🔍 Transparent by design
Every field. Public. No exceptions.
This is the complete list of what the Verdaic agent can send, grouped by the collection category
that controls it. Your admin toggles categories in a signed, versioned, audit-logged manifest — and every
device shows its user the exact payloads locally (http://127.0.0.1:47800).
Never collected, on any plan
Files or file names · messages or email · browsing history or URLs · keystrokes · screen contents ·
window titles · locations. Anonymous mode (default on) additionally drops hostnames and any
per-user linkage. Privacy features are never paywalled.
Inventory On by default · hardware only
| Field(s) | What it is, and why |
manufacturer, model | Device make/model — keys the embodied-carbon and power models |
chassis (+ hypervisor) | laptop / desktop / server / vm — decides measured vs modelled energy |
cpu model, cores, threads | Feeds the power model |
ram_gb, disks (model, size, type), gpus | Capacity and component power deltas |
os_version, hostname | Hostname is dropped when anonymous mode is on |
Power & energy On by default · the core of the product
| Field(s) | What it is, and why |
energy_wh per 5-min bucket | Battery-measured on laptops; modelled from CPU × power curve elsewhere |
energy_src | measured | estimated — the honest-data tag, on every figure |
on_battery | AC vs battery per bucket |
Utilisation On by default · load, never content
| Field(s) | What it is, and why |
cpu_pct, mem_pct per 5-min bucket | Averaged from 60-second samples |
disk_read_mb, disk_write_mb | Volumes only |
idle_seconds, screen_on_seconds | Input-idle time (or CPU heuristic on servers — estimated) |
events: boot / shutdown / sleep / resume | Power-state changes with timestamps |
Per-app energy Pro · off until enabled · names + CPU time only
| Field(s) | What it is, and why |
app name | Executable name (e.g. "chrome") — never window titles, documents or URLs |
cpu_seconds per app per day | Cumulative CPU time |
energy_mwh per app per day | Device energy attributed by CPU share — always tagged estimated |
Software inventory Pro · off until enabled
| Field(s) | What it is, and why |
installed apps: name, version, publisher | From the OS package registry — no usage tracking |
startup items: name, command, location | What launches at boot (startup-bloat insight) |
Health & patch posture Pro · off until enabled
| Field(s) | What it is, and why |
battery_health_pct, battery_cycles | Design vs full-charge capacity — keep-vs-replace signal |
SMART predict_failure per disk | The firmware's own failure verdict; no vendor blobs |
os_build, last_patch_at, last_patch_id, hotfix_count | Patch recency |
pending_reboot | Updates waiting on a restart |
Network volumes Pro · off by default everywhere
| Field(s) | What it is, and why |
net_rx_mb, net_tx_mb per 5-min bucket | Byte counts only — never destinations, domains or content |
How it moves and how long it lives
Samples are aggregated on the device into 5-minute buckets — raw per-second data never leaves the
machine. Buckets upload over TLS with a device-scoped token, are pruned per your plan's retention window,
and daily summaries power the dashboards. Manifest changes are audit-logged; full export or deletion of
your organisation's data is available any time via hello@verdaic.com.