Zenii, developed by Sprklai, is an MCP server tool built primarily with Rust for its core logic. It functions as a local AI backend, providing a platform for agent tooling and local automations. Zenii is distributed as pre-compiled binaries for various platforms including Linux, macOS, Windows, and ARM, or can be built from source. It caters to developers and teams seeking to integrate AI capabilities and custom tools within their workflows, emphasizing local control and privacy. The project also incorporates JavaScript for its frontend components.

What does Zenii connect to?

Zenii operates as both an MCP server and client, connecting to a local AI backend service. As a server, it exposes its tools to compatible AI clients like Claude Code, Cursor, and Gemini CLI. As a client, it consumes external MCP servers such as GitHub, Postgres, and Filesystem, making their capabilities available within its agent loop. Its daemon supports HTTP (REST API) and WebSocket endpoints, while its MCP server functionality uses stdio as a transport protocol. Plugins communicate via newline-delimited JSON-RPC 2.0 over stdin/stdout.

Core features and tools

Zenii exposes a set of callable tools, including 16 base tools or 19 Zenii tools. Specific functionalities are accessible via HTTP API endpoints, such as /chat for AI model interaction, /memory for fact storage, /wiki/ingest and /wiki/query for the LLM Wiki, /workflows/{workflow_id}/run for executing workflows, and /scheduler/jobs for managing tasks. Its Polyglot Agent Runtime (PAR) also allows running external Python or Node.js code as custom tools, extending its capabilities.

Who is it for?

Zenii is for developers building local automations, coding workflows, or agent tooling. It suits teams seeking a single internal AI backend and individuals desiring a native desktop UI with a scriptable local API, especially those who prioritize privacy and local control. It provides a platform for integrating various AI capabilities and custom code. This is not the right fit if you require a system that does not use an API token for access.

How to get started

The server authenticates using an API key passed as an environment variable. This token is provided in an Authorization: Bearer $ZENII_TOKEN header for API calls, with stdio serving as a transport protocol for MCP server functionality.

Use cases

• Users can describe workflows in plain English, and Zenii automatically generates the TOML configuration and opens a visual builder.
• Documents can be ingested into Zenii's LLM Wiki, enabling agents to instantly answer questions against the compiled knowledge base.
• Any prompt or workflow can be triggered on a cron schedule, executing as a full agent turn with access to all available tools.

Zenii provides a versatile platform for local AI backend operations and agent tooling, supporting various integrations and custom code execution. Its focus on local control and privacy makes it a suitable choice for developers and teams. The system's ability to manage scheduled tasks and ingest knowledge offers practical applications.