokerIfNeeded(brokerCommand, brokerArgs, readyTimeoutMs = BROKER_READY_TIMEOUT_MS) {
   ensurePrivateDirectory(INTERCOM_DIR);
   if (await isBrokerRunning()) {
     return;
   }
-  const ownsLock = acquireSpawnLock();
-  if (!ownsLock) {
-    await waitForBroker();
+  const lockOwner = acquireSpawnLock();
+  if (lockOwner === undefined) {
+    await waitForBroker(readyTimeoutMs);
     return;
   }
+  let keepLock = false;
   try {
     if (await isBrokerRunning()) {
       return;
@@ -1590,7 +1629,10 @@ async function spawnBrokerIfNeeded(brokerCommand, brokerArgs) {
     if (launch.kind === "windows-launcher") {
       writeWindowsHiddenLauncher(launch.launcherCommandLine, launch.launcherPath);
     }
-    resetBrokerLog();
+    if (!resetBrokerLog()) {
+      await waitForBroker(readyTimeoutMs);
+      return;
+    }
     const logFd = launch.kind === "direct" ? openSync(BROKER_LOG, "a") : undefined;
     let child;
     try {
@@ -1620,21 +1662,28 @@ ${describeBrokerLog()}`, { cause: error }));
 ${describeBrokerLog()}`));
           return;
         }
-        reject(new Error(`Intercom broker exited before startup with code ${code ?? "unknown"}
-${describeBrokerLog()}`));
+        const exitedBeforeStartup = () => new Error(`Intercom broker exited before startup with code ${code ?? "unknown"}
+${describeBrokerLog()}`);
+        if (code === 0) {
+          checkSocketConnectable().then((answering) => answering ? resolve() : reject(exitedBeforeStartup()));
+          return;
+        }
+        reject(exitedBeforeStartup());
       };
       child.once("error", onError);
       child.once("exit", onExit);
-      waitForBroker().then(() => {
+      waitForBroker(readyTimeoutMs).then(() => {
         cleanup();
         resolve();
       }, (error) => {
         cleanup();
+        keepLock = launch.kind === "direct" && child.exitCode === null && child.signalCode === null && handSpawnLockToBroker(child.pid, lockOwner);
         reject(toError2(error));
       });
     });
   } finally {
-    releaseSpawnLock();
+    if (!keepLock)
+      releaseSpawnLock(lockOwner);
   }
 }
 async function isBrokerRunning() {
@@ -1654,39 +1703,35 @@ async function isBrokerRunning() {
   }
 }
 function checkSocketConnectable() {
-  return new Promise((resolve) => {
-    const socket = net2.connect(BROKER_SOCKET2);
-    const finish = (isConnected) => {
-      clearTimeout(timeout);
-      socket.off("connect", onConnect);
-      socket.off("error", onError);
-      resolve(isConnected);
-    };
-    const onConnect = () => {
-      socket.on("error", () => {});
-      socket.end();
-      finish(true);
-    };
-    const onError = () => {
-      socket.destroy();
-      finish(false);
-    };
-    socket.on("connect", onConnect);
-    socket.on("error", onError);
-    const timeout = setTimeout(() => {
-      socket.destroy();
-      finish(false);
-    }, 1000);
-  });
+  return isSocketAnswering(BROKER_SOCKET2);
+}
+function handSpawnLockToBroker(pid, owner) {
+  if (pid === undefined || !ownsSpawnLock(owner))
+    return false;
+  try {
+    appendFileSync(BROKER_SPAWN_LOCK, `${pid}
+`);
+    return true;
+  } catch {
+    return false;
+  }
+}
+function ownsSpawnLock(owner, lockPath = BROKER_SPAWN_LOCK) {
+  try {
+    return readFileSync(lockPath, "utf-8").startsWith(owner);
+  } catch {
+    return false;
+  }
 }
 function acquireSpawnLock() {
   const maxRetries = 5;
   for (let attempt = 0;attempt < maxRetries; attempt++) {
-    try {
-      writeFileSync(BROKER_SPAWN_LOCK, `${process.pid}
+    const owner = `${process.pid}
 ${Date.now()}
-`, { flag: "wx" });
-      return true;
+`;
+    try {
+      writeFileSync(BROKER_SPAWN_LOCK, owner, { flag: "wx" });
+      return owner;
     } catch (error) {
       if (!(error instanceof Error) || error.code !== "EEXIST") {
         throw error;
@@ -1697,39 +1742,46 @@ ${Date.now()}
         } catch {}
         continue;
       }
-      return false;
+      return;
     }
   }
-  return false;
+  return;
 }
-function isSpawnLockStale() {
-  if (!existsSync(BROKER_SPAWN_LOCK)) {
+function isProcessAlive(pid) {
+  try {
+    process.kill(pid, 0);
+    return true;
+  } catch (error) {
+    return error.code === "EPERM";
+  }
+}
+function isSpawnLockStale(lockPath = BROKER_SPAWN_LOCK, nowMs = Date.now()) {
+  if (!existsSync(lockPath)) {
     return false;
   }
   try {
-    const [pidLine = "", createdAtLine = "0"] = readFileSync(BROKER_SPAWN_LOCK, "utf-8").trim().split(`
+    const [spawnerLine = "", createdAtLine = "", brokerLine = ""] = readFileSync(lockPath, "utf-8").trim().split(`
 `);
-    const pid = Number.parseInt(pidLine, 10);
+    const spawnerPid = Number.parseInt(spawnerLine, 10);
     const createdAt = Number.parseInt(createdAtLine, 10);
-    const ageMs = Date.now() - createdAt;
-    if (Number.isFinite(pid)) {
-      try {
-        process.kill(pid, 0);
-      } catch {
-        return true;
-      }
-    }
-    return !Number.isFinite(createdAt) || ageMs > 1e4;
+    const brokerPid = Number.parseInt(brokerLine, 10);
+    if (!Number.isFinite(spawnerPid) || !Number.isFinite(createdAt))
+      return true;
+    if (nowMs - createdAt > BROKER_SPAWN_LOCK_MAX_AGE_MS)
+      return true;
+    return !isProcessAlive(Number.isFinite(brokerPid) ? brokerPid : spawnerPid);
   } catch {
     return true;
   }
 }
-function releaseSpawnLock() {
+function releaseSpawnLock(owner, lockPath = BROKER_SPAWN_LOCK) {
+  if (!ownsSpawnLock(owner, lockPath))
+    return;
   try {
-    unlinkSync(BROKER_SPAWN_LOCK);
+    unlinkSync(lockPath);
   } catch {}
 }
-async function waitForBroker(timeoutMs = 5000) {
+async function waitForBroker(timeoutMs = BROKER_READY_TIMEOUT_MS) {
   const start = Date.now();
   let interval = BROKER_POLL_INITIAL_INTERVAL_MS;
   while (Date.now() - start < timeoutMs) {
@@ -1742,8 +1794,9 @@ async function waitForBroker(timeoutMs = 5000) {
   throw new Error(`Broker failed to start within timeout
 ${describeBrokerLog()}`);
 }
-var INTERCOM_DIR, EXTENSION_DIR, BROKER_SOCKET2, BROKER_PID, BROKER_SPAWN_LOCK, BROKER_LOG, BROKER_LOG_TAIL_BYTES, BROKER_POLL_INITIAL_INTERVAL_MS = 10, BROKER_POLL_MAX_INTERVAL_MS = 100, INTERNAL_INTERCOM_BROKER_ARG = "--atomic-internal-intercom-broker";
+var INTERCOM_DIR, EXTENSION_DIR, BROKER_SOCKET2, BROKER_PID, BROKER_SPAWN_LOCK, BROKER_LOG, BROKER_LOG_TAIL_BYTES, BROKER_POLL_INITIAL_INTERVAL_MS = 10, BROKER_POLL_MAX_INTERVAL_MS = 100, BROKER_READY_TIMEOUT_MS = 5000, BROKER_SPAWN_LOCK_MAX_AGE_MS = 60000, INTERNAL_INTERCOM_BROKER_ARG = "--atomic-internal-intercom-broker";
 var init_spawn = __esm(() => {
+  init_socket_liveness();
   init_paths();
   INTERCOM_DIR = getIntercomDirPath();
   EXTENSION_DIR = resolveIntercomPackageRoot(import.meta.url);
@@ -6393,7 +6446,7 @@ __export(exports_index_heavy, {
   default: () => piIntercomExtension
 });
 import { APP_NAME, sessionScopedExtensionState } from "@bastani/atomic";
-import { appendFileSync } from "node:fs";
+import { appendFileSync as appendFileSync2 } from "node:fs";
 function isPendingStageUndeliverableEvent(value) {
   if (typeof value !== "object" || value === null)
     return false;
@@ -7352,7 +7405,7 @@ var init_index_heavy = __esm(() => {
     process.env.ATOMIC_INTERCOM_HEAVY_IMPORTED = "1";
   }
   if (process.env.ATOMIC_TEST_LAZY_IMPORT_SENTINEL_FILE) {
-    appendFileSync(process.env.ATOMIC_TEST_LAZY_IMPORT_SENTINEL_FILE, `intercom
+    appendFileSync2(process.env.ATOMIC_TEST_LAZY_IMPORT_SENTINEL_FILE, `intercom
 `);
   }
   INTERCOM_SESSION_ID_ENV = `${APP_NAME.toUpperCase()}_INTERCOM_SESSION_ID`;

package/dist/builtin/intercom/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/intercom",
-  "version": "0.9.28-alpha.4",
+  "version": "0.9.28-alpha.5",
   "private": true,
   "description": "Atomic extension providing a private coordination channel between parent and child agent sessions. Fork of: https://github.com/nicobailon/pi-intercom",
   "contributors": [

package/dist/builtin/mcp/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/mcp",
-  "version": "0.9.28-alpha.4",
+  "version": "0.9.28-alpha.5",
   "private": true,
   "description": "Atomic's pi-compatible MCP integration for server connections, OAuth, tools, resources, and terminal management.",
   "contributors": [

package/dist/builtin/subagents/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/subagents",
-  "version": "0.9.28-alpha.4",
+  "version": "0.9.28-alpha.5",
   "private": true,
   "description": "Atomic extension for delegating tasks to subagents with parallel execution. Fork of: https://github.com/nicobailon/pi-subagents",
   "contributors": [

package/dist/builtin/web-access/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/web-access",
-  "version": "0.9.28-alpha.4",
+  "version": "0.9.28-alpha.5",
   "private": true,
   "description": "Atomic extension for web search, URL fetching, GitHub repo cloning, PDF/video extraction. Fork of: https://github.com/nicobailon/pi-web-access",
   "contributors": [

package/dist/builtin/workflows/CHANGELOG.md

@@ -6,6 +6,12 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 
 ## [Unreleased]
 
+## [0.9.28-alpha.5] - 2026-10-08
+
+### Fixed
+
+- Fixed managed PostgreSQL health checks failing a workflow stage on a healthy but busy host: an unanswered check now retries with a short backoff for up to 15 seconds, the connection limit is 3 seconds instead of 1, and the "did not answer a health check in time" diagnostic clears once PostgreSQL answers again ([#3491](https://github.com/bastani-inc/atomic/issues/3491))
+
 ## [0.9.28-alpha.4] - 2026-10-07
 
 ### Added

package/dist/builtin/workflows/builtin/{chunk-aqv8ny38.js → chunk-9ywkdsa9.js}

@@ -37,7 +37,7 @@ import {
   parseFailureDiagnostics,
   summarizeReviewConvergence,
   reverify_consolidated_batch
-} from "./chunk-p3xmajmj.js";
+} from "./chunk-v4scsq3y.js";
 import {
   VERIFICATION_SCALE
 } from "./chunk-n1910xc4.js";

package/dist/builtin/workflows/builtin/{chunk-p6jagvyf.js → chunk-bnz4knhy.js}

@@ -1,7 +1,7 @@
 import {
   getDurableBackend2,
   getAvailableDurableBackend2
-} from "./chunk-p3xmajmj.js";
+} from "./chunk-v4scsq3y.js";
 export {
   getAvailableDurableBackend2 as getAvailableDurableBackend,
   getDurableBackend2 as getDurableBackend

package/dist/builtin/workflows/builtin/{chunk-br4ac9a3.js → chunk-dx5sjaxe.js}

@@ -39,7 +39,7 @@ import {
   reviewerFailureText,
   summarizeReviewConvergence,
   reverify_consolidated_batch
-} from "./chunk-p3xmajmj.js";
+} from "./chunk-v4scsq3y.js";
 import {
   fold_usage
 } from "./chunk-tgt0s5e5.js";

package/dist/builtin/workflows/builtin/{chunk-p3xmajmj.js → chunk-v4scsq3y.js}

@@ -6542,6 +6542,9 @@ function resolveEmbeddedPostgresTarget(host) {
 // dist/builtin/workflows/src/durable/dbos-postgres-health.ts
 var HEALTH_INTERVAL_MS = 5000;
 var RECOVERY_ATTEMPTS = 3;
+var MONITORING_PROBE_RETRY_WINDOW_MS = 15000;
+var MONITORING_PROBE_RETRY_INITIAL_MS = 250;
+var MONITORING_PROBE_RETRY_MAX_MS = 2000;
 function isMonitoringConnectionTimeout(error) {
   if (!(error instanceof Error))
     return false;
@@ -6569,6 +6572,7 @@ class PostgresHealth {
   attempts = 0;
   nextRecoveryAt = 0;
   failure;
+  failureUnanswered = false;
   healthySinceFailure = false;
   revision = 0;
   listeners = new Set;
@@ -6581,10 +6585,11 @@ class PostgresHealth {
   get endpoint() {
     return this.operations.url?.();
   }
-  retainFailure(error) {
+  retainFailure(error, unanswered = false) {
     const message = error instanceof Error ? error.message : String(error);
     const safe = redactedDatabaseMessage(message, this.endpoint);
     this.failure = new Error(safe);
+    this.failureUnanswered = unanswered;
     this.healthySinceFailure = false;
   }
   dependencyFailure(message) {
@@ -6612,7 +6617,7 @@ class PostgresHealth {
         await this.operations.validate?.(client);
       }
     } catch (error) {
-      this.retainFailure(error);
+      this.retainFailure(error, isQueryReadTimeout(error));
       if (!isQueryReadTimeout(error)) {
         this.invalidate();
         if (this.endpoint === undefined)
@@ -6650,19 +6655,39 @@ class PostgresHealth {
   }
   async probe() {
     const revision = this.revision;
+    const now = this.operations.now ?? Date.now;
+    const wait = this.operations.wait ?? ((ms) => new Promise((resolve) => setTimeout(resolve, ms)));
     let identity;
     try {
       identity = await this.operations.probe();
     } catch (error) {
       if (!(isMonitoringConnectionFailure(error) || isQueryReadTimeout(error)) || this.stopped)
         throw error;
-      try {
-        identity = await this.operations.probe();
-      } catch (retryError) {
-        if (isMonitoringConnectionTimeout(retryError) || !isMonitoringConnectionFailure(retryError))
-          throw retryError;
-        this.retainFailure(retryError);
-        return;
+      const deadline = now() + MONITORING_PROBE_RETRY_WINDOW_MS;
+      let delay = MONITORING_PROBE_RETRY_INITIAL_MS;
+      let failure = error;
+      for (;; ) {
+        if (isMonitoringConnectionTimeout(failure) || isQueryReadTimeout(failure)) {
+          if (now() >= deadline)
+            throw failure;
+          await wait(delay);
+          delay = Math.min(delay * 2, MONITORING_PROBE_RETRY_MAX_MS);
+          if (this.stopped)
+            throw failure;
+        }
+        try {
+          identity = await this.operations.probe();
+          break;
+        } catch (retryError) {
+          if (isMonitoringConnectionTimeout(retryError) || isQueryReadTimeout(retryError)) {
+            failure = retryError;
+            continue;
+          }
+          if (!isMonitoringConnectionFailure(retryError))
+            throw retryError;
+          this.retainFailure(retryError);
+          return;
+        }
       }
     }
     if (revision !== this.revision)
@@ -6677,6 +6702,8 @@ class PostgresHealth {
     this.attempts = 0;
     this.nextRecoveryAt = 0;
     this.healthySinceFailure = true;
+    if (this.failureUnanswered)
+      this.failure = undefined;
     return identity.url;
   }
   async refresh() {
@@ -6686,7 +6713,7 @@ class PostgresHealth {
         return ready;
     } catch (error) {
       if (isQueryReadTimeout(error) || isMonitoringConnectionTimeout(error)) {
-        this.retainFailure(new Error("Managed PostgreSQL did not answer a health check in time."));
+        this.retainFailure(new Error("Managed PostgreSQL did not answer a health check in time."), true);
         throw this.dependencyFailure("Managed PostgreSQL did not answer a health check in time.");
       }
       this.retainFailure(error);
@@ -6954,6 +6981,7 @@ var POSTGRES_IDENTITY_SQL = `SELECT current_setting('data_directory') AS data_di
 	system_identifier::text, current_setting('server_version') AS server_version FROM pg_control_system()`;
 var POSTGRES_TIMEZONE_SQL = "SELECT set_config('timezone_abbreviations', 'Default', true)";
 var POSTGRES_HEALTH_QUERY_TIMEOUT_MS = 3000;
+var POSTGRES_HEALTH_CONNECT_TIMEOUT_MS = 3000;
 async function probePostgresTimezoneData(port) {
   const client = new Client({
     host: "127.0.0.1",
@@ -6962,7 +6990,7 @@ async function probePostgresTimezoneData(port) {
     password: "atomic",
     database: "postgres",
     ssl: false,
-    connectionTimeoutMillis: 1000,
+    connectionTimeoutMillis: POSTGRES_HEALTH_CONNECT_TIMEOUT_MS,
     query_timeout: POSTGRES_HEALTH_QUERY_TIMEOUT_MS,
     statement_timeout: POSTGRES_HEALTH_QUERY_TIMEOUT_MS
   });
@@ -6982,7 +7010,7 @@ async function probePostgresIdentity(port) {
     password: "atomic",
     database: "postgres",
     ssl: false,
-    connectionTimeoutMillis: 1000,
+    connectionTimeoutMillis: POSTGRES_HEALTH_CONNECT_TIMEOUT_MS,
     query_timeout: POSTGRES_HEALTH_QUERY_TIMEOUT_MS,
     statement_timeout: POSTGRES_HEALTH_QUERY_TIMEOUT_MS
   });
@@ -9894,7 +9922,7 @@ function createPendingStageDeliveryStoreMethods(context) {
         const run = context.findRun(input.runId);
         if (run === undefined)
           return;
-        const { getAvailableDurableBackend } = await import("./chunk-p6jagvyf.js");
+        const { getAvailableDurableBackend } = await import("./chunk-bnz4knhy.js");
         const availableBackend = getAvailableDurableBackend();
         const backend = availableBackend === undefined ? undefined : durableBackendForRun(availableBackend, context.state.runs, input.runId);
         const receipt = await backend?.readSettledPendingStageMessage?.(input.runId, input.message.id);

package/dist/builtin/workflows/builtin/goal.js

@@ -1,7 +1,7 @@
 import {
   goal_default
-} from "./chunk-aqv8ny38.js";
-import"./chunk-p3xmajmj.js";
+} from "./chunk-9ywkdsa9.js";
+import"./chunk-v4scsq3y.js";
 import"./chunk-2377r57g.js";
 import"./chunk-z0d03ktm.js";
 import"./chunk-n1910xc4.js";

package/dist/builtin/workflows/builtin/index.js

@@ -16,8 +16,8 @@ import {
 } from "./chunk-kb7s3gx2.js";
 import {
   goal_default
-} from "./chunk-aqv8ny38.js";
-import"./chunk-p3xmajmj.js";
+} from "./chunk-9ywkdsa9.js";
+import"./chunk-v4scsq3y.js";
 import"./chunk-z0d03ktm.js";
 import {
   open_claude_design_default
@@ -27,7 +27,7 @@ import {
 } from "./chunk-94f8zfz1.js";
 import {
   ralph_default
-} from "./chunk-br4ac9a3.js";
+} from "./chunk-dx5sjaxe.js";
 import {
   tournament_default
 } from "./chunk-wf876n16.js";

package/dist/builtin/workflows/builtin/ralph.js

@@ -1,7 +1,7 @@
 import {
   ralph_default
-} from "./chunk-br4ac9a3.js";
-import"./chunk-p3xmajmj.js";
+} from "./chunk-dx5sjaxe.js";
+import"./chunk-v4scsq3y.js";
 import"./chunk-2377r57g.js";
 import"./chunk-z0d03ktm.js";
 import"./chunk-n1910xc4.js";

package/dist/builtin/workflows/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/workflows",
-  "version": "0.9.28-alpha.4",
+  "version": "0.9.28-alpha.5",
   "private": true,
   "description": "Atomic extension for multi-stage workflow authoring and execution.",
   "contributors": [