> For the complete documentation index, see [llms.txt](https://docs.visdom.virtuslab.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.visdom.virtuslab.com/visdom-ai-gateway/known-limitations.md).

# Known limitations

What the gateway does not do yet, so none of it is a surprise.

This page reflects the `visdom-ai-gateway` repository on **14 September 2026** and will drift.

## The Visdom layer is not in the request path yet

* **Teams, keys and budgets are managed in the LLM gateway**, not in the Control Plane. They do not follow the people and projects of the other Visdom products.
* **Gateway calls are not in AI Tracing's signed chain.** The gateway keeps its own spend log. The AI Tracing proxy, which does write the chain, is a separate endpoint for Anthropic Messages and OpenAI Responses only.
* **There is no Policy Gate.** Model allow-lists and budgets are enforced by the LLM gateway; nothing decides per call on content, tool or data.
* **No kill switch** across both gateways; a model or a tool server is disabled in each gateway's own admin.

## Guardrails

* **No PII masking and no content classifiers** in the reference deployment. Secret blocking exists only in the AI Tracing proxy.
* **MCP plugins are not configured.** Context Forge's deny lists, filters and rate limits are available but off.

## MCP

* **Any valid token can call any tool** in a virtual server it can reach. Per-team and per-tool access control is roadmap.
* **No per-user OAuth to tool servers.** A tool is called with the credential stored for its server.
* **MCP specification 2026-07-28 is not supported**; clients must speak 2025-11-25 or earlier.

## Operations

* **Single host.** The reference deployment is Docker Compose with one replica of each gateway. There is no Helm chart and no tested high-availability setup yet.
* **No SIEM export** beyond what each upstream logs.
* **Guardrail and gateway latency have not been measured** by us.
* **Some LiteLLM features are Enterprise-only**, among them per-key guardrails, fine-tuning, batch cost tracking, admin SSO and its audit log. The design covers these with Visdom layers, which are not built yet.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.visdom.virtuslab.com/visdom-ai-gateway/known-limitations.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.