fix(server): wake chat milestones when questions resolve or retire

Co-Authored-By: Paperclip <noreply@paperclip.ing>

doc/plans/2026-10-24-recurring-server-sleep-stack.md

@@ -1472,3 +1472,19 @@ coverage. All 74 selected worker/adapter/policy cases and server production type
 pass. Remaining publication, interaction, configuration, question and monitor
 producer coverage still gates periodic-projector removal. Original inventory,
 remote gates/findings and final master reconciliation remain open.
+
+## Slice 28 draft and publication/interaction coverage
+
+[Draft #17606](https://github.com/paperclipai/paperclip/pull/17606) is stacked on
+#17592. It adds the lifecycle projector worker and covered producer signals;
+production registration and polling removal remain intentionally unfinished until
+the remaining producer audit is complete. Native progress stays a separate slice.
+
+The latest local checkpoint covers real publication success/failure/operator
+resolution and question answer/cancel/withdraw/skip/terminal-expiry paths,
+including historical answers that enqueue no continuation. All 160 milestone and
+interaction-service cases pass on fresh isolated PostgreSQL databases. Four
+existing chat acceptance cases, publication policy cases, and server production
+typecheck passed at their recorded checkpoints. See the detailed milestone audit
+for retained regression/setup failures, controlled ports and remaining coverage.
+The initial draft CI was running; no final green-check or readiness claim is made.

doc/plans/2026-10-25-chat-run-milestone-sleep-audit.md

@@ -721,3 +721,50 @@ called after an interaction mutation cannot replace intent before that earlier
 mutation. Production registration and removal of the combined periodic owner
 remain pending. Publish this slice as a draft for CI feedback, not as a completed
 sleep migration or merge-ready PR.
+
+## Interaction resolution checkpoint
+
+Question answers, question cancellation, withdrawal, skipping and terminal-task
+expiry now signal milestone work before their relevant mutations. Withdrawal,
+skipping and expiry restrict the signal to ask_user_questions and
+request_confirmation, the two kinds used by lifecycle presentation arbitration.
+Existing actor checks, linked tool/secret resolution, pending-state CAS, caller
+hooks and transaction ownership remain in place. No service wrapper or new table
+was introduced. Dismissal alone does not change the arbitration predicate and
+adds no notification.
+
+Historical answers are a distinct case: an ordinary deferred question on a done
+task remains answerable, but deliberately enqueues no question delivery. That
+path cannot rely on a later delivery claim to wake the milestone projector.
+The actual answer transaction now owns its notification without reopening work.
+
+- `/tmp/sleep28-interaction-resolution-before.log`: answer, skip and historical
+  answer all leave an eligible failure notice asleep before the fix.
+- `/tmp/sleep28-interaction-transitions-before.log`: six real producer variants
+  fail the missing-notice assertion: answer, skip, cancel, withdraw, terminal
+  expiry and historical answer. Cancel/withdraw also exposed a fixture cleanup
+  error because they create durable chat actions referencing the conversation.
+- `/tmp/sleep28-interaction-transitions-tests.log`: notice assertions pass after
+  the fix; cancel/withdraw still fail the missing chat-action cleanup. Cleanup
+  now deletes this fixture's actions before its conversations. The old
+  `paperclip_sleep28_worker_final_20261025_test` database retains partial failed
+  fixtures and must not be reused for global worker quiet assertions.
+- `/tmp/sleep28-interaction-final-tests.log`: all 160 cases pass across the full
+  milestone and existing interaction-service suites, on fresh separate databases
+  `paperclip_sleep28_interaction_worker_20261025_test` and
+  `paperclip_sleep28_interaction_service_20261025_test`. The former is the new
+  self-cleaning milestone test database; the latter is exclusive to the service
+  suite's broad cleanup.
+- `/tmp/sleep28-interaction-typecheck.log`: production server typecheck passes;
+  test files are excluded. `git diff --check` passes.
+
+Draft PR #17606 targets #17592. Its first head `f398314ecd406d2e0eebbfb81a5d9dd7edc7c437`
+started CI run 40198190088 and review run 40198190000. At the last inspection,
+24 checks succeeded and 27 remained in progress, with no reported failures.
+That is progress evidence, not final acceptance of this subsequent checkpoint.
+
+Remaining producer work includes confirmation accept/reject and invalidation,
+interaction creation/source changes, provider-specific receipts, conversation
+bindings, configuration and alternate recovery writers. Production registration,
+removal of only the lifecycle polling lane, complete latest-head checks and final
+stack/master reconciliation remain required. No merge or full-goal completion.

server/src/__tests__/chat-run-milestone-work.test.ts

@@ -1,7 +1,7 @@
 import { randomUUID } from "node:crypto";
 import { eq } from "drizzle-orm";
 import { afterAll, afterEach, beforeAll, beforeEach, describe, expect, it, vi } from "vitest";
-import { activityLog, agents, agentWakeupRequests, authUsers, caseHandoffs, cases, companyMemberships, chatConversations, chatEndpoints, chatMessageLinks, chatPublications, companies, completionContracts, createDb, nativeRunFinalizations, nativeRunResults, statusDecisions, statusDecisionEffects, workAssessments, subscribeDatabaseWork,
+import { activityLog, agents, agentWakeupRequests, authUsers, caseHandoffs, cases, companyMemberships, chatActions, chatConversations, chatEndpoints, chatMessageLinks, chatPublications, companies, completionContracts, createDb, nativeRunFinalizations, nativeRunResults, statusDecisions, statusDecisionEffects, workAssessments, subscribeDatabaseWork,
   heartbeatRunEvents, heartbeatRuns, issueComments, issueQuestionResponseDeliveries, issueThreadInteractions, issues, toolApplications, toolConnections } from "@paperclipai/db";
 import { getEmbeddedPostgresTestSupport, startEmbeddedPostgresTestDatabase } from "./helpers/embedded-postgres.js";
 import { resolveExternalTestDatabaseUrl } from "./helpers/external-test-database.js";
@@ -76,6 +76,7 @@ const support = external ? { supported: true } : await getEmbeddedPostgresTestSu
       await db.delete(activityLog).where(eq(activityLog.companyId, companyId));
       await db.delete(caseHandoffs).where(eq(caseHandoffs.companyId, companyId));
       await db.delete(cases).where(eq(cases.companyId, companyId));
+      await db.delete(chatActions).where(eq(chatActions.companyId, companyId));
       await db.delete(chatMessageLinks).where(eq(chatMessageLinks.companyId, companyId));
       await db.delete(chatPublications).where(eq(chatPublications.companyId, companyId));
       await db.delete(chatConversations).where(eq(chatConversations.companyId, companyId));
@@ -632,6 +633,44 @@ const support = external ? { supported: true } : await getEmbeddedPostgresTestSu
     expect((await publications()).map(row => row.idempotencyKey)).toContain(`run:${targetId}:queued:${f.endpointId}`);
     await quiet();
   });
+  it.each(["answer", "skip", "cancel", "withdraw", "expire", "historical answer"] as const)("rechecks an unsent question after %s", async action => {
+    const source = await run(f, { status: "failed", errorCode: "test_run_failed" });
+    const interactions = issueThreadInteractionService(db);
+    const issue = { id: f.issueId, companyId: f.companyId, status: "in_progress" };
+    const interaction = await interactions.create(issue, {
+      kind: "ask_user_questions", continuationPolicy: "wake_assignee", sourceRunId: source.id,
+      title: "Choose", payload: { version: 1, title: "Choose", questions: [{
+        id: "choice", prompt: "Continue?", selectionMode: "single", required: true,
+        options: [{ id: "yes", label: "Yes" }],
+      }] },
+    }, { agentId: f.agentId, runId: source.id });
+    if (action === "historical answer") {
+      await interactions.dismissInteraction(issue, interaction.id, { userId: f.userId });
+      await db.update(issues).set({ status: "done" }).where(eq(issues.id, f.issueId));
+    }
+    await start(); await flush();
+    expect((await publications()).some(row => row.idempotencyKey.startsWith(`run:${source.id}:`))).toBe(false);
+    await quiet();
+    if (action === "skip") {
+      await interactions.skipInteraction(issue, interaction.id, { reason: "No longer needed" }, { userId: f.userId });
+    } else if (action === "cancel") {
+      await interactions.cancelQuestions(issue, interaction.id, { reason: "No longer needed" }, { userId: f.userId });
+    } else if (action === "withdraw") {
+      await interactions.withdrawInteraction(issue, interaction.id, { reason: "No longer needed" }, { userId: f.userId });
+    } else if (action === "expire") {
+      await db.update(issues).set({ status: "cancelled" }).where(eq(issues.id, f.issueId));
+      await interactions.expirePendingInteractionsForTerminalIssue({ ...issue, status: "cancelled" }, { userId: f.userId });
+    } else {
+      await interactions.answerQuestions(issue, interaction.id,
+        { answers: [{ questionId: "choice", optionIds: ["yes"] }] }, { userId: f.userId });
+    }
+    await flush();
+    expect((await publications()).map(row => row.idempotencyKey)).toContain(`run:${source.id}:failed:${f.endpointId}`);
+    if (action === "historical answer") {
+      expect(await db.select().from(issueQuestionResponseDeliveries).where(eq(issueQuestionResponseDeliveries.interactionId, interaction.id))).toEqual([]);
+    }
+    await quiet();
+  });
   it("wakes a question continuation after its in-flight prompt receipt commits", async () => {
     await db.update(chatEndpoints).set({ provider: "discord" }).where(eq(chatEndpoints.id, f.endpointId));
     const source = await run(f, { status: "failed", contextSnapshot: { ...runValues().contextSnapshot, source: "chat:discord" } });

server/src/services/issue-thread-interactions.ts

@@ -1,3 +1,4 @@
+import { notifyChatRunMilestones } from "./chat-run-milestone-notifications.js";
 import { notifyToolReviewChange } from "./tool-review-work-notifications.js";
 import { normalizeEscapedLineBreaks } from "@paperclipai/shared/validators/text";
 import { notifyDeliveryWork, DELIVERY_QUEUES } from "./delivery-work-notifications.js";
@@ -4832,6 +4833,7 @@ export function issueThreadInteractionService(
         // lifecycle reflection.
         const updated = await db
           .transaction(async (tx) => {
+            if (["ask_user_questions", "request_confirmation"].includes(row.kind)) await notifyChatRunMilestones(tx, row);
             if (row.kind === "connection_intent") {
               await tx
                 .delete(toolOauthStates)
@@ -4922,6 +4924,7 @@ export function issueThreadInteractionService(
       // request must not outlive a withdrawn card.
       const updated = await db.transaction(async (tx) => {
         await assertInteractionRunWriteAllowed(tx as unknown as Db, issue, actor);
+        if (["ask_user_questions", "request_confirmation"].includes(current.kind)) await notifyChatRunMilestones(tx, current);
         await resolveLinkedToolActionRequests(tx, current, {
           status: "cancelled",
           fromStatuses: ["pending", "approved"],
@@ -5057,6 +5060,7 @@ export function issueThreadInteractionService(
           if (!historicalAnswer) throw interactionIssueClosedError();
         }
         await assertInteractionRunWriteAllowed(tx as unknown as Db, issue, actor);
+        await notifyChatRunMilestones(tx, current);
         await mutationOptions.beforeResolveInTransaction?.(tx);
         const resolvedAt = new Date();
         const [row] = await tx
@@ -5181,6 +5185,7 @@ export function issueThreadInteractionService(
       const now = new Date();
       const updated = await db.transaction(async (tx) => {
         await assertInteractionRunWriteAllowed(tx as unknown as Db, issue, actor);
+        if (["ask_user_questions", "request_confirmation"].includes(current.kind)) await notifyChatRunMilestones(tx, current);
         await resolveLinkedToolActionRequests(tx, current, {
           status: "cancelled",
           fromStatuses: ["pending", "approved"],
@@ -5278,6 +5283,7 @@ export function issueThreadInteractionService(
       const reason = data.reason?.trim() || null;
       const updated = await db.transaction(async (tx) => {
         await assertInteractionRunWriteAllowed(tx as unknown as Db, issue, actor);
+        await notifyChatRunMilestones(tx, current);
         const resolvedAt = new Date();
         const [row] = await tx
           .update(issueThreadInteractions)