Remote Control: clarify whether the eligibility restrictions are architectural or preview limitations
Documentation Type
Unclear or ambiguous documentation
Documentation Location
https://code.claude.com/docs/en/remote-control
Section/Topic
Requirements, and the "Remote Control is disabled by your organization's policy"
troubleshooting entry.
Current Documentation
The page states that Remote Control:
- is not available on Amazon Bedrock, Google Cloud's Agent Platform, or Microsoft Foundry;
- is disabled when
ANTHROPIC_BASE_URLpoints at a host other thanapi.anthropic.com,
such as an LLM gateway or proxy;
- cannot be enabled by organizations with data retention requirements such as Zero Data
Retention — "Organizations with compliance requirements such as Zero Data Retention
can't enable Remote Control";
- is off by default on Team and Enterprise until an Owner enables it.
What's Wrong or Missing?
Each restriction is stated accurately, but the documentation never says why, nor
whether these are permanent design boundaries or current limitations of a research
preview. Those two readings lead to opposite decisions.
The reason it matters: Remote Control stores the session transcript on Anthropic servers,
which explains the ZDR exclusion as an architectural consequence rather than a policy
choice. If that reading is right, it is unlikely to change, and it should be said plainly.
The ANTHROPIC_BASE_URL restriction has a similar flavour — Remote Control needs a
claude.ai backend to pair with — but the docs leave the reader to infer it.
Concretely, someone evaluating whether to run Claude Code workspaces on their own
infrastructure cannot tell, from the current text, whether they are looking at a
temporary gap worth waiting out or a boundary to design around. That is exactly the
audience most likely to be in one of the excluded categories.
Suggested Improvement
Add a short paragraph to Requirements stating, for each restriction, whether it follows
from the architecture or is a preview limitation. Something like:
Remote Control pairs your local session with a claude.ai backend and stores the session transcript on Anthropic servers so it can be synchronised across your devices. This is why it is unavailable when requests do not go directly to api.anthropic.com (Bedrock, Vertex, Microsoft Foundry, or an LLM gateway), and why organizations with Zero Data Retention cannot enable it. These follow from how the feature works rather than from a policy setting, and are not expected to change while the transcript is stored server-side.
If, on the other hand, any of these are expected to be lifted, saying so would be equally
valuable — it is the uncertainty, not the restriction, that is hard to plan around.