## **📌 What this article covers**

MCP AI Visibility and the in-product AI Visibility experience do not always expose the same platforms at the same time. A platform can appear in quota or overview while **llmv\_submit\_prompts** rejects it. This article explains that behavior for Perplexity, Claude, DeepSeek, and Muse, and what you can request without using a project slot.

## **🚫 Perplexity on submit versus quota**

If **llmv\_submit\_prompts** returns **Platforms no longer available: perplexity**, Perplexity is not accepted for new MCP prompt runs right now. That rejection is a platform availability rule, not a mistake in your prompt text, filters, or project settings.

**get\_quota** can still list Perplexity after submit has stopped accepting it. Quota shows what the catalog last reported; it does not override the submit check. Do not retry the same call hoping the quota list will force acceptance.

Search Atlas does not mark this change as permanent or temporary inside the MCP error. Treat Perplexity as unavailable for MCP submission until that error stops. There is no customer-side setting to turn it back on.

## **🔎 Claude, DeepSeek, and Muse**

Claude, DeepSeek, and Muse can appear in AI Visibility and in **llmv\_get\_overview** while **llmv\_submit\_prompts** and the platforms filter reject them. Overview reports platforms the product knows about. Submit and the platforms filter only accept platforms currently enabled on the MCP connector.

- Seeing a platform in the UI or overview does not mean you can submit prompts to it through MCP.
- A rejected platform name is expected until MCP enablement catches up with the product view.
- No date is returned by these tools. Availability is added when the connector is updated, not by changing filters or creating a new project.

Submit only the platform values the connector currently accepts. Remove rejected names from the platforms filter instead of resubmitting the full set.

## **📊 Results without using a project slot**

Per-platform results for auto-generated, non-branded prompts are stored on a **tracked** project. Tracking uses one project slot. An untracked project does not use a slot, and it does not keep that full result set.

You cannot get a per-platform breakdown for only non-branded auto-generated prompts on an untracked project without using a project slot. To get those results:

1. Track the project so it uses a slot.
2. Use the auto-generated non-branded prompt set on that tracked project.
3. Read per-platform results from the tracked project after the accepted platforms finish.

If you only need a one-off check, stay untracked and accept that MCP will not return the stored per-platform, non-branded-only report. Tracking is required for that report, even when you filter out branded prompts.

## **✅ What to do next**

- Use **llmv\_submit\_prompts** only with platforms it currently accepts.
- Ignore Perplexity in **get\_quota** while submit says it is no longer available.
- Do not expect Claude, DeepSeek, or Muse on MCP submit just because overview lists them.
- Track a project only when you need saved per-platform results.

If you need further assistance, open the chat widget in the bottom-right corner of the platform and send your message. A member of our support team will reply in the chat.