Capabilities and requirements
GigaAgent combines a base UI with optional capabilities. Most useful scenarios need configured providers.
| Capability | Works immediately? | Required setup |
|---|---|---|
| Sign in to the UI | Yes | Nothing for local setup |
| Chat | No | Connector and language model |
| Streaming output | Yes, after model selection | Language model |
| File upload | Yes | User permissions; processing may need tools or a model |
| RAG | No | Embeddings and vector store |
| Memory | No | User embeddings and vector store |
| Python or shell execution | No | Sandbox runtime |
| Image generation | No | Image generator provider |
| Image analysis | No | Vision-capable model |
| Web search | No | Search provider |
| GitHub | No | GITHUB_PERSONAL_ACCESS_TOKEN |
| VK | No | VK_TOKEN |
| Weather | No | OWM_API_KEY |
| Subagents | Partly | Model and scenario-specific secrets |
| MCP tools | No | Tools passed by the UI for the current dialog |
For a shared server, set a long GIGA_AGENT_SECRET_KEY, change the initial admin password, and configure sandbox access conservatively.