Skip to main content

JetBrains Custom Model Clients with Verdictan

Use this pattern only for a plugin that explicitly supports a custom model endpoint. The Verdictan gateway is not a forward proxy. Do not put it in the global proxy settings for the IDE.

Configuration contract

Set the plugin's values to:

Plugin settingValue
API base URLhttp://127.0.0.1:41002/v1
API keyA Verdictan client API token
ModelAn ID returned by /v1/models

A plugin can use a complete endpoint such as http://127.0.0.1:41002/v1/chat/completions. Follow the plugin's documentation. Keep provider credentials in the gateway configuration.

Verify each feature

A plugin can use different services for chat, completion, embeddings, reranking, or telemetry. Make sure that each feature operates independently. A matching Verdictan event verifies this model path. A missing event is diagnostic input, not proof of bypass.

For JetBrains AI Assistant itself, use its supported provider settings in JetBrains AI Assistant with Verdictan.

Verify

Run verdictan events tail --since 10m --follow. Select the custom model in the plugin. Send a small request. If the event tail stays empty, verify gateway event delivery. Then verify the CLI organization, region, event window, and filters. Then verify the active plugin model and endpoint.

Next steps