Editor schemas (autocomplete for hosts.json & settings.toml)¶

otto can generate JSON Schema for the files you edit by hand — hosts.json, settings.toml, and the reservations JSON — so your editor offers field autocomplete and flags typos. The schemas are generated from the pydantic models inside the otto you have installed, so they always match your version. There is nothing to download and nothing that can go stale.

Generate the schemas¶

otto schema export --out schemas

This writes (into schemas/):

Run it again after upgrading otto to pick up new fields. Custom host classes registered via an init module in .otto/settings.toml are included automatically — each gets its own <type>-host.schema.json and an entry in hosts.schema.json. Pass --builtins-only to emit just the built-in types (unix, embedded, zephyr), excluding any custom ones.

VS Code¶

hosts.json and the reservations JSON are covered by the built-in JSON language server. Add to your workspace .vscode/settings.json:

{
  "json.schemas": [
    { "fileMatch": ["**/hosts.json"], "url": "./schemas/hosts.schema.json" },
    { "fileMatch": ["**/reservations.json"], "url": "./schemas/reservations.schema.json" }
  ]
}

For settings.toml, install the Even Better TOML extension and add:

{
  "evenBetterToml.schema.associations": {
    ".*/settings\\.toml$": "./schemas/settings.schema.json"
  }
}

Neovim¶

With the JSON language server (jsonls, from vscode-json-languageserver) via nvim-lspconfig:

require('lspconfig').jsonls.setup({
  settings = {
    json = {
      schemas = {
        { fileMatch = { 'hosts.json' }, url = './schemas/hosts.schema.json' },
        { fileMatch = { 'reservations.json' }, url = './schemas/reservations.schema.json' },
      },
    },
  },
})

For settings.toml, the taplo language server honours schema directives. Either add a directive at the top of the file:

#:schema ./schemas/settings.schema.json

or associate it in the taplo config (.taplo.toml):

[[rule]]
include = ["settings.toml"]
[rule.schema]
path = "schemas/settings.schema.json"

Note on drift¶

The schemas reflect the otto version that generated them. There is no committed copy in the otto repo — regenerate with otto schema export whenever you upgrade so the fields stay in sync with your installed models.