Canyon
Book a Demo
Canyon/Docs/What Canyon can build
Platform

What Canyon can build

Canyon generates complete, production-grade apps: frontend, full-stack, and data-connected. Stacks and components below are examples of what customers are shipping today. The platform is deliberately open-ended — if you need a stack, component or integration we don't list, we will wire it.

BetaUpdated April 2026•Reference•Canyon v1.0
BetaThese docs are in beta
Expect gaps and drift from the live product. Something unclear or missing? Grab 30 minutes with the team and we will walk you through it.
Talk to support

Frontend generation

Production readyReact · TypeScriptResponsiveWCAG 2.1 AA

Every generated app is a modern, typed React application tuned to your design system. The component library, pattern set and accessibility baseline below are the defaults — all of it is extensible at the MCP layer.

Component library
Broad, production-grade component set — tables, forms, dialogs, data grids, command palettes. Accessibility baked in.
Pattern bundles
Whole-page patterns: dashboards, list/detail, wizards, e-commerce flows, chart groups. Each one tuned per theme.
Responsive by default
Mobile-first layouts generated automatically. Works on phone, tablet, desktop, large control-room displays.
Accessibility
WCAG 2.1 AA across every component. Focus order, keyboard navigation, colour contrast — enforced at generation time.
Theming
Light and dark modes driven by the active theme. Every generated app honours your brand tokens automatically.

Full-stack generation

Production readyMulti-runtimeManaged persistence

For apps that need custom business logic, Canyon generates the API, the persistence layer, and the deployment descriptor together — wired to your identity, data and observability stack.

Example applications teams are shipping

Frontend-only React apps
Dashboards, internal tools, review queues. Deployed standalone where no backend is needed.
Live
React + Python API
Operational apps with custom business logic, background jobs and data-connected endpoints.
Live
React + Node API
Apps where the team standardises on a JavaScript runtime end to end.
Live
Apps with a managed database
PostgreSQL on your cloud, schema and migrations generated alongside the code.
Live
Mobile companion views
Responsive layouts optimised for field, floor and phone use.
Live
Vue, Svelte, SolidJS frontends
Framework-agnostic generation on Canyon's roadmap.
Alpha
Go and Java backends
For teams standardising on the JVM or Go toolchain.
Alpha

What gets generated

  • REST and RPC API endpoints with typed request and response validation.
  • Database schema and migrations, wired to a managed PostgreSQL instance in your cloud.
  • CRUD endpoints with input validation and authorization checks at the handler level.
  • Authentication scaffolding wired to your IdP through Canyon’s SSO integration.
  • Container image and deployment descriptor ready for the target environment.
Not seeing your stack?
Canyon’s generation pipeline is stack-agnostic — the frameworks and runtimes we ship with are the ones customers have asked for first. If your team standardises on Vue, Svelte, SolidJS, Go, Java or something else, tell us. New stack templates typically ship in days once we have a representative use case.

Data-aware generation

Production readyCanyon semantic layer

When a project has access to Canyon’s semantic data layer, every chart, table and KPI the platform emits is bound to a governed metric. Row- and column-level permissions are applied automatically; the generated code stays thin and governance stays centralised.

  • KPI dashboards with live connections to your warehouse.
  • Charts and graphs populated from certified measures, not hand-written SQL.
  • Tables with server-side sorting, filtering and pagination out of the box.
  • Real-time refresh for operational dashboards where the data source supports it.
  • Every access logged, every permission evaluated, every query governed.

Extensibility

Canyon is built to be extended at every layer — new stacks, new data sources, new integrations, new observability backends. The extension seam is MCP, the same protocol Claude Code and other agents use to reach tools.

The seam is MCP
New building capabilities — a new database dialect, a new notification sink, a new observability provider — are added as MCP tools without modifying the agent pipeline. Tools are namespaced per server, so conflicts are impossible and governance stays per-tenant.

To add a new capability, expose an MCP server that implements the required tool surface and register it with your organization. It becomes available to every project in that org — including auto-discovery inside the Conversation Agent — on the next request.