> ## Documentation Index
> Fetch the complete documentation index at: https://apidocs.writesonic.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Troubleshooting

> The things that actually go wrong, and what to do about each one.

Match what you see to the section below, then follow the checks in order.

## Pages is empty or nearly empty right after setup

This is expected. Search Console and Analytics backfill over the first few days. AI visibility builds as prompts run. Tasks work immediately, so [start there](/agents/what-to-expect).

## A task has sat In Progress for a while and nothing seems to be happening

In Progress means the agent is working.

Open the task's activity feed to see its steps, and allow several minutes for a first task. [Learn how task states work](/agents/tasks).

## Nothing has happened and the task is waiting on you

There is no Blocked status.

Tasks needing your input appear as Pending Approval or In Review. Those are the two states to watch in your [task list](/agents/tasks).

## The deliverable came back generic

Thin context is usually the cause. Check the [Brand Kit](/agents/set-up-brand-kit), especially the audience and voice. Then check that the content type matches the format. Comment on the deliverable instead of rewriting the task, as explained in [writing good tasks](/agents/writing-good-tasks).

## The agent did the task but ignored the house rules for that format

Format rules belong in content types, not the task description. See [where context lives](/agents/where-context-lives).

## A Knowledge Base page shows Sync failed

Agents couldn't fetch the source.

Select **Sync now** for that source, then check that the page is reachable and isn't blocked. See [Knowledge Base](/agents/knowledge-base).

## Work is not appearing in the CMS

Confirm that a CMS connection exists and is assigned to this project. Work arrives as a draft, never as a published page. Look in your CMS drafts before reconnecting anything. See [publishing](/agents/publishing).

## A change was made and the numbers have not moved

Whether a change worked is a 14 to 28 day question.

First verify that the change shipped and the page was indexed. Then measure the result in the [activity log](/agents/activity-log).

## You cannot tell which of the two audit tools to use

The AI crawler audit checks whether AI systems can reach and read your pages. The SEO audit checks technical and on-page health.

## A playbook produced a weak first result

A playbook is only as specific as the description used to build it. Open the editor, tighten the steps, and run it on one row before running a grid. See [build a playbook](/agents/build-a-playbook).

## What's next

<CardGroup cols={2}>
  <Card title="What to expect" href="/agents/what-to-expect" icon="hourglass">
    What's normal early on.
  </Card>

  <Card title="Glossary" href="/agents/glossary" icon="book">
    Every term, defined once.
  </Card>
</CardGroup>
