Quickstart

Connect your AI surfaces in about five minutes.

Five steps, in order: create a workspace, connect Claude Code (and anything else that speaks MCP), bring in what you already have, install the CLI, and see it pay off.

1

Create your workspace

Sign up with email or Google/GitHub. You get a user and an org together — there's no separate "create a team" step.

2

Connect your AI surfaces

Run this in your terminal — your browser opens to approve access. No tokens to paste. The same endpoint serves every other surface that speaks MCP, so this is the only connect step you repeat per tool.

$ claude mcp add --transport http vincosha https://api.vincosha.com/v1/mcp
3

Import what you have

Already have skills or a rules file like CLAUDE.md? Bring them in from the import wizard in the app, or push them straight from a project directory:

$ vin import --project .
4

Install the CLI + sync

The backend repo is private for now, so hosted binaries aren't available yet — clone the repo you have access to and build from source:

$ go build -o vin ./tools/emptor

The package path still reads emptor — that rename is in flight and the command above is correct today.

Then sign in and pull your assigned assets:

$ vin login --server https://api.vincosha.com
$ vin sync

Sync installs skills, commands, agents, MCP configs, and rules files into ~/.claude/, and tracks every file it wrote in a lockfile at ~/.vin/vin-lock.json — reruns are safe, and vin status shows drift with no network call.

5

See it pay off

  • Recall your first shared decision or session note, from inside any connected surface or the dashboard.
  • Watch spend and budgets in Telemetry (admin).
  • Everything above is one API — the full reference is at https://api.vincosha.com/v1/docs.