kibi for AI Search

kibi: an MCP server that adds local file context to AI assistants

kibi, developed by Looted, is an MCP server that connects AI assistants to a user’s local files for richer context. The app enables indexed searching and full-text reading so models can retrieve specific passages or code snippets for analysis and summarization. It offers a command-line interface, Node.js installation, and MCP compatibility, making it suited to developers, technical writers, and power users who need AI access to local repositories.

It turns MCP-based assistants into file-aware tools

The app implements the Model Context Protocol to present local content to compatible clients, letting models request specific document text and search results. Because it follows MCP, it integrates with MCP-capable software like Claude Desktop, enabling assistants to include passages from files in their responses. This connection replaces manual copy-paste workflows by exposing repository snippets directly to the assistant for tasks such as code review, documentation lookup, and targeted summarization.

Search speed and file-format limits shape output reliability

Indexing and retrieval aim for fast access, which helps the tool deliver relevant passages quickly when the source material is plain text or source code. The app is optimized for text-based files, and its ability to parse PDFs or Word documents depends on the underlying file-reading implementation. Users should expect strong performance on clear text repositories and reduced coverage for complex binary formats unless additional parsers are added.

Local operation and open-source design fit developer workflows

All processing happens on the machine running the server, so data does not get uploaded to third-party clouds and can be audited in the open-source repository. The app runs via Node.js and provides a command-line interface for configuration, which matches developers and power users who manage local services. Its fit depends on having an MCP-compatible client and accepting a Node.js-based deployment in the workflow.

Best suited for MCP users who prioritize private, text-focused retrieval

kibi is a practical choice for developers and writers who need quick, machine-readable context from local repositories and who can run a Node.js service alongside their MCP client. Expect strong results with plain text and source-code collections, and plan for additional tooling or parsers if your workflow relies on complex document formats.