Quickstart

Connect Oriyn MCP for agents, install the CLI for terminal work, authenticate, link your repo to a product, verify status, sync intelligence, then start a research run from the terminal or web app.

Before You Start

• An Oriyn beta account with access to at least one organization.
• A product with analytics or behavioral data ready to connect.
• A local repo or workspace where oriyn.json can be committed for agents and teammates.
• Bun 1.2 or newer if you install the npm package instead of the standalone binary.

Install And Log In

Recommended installer

curl -fsSL https://oriyn.ai/install.sh | bash

oriyn auth login

The installer downloads a standalone binary for macOS and Linux. Package-manager installs are also available, but the published npm bin runs on Bun.

Connect Agents Through MCP

Add Oriyn as a remote MCP server so agents can call product, persona, pattern, and research tools directly.

npx -y install-mcp@latest https://mcp.oriyn.ai --client claude --oauth yes

Replace claude with your MCP client, such as codex, cursor, windsurf, vscode, or claude-code.

Or add the remote server manually:

{
  "mcpServers": {
    "oriyn": {
      "url": "https://mcp.oriyn.ai"
    }
  }
}

Install the skill after MCP if you want coding agents to know when to use Oriyn and how to summarize results:

npx skills add oriyn-ai/skills

Package managers

bun add -g oriyn

Link And Check A Product

Repo setup

cd <your-product-repo>
oriyn link
oriyn status

oriyn link writes an oriyn.json file with the organization and product identifiers. Commit it so agents, CI, and teammates can resolve the same product without guessing.

Run Your First Research Workflow

Research run

oriyn sync
oriyn personas
oriyn research ab-test --question "Which onboarding offer is clearer?" --a "Free trial" --b "Freemium tier"

A good first research run is one clear product question with a user-visible outcome. Oriyn returns structured results, participant summaries, and caveats.