tablitz

Recover, manage, search, and back up your OneTab tabs — in Rust

What is tablitz?

tablitz is a Rust-powered CLI tool and MCP server for recovering, managing, and searching your OneTab browser tab data. It extracts 20,000+ tabs from Chrome/Edge/Brave LevelDB stores, stores them in a local SQLite database for fast querying, and exposes them to AI assistants like Claude Desktop and Claude Code through the Model Context Protocol.

Features

• LevelDB Recovery — Extract OneTab data from Chrome, Edge, Brave, and Comet browser profiles
• Import/Export — OneTab pipe (.txt) and markdown (.md) format support, plus JSON/TOML
• Powerful Search — Fuzzy search with scoring and SQL full-text search across titles and URLs
• Deduplication — Three strategies: exact URL, normalized URL, URL+title combination
• Multiple Export Formats — Export tabs to JSON, Markdown, or TOML
• Git-Backed Snapshots — Version-controlled backups with full restore capability
• MCP Server Integration — Expose your tab collection to Claude Desktop and Claude Code

Installation

Build from source with the MCP feature enabled:

cargo build --release --features mcp

The binary will be available at target/release/tablitz.

Quick Start

# 1. Recover tabs from your browser
tablitz recover --browser chrome

# 2. View store statistics
tablitz stats

# 3. Search for tabs
tablitz search "rust tutorial"

# 4. Export results
tablitz export --format markdown --out tabs.md

Commands Reference

Command	Description	Key Flags
recover	Recover OneTab data from browser LevelDB	--browser, --profile, --dry-run, --out
import	Import tab data into the store	--from-onetab-export, --from-onetab-leveldb
export	Export tab data from the store	--format, --out, --filter
search	Search tabs with fuzzy or full-text mode	--mode, --limit
list	List tab groups with optional filtering	--filter, --limit
dedup	Deduplicate tabs using configurable strategy	--strategy, --normalize-titles, --dry-run
init	Initialize tablitz config and data directories	(none)
stats	Show store statistics and top domains	(none)
serve	Start MCP server for AI assistant integration	--port
snapshot	Create git-backed snapshot of the store	--repo, --filename
restore	Restore store from git-backed snapshot	--repo, --commit, --filename
snapshots	List recent snapshots in a git repo	--repo, --limit

MCP Server

To use tablitz with Claude Desktop or Claude Code, add it to your MCP configuration:

{
  "mcpServers": {
    "tablitz": {
      "command": "tablitz",
      "args": ["serve"]
    }
  }
}

The MCP server exposes these tools:

• search_tabs — Fuzzy search tabs by query
• list_groups — List tab groups with optional filtering
• get_stats — Get store statistics and top domains
• recover_from_browser — Recover tabs from browser LevelDB
• import_onetab_export — Import from OneTab export files

Data Formats

tablitz understands OneTab's native formats. See docs/ONETAB-FORMAT.md for detailed format specifications.

Architecture

tablitz uses a multi-crate workspace architecture:

• tablitz-core — Shared types, session models, and domain logic
• tablitz-recover — LevelDB extraction and OneTab format parsing
• tablitz-store — SQLite persistence with FTS5 full-text search
• tablitz-search — Fuzzy search, deduplication, and title normalization
• tablitz-sync — Git-backed snapshot and restore functionality
• tablitz-cli — Command-line interface and MCP server

See docs/ARCHITECTURE.md for architectural details.

Recovery Guide

Need to recover OneTab data from your browser? See docs/RECOVERY.md for step-by-step instructions covering browser profiles, LevelDB paths, and troubleshooting.

License

MIT License — see LICENSE for details.