All jobs

Add User Settings to a VS Code Extension

Yes. An agent adds a contributes.configuration section to your VS Code extension, reads the settings with the workspace API, reacts to changes and opens a PR.

Run this on RushOpens the console with this prompt ready to send, in a Project that has your repository.

What the agent does

The agent reads the extension and the configuration sample that ships beside it. In an extension like helloworld-sample in microsoft/vscode-extension-samples, package.json declares engines.vscode ^1.100.0, one contributed command extension.helloWorld titled Hello World, an empty activationEvents list because VS Code infers activation from the command, main at ./out/extension.js, and scripts compile (tsc -p ./), watch and lint (eslint 9 with typescript-eslint). tsconfig.json compiles to CommonJS with strict on, and src/extension.ts registers the command and calls vscode.window.showInformationMessage('Hello World!') with the greeting as a literal. The same repository's configuration-sample shows the target shape: a contributes.configuration array with an id, a title, an order and typed properties, each with a type, a default, a description and a scope of window, resource or language-overridable; an enum setting with a default; an object setting with additionalProperties: false; code that reads with vscode.workspace.getConfiguration().get(...), writes with .update(key, value, vscode.ConfigurationTarget.Global) or Workspace, and listens with onDidChangeConfiguration and e.affectsConfiguration(...).

From that the agent adds a helloworld configuration section to helloworld-sample's manifest with a title for the Settings editor and a small set of properties the prompt names: helloworld.greeting, a string whose default is the current literal Hello World! so behaviour does not change for anyone who has not touched settings; helloworld.showAsModal, a boolean defaulting to false that switches the message to a modal dialog; and helloworld.enabled, a boolean that lets the command be turned off. Each has a description or a markdownDescription where a link helps, and the scope is window, because a greeting cannot sensibly vary per folder; the pull request explains when resource would be the right choice instead.

In src/extension.ts the literal becomes a read of vscode.workspace.getConfiguration('helloworld') with typed getters, get<string>('greeting', 'Hello World!'), taken at command time rather than cached, so a changed setting applies on the next invocation without a reload. The showInformationMessage call gains the modal option from the boolean. Following configuration-sample, a second command, extension.configureGreeting, asks with showInputBox and writes the value with update to the Global target, or offers User and Workspace through showQuickPick when a workspace is open. An onDidChangeConfiguration listener filtered with affectsConfiguration('helloworld') is pushed to context.subscriptions, as the existing command disposable already is, so it is disposed on deactivation.

The checks are the repository's own scripts: npm run compile and npm run lint pass with the stricter typescript-eslint rules the sample configures, including curly and the semicolon rule. The README gains a settings table with each key, type, default and meaning, matching how configuration-sample's README documents its manual testing steps. A headless extension test needs the @vscode/test-cli setup that helloworld-test-cli-sample in the same repository shows, plus xvfb-run on Linux, so the agent adds one only when the prompt asks and says so otherwise.

Start it with one API call

POST https://api.phoenix.so/o/{org}/sessions with this body:

{
  "agent": "claude",
  "input": "Add user settings to this VS Code extension. Read package.json, src/extension.ts, tsconfig.json, eslint.config.mjs and the README first, and use the configuration-sample in this repository as the model. Add a contributes.configuration section with a title and three properties: a string greeting whose default is the message the extension shows today, a boolean that shows the message as a modal dialog, and a boolean that disables the command. Give each a type, default, description and window scope. Read the settings at command time with workspace.getConfiguration and typed getters, add a command that updates the greeting through showInputBox and update with a configuration target picker when a workspace is open, and register an onDidChangeConfiguration listener filtered with affectsConfiguration that is pushed to context.subscriptions. Run npm run compile and npm run lint, document the settings in a README table, and open a pull request. Do not add a test harness unless asked; say so in the pull request.",
  "resources": [
    {
      "url": "https://github.com/your-org/your-app"
    }
  ]
}
  1. Get an API key. In the console, open Settings, then API keys, and create a key for your org. Send it in the Authorization header as a Bearer token.
  2. Send the request. POST the body in request.json to https://api.phoenix.so/o/{org}/sessions, with your repository as the resources URL. The run starts in its own cloud sandbox.
  3. Read what came back. Follow the run on its live stream or wait for the run.completed webhook, then read the result and its trajectory.

What comes back

What comes back is a pull request on its own branch with package.json (the contributes.configuration section and the second command under contributes.commands), the edited src/extension.ts, and the README table. tsconfig.json, eslint.config.mjs and the lockfile are untouched, because nothing new is installed.

The description lists each setting with its key, type, default and scope and why that scope, shows the compile and lint output, and names the manual steps a reviewer follows in the Extension Development Host: change helloworld.greeting in Settings, run the command, see the new text. The commands are in the transcript at GET /o/{org}/sessions/{id}/messages.

Before merging, a reviewer presses F5 with the repository's existing Run Extension launch configuration, tries the settings in both User and Workspace scope, and decides whether the settings keys match the publisher's naming convention for other extensions. A follow-up message to the same session adds a fourth setting with the same pattern.

Run it on every Linear issue labelled settings

A team that tracks extension requests in Linear starts one run per issue from an automation with a linear trigger on a settings label. The automation adds the issue after this prompt, so the setting a user asked for, with its name and default from the issue, is what the agent adds, and the run starts in the repository the automation names.

Each run opens its own pull request from its own sandbox, and an automation run has its own max_runs_per_hour cap. The pull request links the issue and lists the keys it added, so review is a read of the manifest section and the getter calls.

The use case this job follows, with its automation and trigger: Turn a Linear label into a PR.

Why run it on Rush?

Settings touch three places that have to agree: the manifest schema, the code that reads it and the documentation. A key spelled differently in any one of them fails quietly, and the only test is to launch VS Code. The agent uses the repository's own sample as the pattern, keeps the default equal to today's behaviour, and runs the compiler and the linter in its own sandbox so the type of each getter matches the schema before anyone opens the Extension Development Host.

The pull request carries that proof in the transcript: compile and lint output sit beside the commands that produced them, and the limit, no headless UI test unless asked, is stated. Several extensions can be handled at once, each in its own sandbox, with one pull request each.

Frequently asked questions

Will existing users see a change?

No. The greeting setting defaults to the exact string the extension shows today, the modal option defaults to off and the command stays enabled, so an installation with no settings behaves as before. Only someone who changes a value in Settings sees anything new.

Why window scope and not resource?

A resource setting can differ per workspace folder and must be read with the document's URI, as configuration-sample does for its empty-last-line setting. A greeting has no per-folder meaning, so window is correct and simpler. The pull request explains the choice and when the other scope applies.

Does the setting apply without reloading the window?

Yes. The command reads the configuration each time it runs instead of caching a value at activation, and the onDidChangeConfiguration listener is there for any future state that does need refreshing. No reload prompt is needed.

Can the agent add a test for the settings?

It can, using the @vscode/test-cli setup the helloworld-test-cli-sample in the same repository shows, which downloads VS Code and runs Mocha inside it under xvfb-run on Linux. That is a separate piece of setup, so the agent does it only when the prompt asks and otherwise relies on compile, lint and the manual steps.

More VS Code Extension jobs

More jobs like this: Turn a Linear label into a PR

Keep reading