# Plans: open source or hosted preview

> One method, two ways to have it. The difference is who runs it and what it costs us to run, not which skills you get.

<a id="open-source"></a>

## Open source — for individuals and developers. Free, permanently.

The method is open so any architect can read it, run it and fork it. Every skill, tool, agent and workflow; runs in Claude Code or Codex with no account; firm and project records as files in your workspace; extend by forking.

- [Fork in GitHub](https://github.com/AlpacaLabsLLC/skills-for-architects)
- [Install the plugin](https://architecturestudio.ai/docs/install?method=plugin)

<a id="hosted"></a>

## Hosted preview — for firms on Claude or ChatGPT. Free for firms during the preview.

Firms work in Claude and ChatGPT, not in a terminal. The hosted connection brings the same method there, with an account per firm, and stays free while we build what a firm needs and a fork cannot give it: shared memory, telemetry and permissions.

What you get today: the same skills over MCP, managed release updates, Norma as coordinator, email support, no service commitments yet. Your Claude or ChatGPT subscription and usage stay with your provider.

What we are working on: shared memory (firm rules, decisions and project records every skill reads, for everyone in the firm), telemetry (who uses what, which skills carry the work, monthly) and permissions (seats, roles and what each person and the assistant can access).

- [Configure MCP](https://architecturestudio.ai/docs/install)