JSON Schema for `jarvy.toml`¶

Jarvy publishes a JSON Schema for jarvy.toml. Wire it into your editor and you get inline autocomplete, hover descriptions, and live validation while you edit.

Schema URL:

https://jarvy.dev/schema/jarvy.schema.json

VS Code¶

Install the Even Better TOML extension, then add to your settings:

.vscode/settings.json

{
  "evenBetterToml.schema.associations": {
    "jarvy.toml": "https://jarvy.dev/schema/jarvy.schema.json"
  }
}

Or use the inline schema directive at the top of jarvy.toml (works without any settings):

jarvy.toml

#:schema https://jarvy.dev/schema/jarvy.schema.json

[provisioner]
node = "20"

You'll get:

Autocomplete for [provisioner], [hooks], [env.vars], every block
Inline hover hints with the description from the schema
Squiggly underlines on unknown sections, invalid version syntax, missing required fields

JetBrains (IntelliJ, WebStorm, GoLand, RustRover, …)¶

JetBrains has built-in JSON Schema support for TOML. Add the schema in Settings → Languages & Frameworks → Schemas and DTDs → JSON Schema Mappings:

Field	Value
Name	`jarvy.toml`
Schema file or URL	`https://jarvy.dev/schema/jarvy.schema.json`
Schema version	JSON Schema version 7
File path pattern	`jarvy.toml`

Or per-project in .idea/jsonSchemas.xml if you want to commit the binding.

Helix / Neovim (LSP)¶

Use taplo as a TOML language server with the schema:

.taplo.toml

[[schema.rule]]
url = "https://jarvy.dev/schema/jarvy.schema.json"
formats = ["toml"]
include = ["**/jarvy.toml"]

Then start your editor with taplo configured as the TOML LSP. You'll get hover, completion, and diagnostics through any LSP-aware editor.

Offline / pinned versions¶

If your team is air-gapped or wants to pin to a specific Jarvy version:

curl -fsSL https://jarvy.dev/schema/jarvy.schema.json -o .jarvy/jarvy.schema.json

Then point the schema URL at the local file:

#:schema ./.jarvy/jarvy.schema.json

Commit .jarvy/jarvy.schema.json and refresh on Jarvy upgrades.

What the schema covers¶

The published schema describes every TOML section Jarvy accepts:

[provisioner] — tool entries with version specs
[privileges] — sudo behavior
[hooks] — global + per-tool hooks
[env.vars], [env.secrets] — environment
[services] — Docker Compose / Tilt integration
[roles.*] — role definitions and inheritance
[npm], [pip], [cargo] — language packages
[git] — Git configuration
[network] — proxy + TLS
[drift] — baseline + version policy
[telemetry], [update], [logging], [commands]

Each field has a description, type, default value, and (where applicable) an enum constraint.

What's not covered¶

The schema validates shape, not semantics:

It accepts any string for a tool name. jarvy validate checks the registry; the schema can't.
Version strings are validated by length, not by parsing. jarvy validate does the SemVer check.
Role inheritance cycles are runtime checks, not schema rules.

So the editor catches typos and unknown fields; jarvy validate catches everything else. Run both.

Reporting schema issues¶

The schema is hand-maintained alongside the Rust config struct. If you see a field accepted by jarvy validate that the schema rejects (or vice versa), open an issue — schema drift is a bug.

JSON Schema for jarvy.toml¶