Eidon
Self-hosted BYOK AI assistant workbench supports conversation, tools, memory, automation, MCP and browser automation.
Integrated lazy cat oidc login, or default account admin/12345678 Eidon is a self-hosted AI assistant workbench. It runs in Lazy Cat Micro Service, data is stored in the local application directory, and users can access their own model services for daily dialogue, writing, code assistance, data collation, automated tasks, MCP tool calls and multi-person collaboration. ##Who is suitable for use - Users who want to run private AI assistants on their lazy cat microservice. - Users who already have OpenAI compatible interfaces, DeepSeek, OpenRouter, Ollama Cloud, GLM and other model service accounts. - Small teams that need to save conversations over time, configure multiple models, use tools and automate tasks. - Self-managed users who want to keep their data as much as possible on their devices. ##Configure Model Services Eidon does not have any built-in model API keys by default and needs to be configured by users themselves. Entry path: ```text Settings -> Providers ``` Common configuration methods: 1. Click Add or Edit Provider. 2. Select the Provider type **OpenAI compatible**. 3. Provider preset can be selected or used **Manual configuration**. 4. Fill in the API base URL, Model, and API key. 5. Select the correct API mode. 6. Click Save before Test. ### DeepSeek Recommended Configuration If using DeepSeek, we recommend: ```text API base URL: https://api.deepseek.com/v1 Model: deepseek-chat API mode: chat_completions ``` Do not use the `responses` pattern to test DeepSeek, otherwise 404 may be returned. ###Picture understanding instructions Ordinary 'deepseek-chat' cannot directly look at the picture. If the model replies "Unable to directly view the picture content" after uploading the picture, this is a limitation of the model's capabilities and not a lazy cat deployment error. To understand the picture, one of the conditions needs to be met: - Switch to a model that supports visual input and configure Vision mode to native visual mode. - Configure MCP services that support image analysis and switch Vision mode to MCP.
