Concepts

01Core objects

02Organization

Your workspace. Everything (properties, issues, runs, and tasks) is scoped to an organization.

03Property

The app/site you’re testing. A property typically maps to a domain or environment (production, staging).

A property also includes an Application Context workspace, where you can bootstrap and manage structured context for that property. Use it to review live findings, edit entries manually, and restore earlier versions when needed. A property can also build a Site Graph from observed page activity so you can understand how pages connect. Learn more in the Application Context reference.

04Credentials

Saved login information that allows Canary to access authenticated areas of your app.

05Crawl

A crawl discovers pages and builds a map of what Canary can reach.

06Pages

Discovered URLs (grouped into canonical pages) with observed history and QA status.

07Site Graph

A property-level map built from observed page activity. Use Site Graph to understand how pages connect across your property and spot relationships that are not obvious from a simple page list.

08Synthetic Users

AI personas that explore your property like real users. A Synthetic User can return across multiple days, build on what it has already learned, and broaden coverage over time instead of acting like a single isolated session.

Synthetic Users operate inside a universe, which is the shared exploration space for that persona and property. The universe keeps the accumulated app knowledge, discovered areas, and coverage history that future run days use to decide what to revisit or explore next.

09Runs

Automated QA executions. Runs generate artifacts (screenshots, traces) and can open/close issues. Runs can also execute flows and agent-driven task work.

A Synthetic User run can also span multiple orchestrated days. In that model, the run is the top-level execution record, and each day contributes new findings, updated knowledge, and end-of-day reporting back into the same universe.

For organizations with the evals product enabled, eval runs are also visible as a run object. Use them to review read-only eval execution details. Learn more in the Evals reference.

10Tasks

Agent-driven work tracked in a central tasks area for organizations with access. Use tasks to view, filter, claim, trigger, complete, and track work, and follow related execution details. Learn more in the Tasks reference.

11Issues

Problems detected by Canary (e.g., broken UI, auth failures, regressions). Issues are deduped and tracked over time.

Issues from Synthetic User activity can be reviewed per run day or as aggregated findings across the full run, which helps you separate a one-day problem from a repeated issue seen across the universe’s exploration.

12Flows

Repeatable user journeys you define (e.g., “checkout”, “invite teammate”). Flows can be executed in runs.

Flows can also define workflow inputs. Use inputs to pass run-time values into a flow instead of hard-coding them, so you can reuse the same flow in different scenarios.

13Seed Data

Reusable setup data and workflows you manage from a dedicated catalog experience. Use Seed Data to prepare a property for testing and add paired teardown workflows when you need cleanup after a run. Learn more in the Seed Data reference.

14Test suites (smoke suites)

A test suite is a set of flows run together. A smoke suite is the small “must-pass” set you run on every release to catch breakages quickly.

15Requirements

Product requirements (often imported from Jira) that issues and flows can be associated with.

16Attachments

Uploaded artifacts (images/PDFs) used as evidence or training/context for QA and investigations.

17API Catalog

Captured API requests/responses observed during QA for debugging and understanding app behavior.

Core objects with supported shared prefixed IDs can now open directly from their ID links. Use the Universal ID Lookup reference to see which record types are supported and how routing works across organizations.

18How objects relate

Use these relationships to understand where work starts and how results get tracked.

Object	Relates to
Organization	Contains your properties, runs, issues, and tasks.
Property	Contains the pages, flows, runs, and other QA activity for a specific app or environment. A property also includes its Application Context workspace and Site Graph.
Page	Belongs to a property. Pages also contribute observed activity that helps build the Site Graph.
Flow	Defines a repeatable journey that you can execute in a run. Flows can also accept workflow inputs so you can reuse them with different run-time values.
Synthetic User	Explores a property as an AI persona. A Synthetic User operates inside a universe, where it accumulates shared knowledge and coverage across multi-day run activity.
Run	Executes QA work for a property. A run can execute flows, supply workflow inputs, generate artifacts, and produce issues. Synthetic User runs can also span multiple days while contributing findings and knowledge back to a universe. Runs that support shared prefixed IDs can also be opened directly from those IDs.
Task	Tracks agent-driven work. A task can trigger or follow work that executes in runs and may use flows as part of that execution. Tasks that support shared prefixed IDs can also be opened directly from those IDs.
Issue	Captures a problem found during a run or related task execution. Issues can be reviewed from individual run days or as aggregated findings for a longer Synthetic User run. Issues that support shared prefixed IDs can also be opened directly from those IDs.
Seed Data	Stores reusable setup and teardown workflows for a property so you can prepare and clean up test state consistently.

In practice, you often start with a property, review its Application Context and Site Graph to understand the app, define a flow, and then execute it in a run. When you need the same flow to work across multiple scenarios, pass workflow inputs at run time instead of creating duplicate flows. After execution, review the resulting issues and artifacts. When you use agent-driven work, tasks give you a dedicated place to manage that work and track the runs connected to it.

When you use Synthetic Users, start with the persona and its universe. Then review how each run day expands what that Synthetic User knows about your property, what it explored, and which issues repeated or changed over time.

You can also open supported records directly from a shared prefixed ID link when someone sends you an object ID. Canary routes you to the correct record page automatically and prompts you to switch organizations if needed. For supported record types and routing details, see the Universal ID Lookup reference.

19Related

• Learn more about eval runs in the Evals reference.
• Learn more about managing reusable setup and teardown workflows in the Seed Data reference.
• Learn more about property context in the Application Context reference.
• Learn more about tasks in the Tasks reference.
• Learn more about opening supported records from shared prefixed IDs in the Universal ID Lookup reference.