ReDock + OpenAI
GPT, at the helm.
ReDock speaks the OpenAI-compatible API, so GPT models (and any service that mimics the API) can drive the agent with the same tools, fence and receipts.
Free · works with OpenAI as OpenAI · official marks, interoperability only What ReDock does with it
GPT models as your agent.
Your key Stored in your system credential store, sent only to the endpoint you configured.
OpenAI-compatible Point the provider at a compatible endpoint and bring your own models.
Same fence Permissions, receipts and undo do not change with the model.
Build it in an afternoon
Set it up
Three steps, one window, receipts for the parts the agent did.
- 01Settings → AI → OpenAI → paste the key (or an endpoint + key).
- 02Pick the model at the composer.
- 03Ask; read the receipt.
OpenAI on the blog
MCP Claude Desktop + your local sites: connecting an MCP server that can ship Connect Claude Desktop to a local MCP server at localhost:47200 and it can create, run, author, share, and deploy real websites, no terminal, no API keys. Read → MCP Claude MCP tutorial: build and publish a website by chatting A practical Claude MCP tutorial: connect Claude to ReDock's local MCP server and go from empty Mac to a published website in one conversation. Read → MCP Cursor + MCP: letting your editor's agent run real websites Cursor's agent can call MCP tools. Point it at ReDock on localhost:47200 and it can create, run, author, and deploy real local sites, without leaving the editor. Read → MCP Easy MCP setup: zero config, no terminal, no JSON files Easy MCP setup means pointing a client at a running server, no JSON editing, no terminal. ReDock exposes its whole surface at localhost:47200. Read →
Calm ships tonight.
Download ReDock, berth your first site, and watch the AI work, receipts and all. Five minutes, no terminal.
Free · macOS and Windows · your sites stay on your machine