Developer pages in the dashboard
Webhooks, request logs, the MCP server page, custom LLM and data export: what each page does and where the full guide is.
The Developers row in the sidebar has five tabs. Each is a dashboard view onto something you build with from code.
| Tab | What it does | Full guide |
|---|---|---|
| Webhooks | Endpoints VoiceLayer calls when something happens | Receive with webhooks |
| Request logs | Every API request made with a key | Errors and retries |
| MCP server | Setup for AI clients | MCP |
| Custom LLM | Put the voice on your own LLM | Brain connectors |
| Data export | Send LLM and telemetry data to your own tools | below |
Webhooks
- Open Webhooks and choose Add endpoint.
- In New webhook endpoint, enter the Endpoint URL and an optional description.
- Choose the events to send. Leave all unticked to receive every event except memory events, which you must tick on purpose.
- Save. The Signing secret is shown once. Copy it: you use it to check each delivery is from VoiceLayer.
The table has an on/off switch for each endpoint, and Delete endpoint (you confirm). Verify the signature on every delivery; the guide has working code.
Request logs
A read-only table of API requests made with your keys in the last 14 days: route, status, error code, which key, duration and request id. Filter by All, Errors or Succeeded, and load more at the bottom. An error badge links to the page that explains that error and how to fix it.
MCP server
Shows whether your workspace has an MCP key and the setup steps for each AI client, with copy buttons. To let ChatGPT, Claude or Cursor in with a sign-in instead of a key, see Connect VoiceLayer to ChatGPT, Claude, Cursor. The tools available are listed in MCP.
Custom LLM
Lists the connectors that let a call use your own model. Add connector takes a name, a connection (URL and key) and a model name. The connector token is shown once. A row can be turned off, have its token rotated, or be deleted (you confirm). Full steps: Brain connectors.
Data export
Forward LLM and telemetry data to Langfuse, PostHog or any OTLP endpoint.
- Open Data export and add a target.
- Choose its Kind (Langfuse, PostHog or Generic OTLP). You cannot change the kind later.
- Fill in the Config (JSON), starting from the template shown for that kind. Secrets you enter are write-only: they are not shown again.
- Keep Enabled on.
Each target has an on/off switch, Edit, and Remove (you confirm).