# Rhylthyme

> Rhylthyme schedules multi-step work that a person carries out against a clock: cooking several dishes, lab protocols, event run-of-shows, workouts. A schedule is a JSON "program" of parallel tracks, timed steps, start triggers and shared-equipment limits. The deliverable is a live timeline URL the person follows on a phone.

If you are an AI assistant asked to "use Rhylthyme": call the hosted MCP server below. Do not drive this website in a browser and do not look for a local server. No account or API key is needed for anything on this page.

## Hosted MCP server (Streamable HTTP, stateless)

- Endpoint: https://mcp.rhylthyme.com/mcp
- All endpoints: https://mcp.rhylthyme.com/mcp (general), /kitchen/mcp, /lab/mcp, /events/mcp, /gym/mcp
- New schedule: build the program JSON, call `validate_program` and fix every finding, optionally `analyze_schedule` with `finishAt` for clock times, then `visualize_schedule`. Its `structuredContent.url` is the live timeline; give the user that link rather than describing the schedule in prose.
- Existing content: `search_public_recipes`, then `load_public_recipe`.
- The program shape is described on `validate_program`'s `program` argument in `tools/list`; `resources/read` of `rhylthyme://guide/authoring` is a one-page guide.
- Without an MCP client, one POST works with no handshake:

```
curl -s https://mcp.rhylthyme.com/mcp -H 'Content-Type: application/json' -H 'Accept: application/json'   -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"validate_program","arguments":{"program":{}}}}'
```

## Connecting an assistant

- Claude: Settings > Connectors > Add custom connector > https://mcp.rhylthyme.com/mcp
- ChatGPT: Settings > Apps & Connectors > Advanced settings > Developer mode, then create a connector with https://mcp.rhylthyme.com/mcp
- Claude Code: `claude mcp add --transport http rhylthyme https://mcp.rhylthyme.com/mcp`
- Clients that only launch a command: `pip install rhylthyme-mcp`, command `rhylthyme-mcp` (a stdio bridge to the hosted server)

## Command line

- `pip install rhylthyme`, then `rhylthyme validate p.json`, `rhylthyme analyze p.json --finish-at 18:00`, `rhylthyme publish p.json` (prints the live-timeline URL)

## Docs

- [MCP server](https://docs.rhylthyme.com/web-app/mcp/)
- [Program schema](https://docs.rhylthyme.com/development/schema/)
- [Concepts](https://docs.rhylthyme.com/getting-started/concepts/)
- [Source: MCP server](https://github.com/rhylthyme/rhylthyme-mcp), [CLI and Claude skill](https://github.com/rhylthyme/rhylthyme-cli-runner)