Developers

Build against Canopy, not just click through it.

Whatever you're building - a script, a CI step, an internal tool, or an AI agent - Both products expose first-class programmatic interfaces, including APIs, dashboards, CLI and SDK surfaces, with MCP tooling where a capability supports it. This page is the starting point; reference material lives on /docs.

Where to start

  • Building grounded AI answers or workflows? Start with Cortex.
  • Deploying an app? Start with Canopy Host.
  • Want the full picture first? Read Platform.

Interfaces

Programmatic interfaces on both products, with capability-specific coverage.

DashboardThe default way most people use either product - no code required to get started.
REST APIThe interface everything else here is built on top of. Nothing on either product is dashboard-only.
CLI`canopy` for Canopy Host, a matching CLI for Cortex - scripting, CI, and local setup.
SDKA typed TypeScript client, `@canopy/sdk`, for building your own tools against either product.
MCP serverLets AI coding agents call either product’s capabilities as tools, under the same auth model as the API.

Cortex for developers

OpenAI-compatible, drop-in ready.

Point existing OpenAI-client code at your Cortex project endpoint and get retrieval-augmented, cited answers back. Cortex has its own CLI and MCP server, separate from Canopy Host's.

Canopy Host for developers

Push to deploy, script the rest.

Everything the dashboard does - projects, environments, domains, variables, builds, releases - is also available from the canopy CLI and the REST API. Canopy Connect pipelines run identically whether triggered by a push, the CLI, or the API.

For platform and infrastructure engineers

Curious what actually runs your deploy?

You don't need to know this to ship on Canopy Host. If you want it anyway: Canopy Forest is the execution layer underneath, with its own page covering desired-vs-observed state, build/runtime separation, and the real infrastructure primitives it's built on.

Reference material

Full API and CLI reference.

Endpoint-level documentation, request/response examples, and error handling live on the docs page.