MCP Server

alegra-cli can expose its entire command tree as a Model Context Protocol server, so AI agents (Claude, etc.) can drive Alegra through well-described tools — one per command.

alegra mcp --help

Each CLI command becomes an MCP tool named alegra_<resource>_<action> (e.g. alegra_invoices_list, alegra_contacts_create) whose input schema is derived from that command's flags and arguments.

Credentials are resolved exactly as for the CLI (profile → keyring/env), and the sensitive --show-token and --profile flags are excluded from the exposed tool schemas.

Warning

The MCP server can create, modify, and delete real accounting records. Run it against a sandbox profile while developing, and review the actions your agent is allowed to take.

Wire it into your editor automatically

Rather than editing config files by hand, let the CLI write them for you. There are subcommands for the common MCP hosts:

alegra mcp claude enable     # add the server to Claude Desktop's config
alegra mcp cursor enable     # Cursor
alegra mcp vscode enable     # VS Code

alegra mcp claude list       # show configured servers
alegra mcp claude disable    # remove it again

Each enable writes the host's MCP config to launch alegra mcp start (the stdio server) with your current setup.

Transports

Command	Transport	Use
alegra mcp start	stdio	The default; what the editor wiring launches. Also: claude mcp add alegra -- alegra mcp start.
alegra mcp stream	HTTP	Long-running server for remote/shared agents. Flags: --host, --port (default 8080), --log-level.
alegra mcp tools	—	Print the full tool schema as JSON (one tool per command) — handy for inspection or feeding another system.

# Run the HTTP server on a custom port
alegra mcp stream --port 9000 --log-level debug

# Inspect the exposed tools
alegra mcp tools | jq '.[].name'

Then ask the agent to, for example, "list this month's open invoices in Alegra".