AddThisFeature

Moderation Queue

Turn reports into a queue moderators can work through consistently.

involved Admin & Operations

What it adds

A permission-restricted review queue with assignment, locking, policy reasons, escalation, and a full audit trail of every decision.

What your agent is told to do

5
  1. 1

    Build a queue of open items with filters for reason, severity, age, and target type. Sort so the oldest and the most-reported surface first, and make the current backlog size visible.

  2. 2

    Assign or lock an item to one moderator while they work it, with a visible expiry so an abandoned lock releases itself. Two moderators applying opposite decisions to the same item is the failure this feature exists to prevent.

  3. 3

    Require a policy reason on every decision, chosen from a maintained list, plus optional private notes. Free-text-only decisions make patterns impossible to review later.

  4. 4

    Support escalation to a senior queue or a second reviewer, carrying the full history rather than starting a new item.

  5. 5

    Do not let this queue become the app's generic approval tool. Per-record approve and reject workflows are owned by Approval Requests, and the reporter-facing submission flow is owned by Content Reporting; extend those rather than rebuilding them here.

Edge cases it handles

6
  • Evidence must be the snapshot taken at report time. A moderator must be able to see what was reported even after the author edits or deletes it.
  • Multiple reports about the same target should group into one review item while preserving each reporter's reason and count.
  • A decision must be reversible, and the reversal must itself be audited — moderators make mistakes and the original state has to be recoverable.
  • A moderator must not be able to review an item they reported, or one about their own content.
  • Bulk actions need a confirmation that names the count and the reason, and must not silently skip items another moderator has locked.
  • Actions on a deleted or already-actioned target must fail cleanly with an explanation, not throw or double-apply.

Definition of done

8
  • Only users with the moderator role can open the queue or act on an item.
  • An item can be locked to one moderator, and locks expire automatically.
  • Evidence snapshots remain viewable after the original content is edited or deleted.
  • Every decision records actor, policy reason, timestamp, and target, and reversals are audited too.
  • Escalation moves an item with its full history intact.
  • Reporting submission and generic approve/reject flows are not duplicated here.
  • The feature matches the existing design system.
  • No existing functionality is broken.

Related features

How it works

  1. 1

    Copy the link

    Grab the Markdown instruction URL for this feature.

  2. 2

    Give it to your AI

    Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.

  3. 3

    It inspects, then implements

    Your agent reads your existing app first, then adds the feature to fit it.

Works with your stack

These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.

Need it tighter than that? Customize the feature and tell it exactly what you're running.