> ## Documentation Index
> Fetch the complete documentation index at: https://docs.insaion.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Glossary

Common terms used in Insaion.

* Device: a robot or machine running the agent.
* Agent: lightweight software that connects a device to Insaion.
* Site: a group of devices by location or project.
* Tag: a `key:value` pair that describes a device, such as `customer:acme` or `software_version:1.4.2`. Every tag keeps a history of its values.
* Detected tag: a tag the agent reports on its own, such as `os.name`, `ros.distro` or `agent.version`. It can't be edited by hand.
* Manual tag: a tag set by people in your organization, now or from a past date.
* Data tag: a tag on each data point of a metric or log, such as `cpu:cpu-total` or `severity:ERROR`.
* Where: the tag filter of a panel query or an alarm, written in the shared filter language.
* Dashboard variable: a picker for one device tag above a dashboard's panels, applied to every panel.
* Compare panel: a dashboard panel that sets two groups of robots side by side, two tag values or two time ranges, and ranks metrics by how much they changed.
* Device-hour: one device reporting for one hour. Compare panel metrics use it to make groups of different sizes comparable.
* Template: reusable data collection configuration (monitored topics and recording setup) applied to devices.
* Alarm: rule that detects conditions and notifies.
* Incident: a tracked occurrence of an alarm.
* Data Collection: the device configuration section where each topic is set to be monitored, recorded, or both.
* Monitor: ingest a topic as live telemetry, for dashboards, queries and alarms.
* Monitoring stopped: a device whose live telemetry has been switched off from the UI. Nothing streams to the cloud, but recording carries on and the topic selection is kept.
* Record: write a topic to MCAP files in the device's rolling buffer, for playback and upload.
* Rolling buffer: on-device storage holding recent MCAP recordings, trimmed automatically by a retention policy.
* Recording set: one group of recorded topics inside the rolling buffer, recorded into its own MCAP recordings and uploaded as its own file. Called a pipeline in the agent configuration and the recordings list.
* Rule: a glob such as `/camera/*` that records every topic it matches, including ones that appear later.
* Downsampling: keeping only a fraction of a monitored topic's messages, either one in every N (stride) or at most N per second (max rate).
* Host metrics: CPU, memory, disk and network telemetry the agent always collects, published under `/host_metrics/`.
* Telemetry buffer: disk space on the device (1 GiB by default) where live telemetry waits while the device is offline, delivered once the link returns.
* Segment: one uploaded time window of a recording, stored in the cloud as an MCAP file per recording set.
* Upload first: an upload priority. The upload goes before the device's other uploads and takes over from one in progress, which resumes afterwards. Copilot incident uploads always use it.
* Evicted: recorded data the rolling buffer has overwritten to make room for newer data. It can no longer be uploaded.

If something's unclear or you hit a blocker, contact us on

[Discord](https://discord.gg/ZJMPfMmu3S)

or [[contact@insaion.com](mailto:contact@insaion.com)](mailto:contact@insaion.com)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.