Synchzor

Keep your agents in sync. A distributed knowledge store where AI agents and humans share files, search content semantically, and coordinate work through task queues. Built on Keryx.

Features

• Shared Filesystem — Markdown, PDFs, images, and code in a unified filesystem with absolute paths
• Semantic Search — Natural-language queries powered by in-process embeddings (no external LLM API)
• Full-Text Search — Regex content search plus PostgreSQL tsvector + GIN indexes for full-text queries
• Task Queues — Priority-based tasks for coordinating work between agents and humans
• Strict RBAC — Org roles for the control plane, project roles for data access, no implicit permissions
• Audit Logging — Immutable before/after snapshots for every state change

Access Methods

| Interface | Description | |-----------|-------------| | MCP Tools | Drop into Claude Code, Claude Desktop, or any MCP-compatible agent | | CLI | bun install -g synchzor — standalone thin client, zero backend deps | | HTTP API | RESTful JSON with auto-generated OpenAPI docs | | Web Dashboard | File browser, search, tasks, team management, audit logs |

Quick Start

bun install -g synchzor
synchzor login --url https://api.synchzor.com
synchzor file write /docs/hello.md "# Hello World"
synchzor search semantic "hello"

Development

bun install
bun dev          # Start backend + frontend
bun run test     # Run backend tests
bun run test:cli # Run CLI tests

Learn More

Full documentation, quickstart guide, and API reference at synchzor.com.