First-run setup
Open /setup, configure model endpoints, bootstrap memory, and restart connected runtimes.
After the server starts, open /setup on the Lore Web server.
The Web Console is the browser-based UI for Lore at http://127.0.0.1:18901. /setup is part of it — the first-run configuration page.
Before you begin
You need an OpenAI-compatible API key for embeddings. Any provider that exposes an /v1/embeddings endpoint works: OpenAI, DeepSeek, a local Ollama instance, or a self-hosted embedding service. You will also need a chat model API key for the View LLM (used to generate searchable memory views and Dream suggestions).
If you only have one key that works for both embeddings and chat, you can reuse it across both fields in /setup.
Setup flow
The setup flow writes runtime settings into the app_settings table. It is not the same thing as editing the server .env file.
- Open
http://127.0.0.1:18901/setupor your server URL. - Configure the embedding provider. Lore uses embeddings for semantic retrieval and generated memory views.
- Configure the View LLM provider. Lore uses it to generate searchable views such as gists and questions from memory content, and Dream depends on it.
- Review boot memory defaults. The fixed startup baseline includes global agent rules, soul/style context, user preferences, and active client-specific agent rules.
- Save settings, then restart connected agent runtimes so hooks and MCP clients reload server config.
If the server uses API_TOKEN, browser and API clients must authenticate. Agent clients need the same token in their MCP or plugin config.
Do not skip setup just because the Web Console loads. A server can be online while embeddings, View LLM, cache, backup, or recall settings are still incomplete.