kiwifs/kiwifs

Markdown filesystem for agents and teams.

What it solves

KiwiFS addresses the gap between human-readable documentation and agent-accessible data. While Markdown is ideal for both, raw files lack the search, versioning, and structure needed for production AI agents. It replaces the need to choose between databases that agents cannot read and read-only retrieval layers that agents cannot write to.

How it works

It functions as a markdown filesystem where files on disk serve as the single source of truth. It wraps these files in a single Go binary that provides multiple access protocols (REST, MCP, NFS, S3, WebDAV, FUSE) and a built-in web UI. To enable AI capabilities, it implements full-text search (BM25) and pluggable vector search (supporting OpenAI, Ollama, ONNX, etc.), while using Git for atomic versioning and audit trails.

Who it’s for

  • AI agent builders who need a persistent, writable workspace where agents and humans can collaborate on knowledge.
  • Teams seeking a self-hosted, vendor-lock-in-free alternative to Notion or Confluence.
  • Compliance-heavy industries requiring an immutable audit trail of every change via Git.
  • DevOps teams wanting agents to maintain live runbooks that humans can review.

Highlights

  • Native MCP Integration: Includes 62 tools for seamless use with Claude, Cursor, and other MCP clients.
  • Hybrid Search: Combines SQLite FTS5 for full-text search with flexible vector store options (e.g., sqlite-vec, Qdrant, Pinecone).
  • Automatic Versioning: Every write is recorded as an atomic Git commit, allowing for diffs and point-in-time restores.
  • DQL Queries: Supports SQL-like queries over markdown frontmatter for structured data retrieval.
  • Zero-Config Deployment: Distributed as a single binary with an embedded web UI and multiple data importers.

Related

  • Project
    invertase/docs.page

    An open-source platform that transforms public GitHub Markdown files into modern, agent-ready documentation sites with built-in AI chat and MCP support.

  • Project
    kanwas-ai/kanwas

    A multiplayer workspace for teams and AI agents to collaborate on shared documents and evidence using a Git-backed markdown filesystem.

  • Project
    iwe-org/iwe

    A memory system that turns local Markdown files into a queryable knowledge graph for both humans and AI agents, providing structured retrieval via CLI and MCP.

  • Project
    makenotion/notion-mcp-server

    An MCP server that enables AI agents to read, search, and edit Notion workspaces using token-efficient Markdown tools.

  • Project
    arabold/docs-mcp-server

    Grounded Docs MCP Server is a local, version‑aware documentation indexer that scrapes official docs (web, GitHub, npm, PyPI, local files), converts them to searchable text, and optionally adds vector embeddings. It exposes a simple MCP (SSE) API so AI coding assistants (Claude, Cline, Copilot, Gemini‑CLI, etc.) can retrieve exact documentation passages, reducing hallucinations while keeping data private.