OUT-3663 | Auto-archive: daily cron job & activity log

[arpandhakal]

Changes

• Cron script for auto archiving tasks.
• What the job basically does is:
  - Auto archives eligible tasks for a workspace. For a task to be eligible in a workspace: workspace's autoArchiveAfter days should be a valid number and:
  - Task should not be deleted.
  - Task should not be archived.
  - Task should be completed
  - Task should not have open subtasks.
  - Task should only be a parent task.
  - Task's completedAt should be older than workspace's autoArchiveAfterDays.
  - If a parent task is eligible after checking the above conditions, their subtasks are also auto archived.
  - Activity logs for all auto archived tasks + subtasks.

Testing Criteria

• Test case for workspace 'us-west-2_M5FSvMUap' having 7 autoArchiveAfterDays.

1. Fetch eligible tasks:

WITH eligible_parents AS (
  SELECT
    t.id,
    t.title,
    t."parentId",
    t."workspaceId",
    t."completedAt",
    t.path,
    EXTRACT(EPOCH FROM (NOW() - t."completedAt")) / 86400 AS days_since_completed
  FROM "Tasks" t
  JOIN "WorkflowStates" ws ON ws.id = t."workflowStateId"
  WHERE t."workspaceId" = 'us-west-2_M5FSvMUap'
    AND t."parentId" IS NULL
    AND ws."type" = 'completed'::"StateType"
    AND t."isArchived" = false
    AND t."deletedAt" IS NULL
    AND t."completedAt" IS NOT NULL
    AND t."completedAt" < NOW() - (7 * INTERVAL '1 day')
    AND NOT EXISTS (
      SELECT 1
      FROM "Tasks" sub
      JOIN "WorkflowStates" sub_ws ON sub_ws.id = sub."workflowStateId"
      WHERE sub."parentId" = t.id
        AND sub."isArchived" = false
        AND sub."deletedAt" IS NULL
        AND sub_ws."type" <> 'completed'::"StateType"
    )
),
cascade_descendants AS (
  SELECT
    t.id,
    t.title,
    t."parentId",
    t."workspaceId",
    t."completedAt",
    t.path,
    EXTRACT(EPOCH FROM (NOW() - t."completedAt")) / 86400 AS days_since_completed
  FROM "Tasks" t
  JOIN "WorkflowStates" ws ON ws.id = t."workflowStateId"
  WHERE t."workspaceId" = 'us-west-2_M5FSvMUap'
    AND t."isArchived" = false
    AND t."deletedAt" IS NULL
    AND ws."type" = 'completed'::"StateType"
    AND t.path IS NOT NULL
    AND EXISTS (
      SELECT 1 FROM eligible_parents ep
      WHERE ep.path IS NOT NULL
        AND t.path <@ ep.path
        AND t.id <> ep.id
    )
)
SELECT 'eligible_parent' AS reason, id, title, "parentId", "workspaceId", "completedAt",
       ROUND(days_since_completed::numeric, 2) AS days_since_completed, path
FROM eligible_parents
UNION ALL
SELECT 'cascade_descendant' AS reason, id, title, "parentId", "workspaceId", "completedAt",
       ROUND(days_since_completed::numeric, 2) AS days_since_completed, path
FROM cascade_descendants
ORDER BY reason, "completedAt";

result : 134

2. Total archived and unarchived tasks in database:

SELECT *
FROM "Tasks"
WHERE "isArchived" = true || false
ORDER BY "lastArchivedDate" DESC
LIMIT 1000000;

Archived : 163

UnArchived : 11030

3. Run the auto archived script manually:

Only one workspace has autoArchived enabled:

Entries to be executed (numbers are correct):

Auto Archived completed:

4. Re-querying (1) and (2) for matching the numbers.

• Eligible task to be archived after running the query : 0

• Number of Archived tasks: 297

• Number of Unarchived tasks: 10896

5. Activity log in recently archived tasks

[linear-code]

OUT-3663 Auto-archive: daily cron job & activity log

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
tasks-app	Ready	Preview, Comment	May 5, 2026 5:11pm

[greptile-apps]

Greptile Summary

This PR introduces a nightly Trigger.dev cron job that auto-archives completed tasks (and their descendants via ltree path) for workspaces with autoArchiveAfterDays configured, and wires up activity log entries for all archived tasks with a null userId to mark them as system-initiated. To support this, ActivityLogs.userId is made nullable across the DB migration, Prisma schema, Zod schemas, and the ActivityLog UI component.

Confidence Score: 5/5

Safe to merge; no P0 or P1 issues found — only minor UI consistency and forward-compatibility P2 notes.

All previously flagged concerns (activity log entries, subtask cascade, per-workspace error isolation) have been addressed. Remaining findings are P2: a missing DotSeparator in the system-activity render path and an overly specific isSystemActivity type guard.

No files require special attention beyond the P2 notes on ActivityLog.tsx.

Important Files Changed

Filename	Overview
src/jobs/auto-archive/auto-archive-completed-tasks.ts	New cron job archiving completed tasks in batches with per-workspace error isolation and activity log creation; LIMIT applies only to eligible_parents so UPDATE and createMany can touch far more than BATCH_SIZE rows when parents have many descendants.
src/app/detail/ui/ActivityLog.tsx	System activity path renders 'Task was auto-archived' but omits DotSeparator, creating a minor visual inconsistency; isSystemActivity check is also scoped only to ARCHIVE_STATE_UPDATED type.
prisma/migrations/20260501000001_make_activity_log_user_id_nullable/migration.sql	Straightforward migration to make ActivityLogs.userId nullable, matching the updated Prisma schema.
src/app/api/activity-logs/services/activity-logger.service.ts	createdBy.userId now accepts null; userId resolution correctly distinguishes between 'createdBy not provided' (infer from token) and 'createdBy provided with null userId' (system activity).
src/app/api/activity-logs/services/activity-log.service.ts	Type signatures widened to accept null userId; isIuLog filter correctly makes system activities visible to clients, and null userId doesn't accidentally match any real client/company member.
src/cmd/backfill-completed-tasks/index.ts	Added null guard on latestLog.userId before assigning completedBy, correctly preventing null from being propagated as a user identity.
src/app/api/activity-logs/const.ts	DBActivityLogSchema updated to make userId nullable, aligning with schema and migration changes.
src/app/api/activity-logs/schemas/LogResponseSchema.ts	LogResponseSchema userId made nullable, consistent with other schema updates.
src/jobs/auto-archive/index.ts	Simple barrel export for the new auto-archive job.

Sequence Diagram

sequenceDiagram
    participant Trigger as Trigger.dev (2 AM cron)
    participant Job as auto-archive job
    participant DB as PostgreSQL

    Trigger->>Job: run(payload)
    Job->>DB: SELECT workspaces WHERE autoArchiveAfterDays > 0
    DB-->>Job: [{workspaceId, autoArchiveAfterDays}, ...]

    loop per workspace
        loop batch until 0 rows returned
            Job->>DB: CTE eligible_parents (LIMIT 200) + cascade_targets<br/>UPDATE Tasks SET isArchived=true RETURNING id
            DB-->>Job: archivedRows[]
            Job->>DB: activityLog.createMany({userId:null, userRole:internalUser})
        end
        Job->>Job: log workspace totals
    end

    Job->>Trigger: return {totalArchived, workspaceCount}

Loading

_{Reviews (2): Last reviewed commit: "fix(OUT-3663): system activity logs uses..." | Re-trigger Greptile}

[vercel]

Deployment failed with the following error:

Deploying Serverless Functions to multiple regions is restricted to the Pro and Enterprise plans.

Learn More: https://vercel.link/multiple-function-regions

[arpandhakal]

@greptileai

[arpandhakal]

@greptileai please review

[priosshrsth]

lgtm @arpandhakal