Skip to main content

​
First CLI & MCP

Use the CLI for terminal workflows and scripts. Use MCP when your assistant (Cursor, Claude Desktop, etc.) should call AUVY tools directly.

​
CLI — 60 seconds

npm install -g @auvy-synapse/cli
auvy login --api-key ak_live_...
auvy health
auvy receptors list --output json
auvy receptors execute support-bot --message "Hello"
Machine-readable output: 
auvy commands list --output json
auvy schema commands --output json
auvy jobs stream <jobId> --output jsonl
Full overview: CLI introduction.

​
MCP — connect an assistant

Endpoint: 
https://mcp.auvy.ai/mcp
Send your API key as a Bearer token (or use OAuth when your client supports interactive login). Typical flow:
  1. Add the MCP server in your client settings.
  2. Authenticate with an API key from cortex.auvy.ai.
  3. Ask the assistant to list receptors, invoke agents, or search workspace resources.
Details: MCP Server.

​
Choose CLI vs MCP

SurfaceBest for
CLICI, cron, local debugging, JSON pipelines
MCPIn-IDE agents, chat assistants, tool discovery
SDKApplication code — First SDK call

​
Next steps