Configuration

opx reads .opx/config.json from the repository root. opx init and opx enable write it; you can edit it by hand but you don't have to.

{
  "$schema": "https://unpkg.com/@allons-y/opx/schema/config.json",
  "version": 1,
  "lint": { "js": true, "json": true, "md": true },
  "release": true,
  "test": false
}

Keys

Key	Type	Description
`$schema`	string	Optional. Enables editor validation and completion.
`version`	`1`	Required. The config format version.
`lint`	object	Lint plugins, keyed by short name
`build`	object	Build plugins, keyed by short name
`release`	plugin setting	Enables the release plugin
`test`	plugin setting	Reserved for a future test plugin

Short names must be lowercase letters and digits (js, md).

Plugin settings

Wherever a plugin is configured, the value can be any one of:

Value	Meaning
`true`	Enabled with defaults
`false`	Disabled
`"./path.js"`	Enabled, using your own config file. Resolved from the repository root.
`{ ... }`	Enabled, with inline settings the plugin understands

Examples

Use your own ESLint flat config for JavaScript, and lint JSONC and JSON5 too:

{
  "version": 1,
  "lint": {
    "js": "./eslint.config.js",
    "json": { "dialects": ["jsonc", "json5"] }
  }
}

The settings each plugin understands are documented on its page: js, json, md, release.

Defaults & errors

If .opx/config.json doesn't exist, opx behaves as if every task were empty and release and test were false.
If it exists but isn't a valid JSON object, opx stops with an error that names the file.

State file

.opx/state.json is written by opx for its own bookkeeping. You shouldn't need to edit it, and it shouldn't be committed.

{
  "version": 1,
  "lastScanCommit": "a1b2c3d",
  "deferrals": {
    "md": { "branch": "main", "skipUntilDate": "2026-10-07", "reason": "later" }
  }
}