This page is a reference, not an introduction. Each term is also defined where
it first appears on the page that uses it, because nobody clicks “Terminology”
in a sidebar while they are trying to get something done.
Why this page exists
These docs once used four labels for the single act of scanning a project, two unrelated meanings for “workspace”, five for “rule” and three for “gate”. The word “finding” appeared 172 times across 37 pages without being defined anywhere. Each page was internally consistent, so nothing looked wrong until the whole corpus was read at once.The words that carry the product
Finding - one thing Iris Code reports about your code. Every finding is either a blocker or a warning. It is not called an issue. Blocker - a serious finding, such as a committed credential or security smell. Review it before sharing the code. Warning - a finding that needs attention, such as a long file or too many imports. A warning can still affect the health score or a configured threshold. The gate uses your thresholds to decide whether a push or build passes. The blocker or warning label alone does not decide that result. Gate - the decision about whether a change is allowed through, made by comparing a scan against your configured thresholds. There is one gate. “Quality gate” is not a second thing; it is only used to describe what kind of product Iris Code is. Threshold - a configured cap the gate checks against, such asminHealthScore or gateMaxComplexity. Thresholds are not rules. Confusing the
two is the single most common mistake made about enforcement, which is why
the accuracy benchmark opens by separating them.
Health score - a file’s score out of 100. It starts at 100 and decreases as
findings are applied, flooring at 0. Use it to find files that need attention; a high score does not prove that the code works correctly. See Scoring.
Scan - the act of analysing code, at any scope: a file scan, a folder scan,
a workspace scan, a cloud scan. It is not called a review, an audit or an
analysis.
Workspace - the project folder open in your editor. Nothing else. VS Code
owns this word (workspace folders, multi-root workspaces), so Iris Code does not
use it for anything but the folder.
Team - the entity that owns seats, connected repositories, the published
standard, the audit trail and the billing relationship. This used to be called a
workspace, which collided with the editor sense above.
Rules, in their five senses
“Rule” is the one word that legitimately means several things. Each sense is qualified, always:The rest
Snapshot - one saved record of a scan, kept for the daily trend. The saved free scan and the HTML export are not snapshots. Readiness - whether the current scope would pass the gate as it is configured right now. Hotspot - a file that ranks high for size or complexity within a scan. Entitlement - which plan a person’s editor resolves to, from an individual subscription, an individual trial, or a team seat, in that order. See Entitlement. Analysis engine -@iris-code/core, the single implementation every surface
calls, so a finding in JetBrains is the same finding in VS Code, on the same
line.
Code smell - a pattern that is not a defect but makes a defect more likely.
Dependency audit - the separate act of checking package manifests against
OSV advisories. This is the one place “audit” survives as a verb, because
auditing dependencies is genuinely not the same act as scanning source.
Audit trail - the record of what ran, on which commit, under which policy
version. A record, not an act.
Spelling
The product is Iris Code, with a space, in all prose. NeverIrisCode,
never bare Iris.
Machine forms are not prose and are unchanged: iriscode.co, iris-code,
iris (the CLI binary), iris.* settings and command ids, .irisconfig.json,
@iris-code/core, @iris-code/cli.