Connect Claude to your ElevenLabs workspace to create and manage voice agents. Find existing agents, review their configuration, update prompts and voices, duplicate or delete agents.
The feature I did not expect here is estimating LLM usage before you push a change. Managing a fleet of ElevenLabs agents from the chat interface is convenient, but the thing that actually bites in production is a prompt edit that quietly shifts cost or latency per call. I run daily check-in voice calls for older adults, so prompt drift across agent variants is a real operational problem on my side.
Two things I would want to know: does the MCP expose agent versioning, so I can diff a prompt against whatever was live yesterday and roll back if the new one regresses? And is there any confirmation step on the delete tool, given an agent taking live calls is one loose instruction away from vanishing?
Report
if two teammates both have Claude open, what stops one person's duplicate/delete from stepping on an agent the other is mid-edit on? that's the part that'd worry me most about exposing this via chat
Report
Prompt edits are the thing I'd iterate on most, and iterating means occasionally making an agent worse. Does a change made through chat get versioned on the ElevenLabs side so I can roll back to yesterday's prompt, or is the previous version gone once Claude writes it?
Report
The usage estimate before making changes is a useful safety rail. For prompt or voice updates, does the MCP show a diff and require confirmation when projected usage rises or a handoff rule changes? That boundary would matter a lot for production voice agents.
Report
I use Claude to write prompts for ElevenLabs voice generation — is there already something that lets Claude generate the voice directly too, not just agents?
Report
the "estimate LLM usage before making changes" bit is the one I'd want most - is that a token-diff estimate off the prompt edit or does it actually simulate a call against the new config?
ElevenLabs MCP brings voice-agent management directly into Claude, so teams can manage their agents from the chat interface.
It reduces the need to switch between tools by letting users create, review, update, duplicate, and delete ElevenLabs agents in one place.
Key Features
Create, update, duplicate, and delete agents.
List agents and review their summaries.
Check agent knowledge size, widgets, and links.
Estimate LLM usage before making changes.
Benefits
Simplifies voice-agent management.
Makes agent configuration easier to review and maintain.
Helps teams understand potential LLM usage before updates.
Who It’s For: Teams working in code, productivity, design, and sales and marketing.
Sign in to ElevenLabs through Claude to get started.
Refocus
The feature I did not expect here is estimating LLM usage before you push a change. Managing a fleet of ElevenLabs agents from the chat interface is convenient, but the thing that actually bites in production is a prompt edit that quietly shifts cost or latency per call. I run daily check-in voice calls for older adults, so prompt drift across agent variants is a real operational problem on my side.
Two things I would want to know: does the MCP expose agent versioning, so I can diff a prompt against whatever was live yesterday and roll back if the new one regresses? And is there any confirmation step on the delete tool, given an agent taking live calls is one loose instruction away from vanishing?
if two teammates both have Claude open, what stops one person's duplicate/delete from stepping on an agent the other is mid-edit on? that's the part that'd worry me most about exposing this via chat
Prompt edits are the thing I'd iterate on most, and iterating means occasionally making an agent worse. Does a change made through chat get versioned on the ElevenLabs side so I can roll back to yesterday's prompt, or is the previous version gone once Claude writes it?
The usage estimate before making changes is a useful safety rail. For prompt or voice updates, does the MCP show a diff and require confirmation when projected usage rises or a handoff rule changes? That boundary would matter a lot for production voice agents.
I use Claude to write prompts for ElevenLabs voice generation — is there already something that lets Claude generate the voice directly too, not just agents?
the "estimate LLM usage before making changes" bit is the one I'd want most - is that a token-diff estimate off the prompt edit or does it actually simulate a call against the new config?