dataroo.app

// macOS agent + optional fleet service

dataroo --live --learn — auto-learn your Macs. Inventory everything. Own your data.▍

Every app works stand-alone on one Mac. dataroo.app is the optional sync + fleet service on top — never required.

Download for macOS View fleet demo

> capabilities

Six agents. One index. Zero cloud dependency.

Each piece runs standalone on a single Mac — combine as many as you want.

inventory

Auto-inventory

Hardware, installed software, Homebrew/npm/pip/gem packages, and dotfiles — scanned locally, versioned, and diffed on every run.

dataroo scan --hw --sw --brew --dotfiles

catalog

File catalog + dedupe

Every file gets a content hash, Finder tags, and Spotlight comments. Duplicates surface automatically — no upload required.

dataroo catalog ~/Documents --dedupe

local-ai

Local-AI processing

Runs llama.cpp with Qwen or GLM on your own Macs. Classification, summarization, and tagging — zero cloud tokens spent.

dataroo infer --model qwen2.5-14b --local

wiki

llm-wiki

Promoted files become a queryable local wiki. Claude, Codex, Gemini, and Apple Intelligence can all query it directly.

dataroo wiki query "Q3 roadmap"

fleet

Fleet dashboard + telemetry

Every enrolled Mac streams hardware, software, and file metadata into one dashboard — live wire, host drilldowns, telemetry.

dataroo fleet status

security

Security

CVE exposure cross-referenced against cvedb.io. Keychain and Passwords.app inventory — values never leave the Mac.

dataroo audit --cve --keychain

> fleet demo

What the fleet dashboard looks like

Synthetic data below — click the sidebar to preview each surface.

dataroo.app — fleet overview
~ fleet
  • ├overview
  • ├hosts
  • └telemetry
> inventory
  • ├hardware
  • ├software
  • ├packages
  • └dotfiles
> data
  • ├files
  • ├wiki
  • └sources
> security
  • ├cve-exposure
  • └credentials
~ settings
  • ├agents
  • └theme
5 / 5
hosts_online
1,842,006
files_indexed
21,384
dupes_found
118.4 GB
reclaimable
files_ingested // 14d
enrolled_hosts
hostmodelmacOSdisk usedlast seenstatus
rdmbair15m5MacBook Air 15" M526.1412 GB / 1 TB2m agoonline
rdmbair13m5MacBook Air 13" M526.1288 GB / 512 GB5m agoonline
rdmsM4xMac Studio M4 Max26.01.8 TB / 4 TBjust nowonline
rdmpw3265mMac Pro (Xeon W-3265M)15.63.1 TB / 8 TB14m agoupdate due
rdmpw3275mMac Pro (Xeon W-3275M)15.64.4 TB / 8 TB1h agoupdate due
fleet_cpu % // 60min
tail -f live.wire
14:02:11PKGbrew upgrade node 22.3→22.4 · rdmsM4x
14:01:47FILEindexed 214 files · rdmbair15m5
14:00:59CFG~/.zshrc drift detected · rdmpw3265m
13:59:32FILEmetadata catalog synced · 1,842 rows
13:58:15PKGnpm -g update · 4 packages bumped
13:57:03CFGlaunchd agent com.dataroo.scanner reloaded
> hardware

> architecture

Stand-alone by design

The fleet dashboard is a layer on top — every feature works with it turned off.

$ ~/.dataroo/index.db

Every dataroo agent keeps its own SQLite index on disk. No daemon phones home, no background service you can't inspect — the index is yours to query with plain SQL.

$ xattr -l ~/Documents/file.pdf

Tags and summaries are written as extended attributes and Finder comments, not locked in a proprietary database. Spotlight sees them — so does any AI agent that can read a file system.

$ dataroo sync --off

dataroo.app is the fleet layer on top of the local agent — sync is opt-in, per host, per source. Turn it off and every feature above keeps working exactly the same, minus the dashboard.