UniRoute
Docs

JetBrains AI Assistant API setup

Connect JetBrains AI Assistant to UniRoute with an OpenAI-compatible provider, API key and model assignments.

JetBrains AI Assistant · pay-as-you-go API

Connect AI Assistant in your JetBrains IDE to UniRoute. Add an OpenAI-compatible provider, then choose models for chat and editor actions.

Best for: Developers using IntelliJ IDEA, PyCharm, WebStorm or another JetBrains IDE.

Start setupGet an API key

Quick setup

Sign in to create a key and fill it in automatically Sign in

Copy each value into its matching client field.

Provider

text
OpenAI-compatible

URL

text
https://api.uniroute.app/v1

API Key

text
YOUR_UNIROUTE_API_KEY

Core features

text
deepseek-v4-flash

Instant helpers

text
deepseek-v4-flash

Setup steps

1. Open AI Assistant providers

Open Settings → Tools → AI Assistant → Providers & API keys. Under Third-party AI providers, set Provider to OpenAI-compatible.

Use AI Assistant’s compatible provider for UniRoute; Junie’s native BYOK slots accept keys issued directly by OpenAI or Anthropic.

2. Enter the URL and key

Paste the URL and API Key from Quick setup. Enable Tool calling when using an MCP tool workflow, then click Test Connection and Apply.

3. Choose a chat model

Open AI Chat and its model selector. Choose deepseek-v4-flash in the UniRoute provider section and start a new chat.

4. Assign editor features

Return to Providers & API keys → Models Assignment. Select deepseek-v4-flash for Core features and Instant helpers, then click Apply.

Inline completion and next edit suggestions use the separate AI Completion provider and a FIM-capable model.

What these fields do

FieldPurpose
ProviderOpenAI-compatible connects to a custom Chat Completions endpoint.
URLThe API base ending in /v1; the request uses /chat/completions.
API KeyYour UniRoute request credential.
Tool callingAllows the selected model to call configured MCP tools.
Core features / Instant helpersManual model assignments for editor generation and lightweight helper tasks.

Choose a model

DeepSeek V4

deepseek-v4-flash

Text chat, code questions and lightweight editor helpers.

Claude Sonnet 5.5

claude-sonnet-5-5

Code explanation, review and longer implementation discussions.

Alternative: Claude Code plugin

Install the Claude Code plugin from JetBrains Marketplace and follow the Claude Code setup guide. Merge the following into your existing Claude user settings, preserving unrelated fields:

json
{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.uniroute.app",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_UNIROUTE_API_KEY"
  },
  "model": "claude-sonnet-5-5"
}

Remove an old ANTHROPIC_API_KEY from those settings. Launch claude in the IDE terminal; use /ide to connect an external Claude Code terminal to the IDE.

Confirm the connection

Send a short test: Reply only: hello

Open usage logs to review the model, request status and charge. The test is a real, billed API request.

Check the selected provider in AI Chat, read the streamed response, and match its model in the UniRoute usage log.

Switch back / restore a backup

In Providers & API keys, restore the previous provider and model assignments, then click Apply. Select the previous model in AI Chat.

Not installed yet?

Install the AI Assistant plugin offered for your IDE build, then open its AI Chat tool window.

FAQ

Do I need a JetBrains AI subscription?

This provider connection uses your UniRoute API key. A JetBrains AI subscription supplies JetBrains-hosted features separately.

Why do chat models need feature assignments?

OpenAI-compatible models are available in AI Chat after connection. Choose Core features and Instant helpers manually to use them for the corresponding editor actions.

Where is the provider option in a managed IDE?

Your organization controls third-party-provider access through its JetBrains administration settings. Ask the administrator to enable the compatible provider.

Troubleshooting

Chat works, editor actions do not

The compatible models have no feature assignment.

Choose Core features and Instant helpers under Models Assignment and click Apply.

401 / authentication failed

The API key is incomplete, revoked or lacks permission.

Replace API Key with a complete active UniRoute key and test the connection again.

404 / incorrect endpoint

The URL contains a repeated version or the full chat path.

Set URL to https://api.uniroute.app/v1.

Inline completion uses another model

AI Completion has its own provider settings.

Configure AI Completion with a FIM-capable model for that workflow.

References

JetBrains custom-model setup · Agent activation · Claude Code JetBrains plugin

Related links

All integrations · Manage API keys · Protocols and addresses · Model prices · Usage records