Skip to content

Reference

Open VS Code settings (Ctrl+, or Cmd+, on macOS) and search for Prompty to find all extension settings.

SettingTypeDefaultDescription
prompty.currentModelConfigurationstring"default"The active model configuration name
prompty.showV1DeprecationWarningsbooleantrueShow deprecation warnings when loading .prompty files that use the legacy v1 format
prompty.envFilePathstring""Custom path to a .env file. If empty, the extension searches the workspace hierarchy
promptyLanguageServer.trace.serverstring"off"Trace level for VS Code ↔ language server communication. Values: off, messages, verbose

Example settings.json:

{
"prompty.currentModelConfiguration": "my-azure-connection",
"prompty.showV1DeprecationWarnings": true,
"prompty.envFilePath": "${workspaceFolder}/.env",
"promptyLanguageServer.trace.server": "off"
}
ActionShortcutContext
Run PromptF5When a .prompty file is active
New PromptyCtrl+Shift+P → “Prompty: New Prompty”Any time
PreviewCtrl+Shift+P → “Prompty: Preview”When a .prompty file is active
Add ConnectionCtrl+Shift+P → “Prompty: Add Connection”Any time

The complete list of commands registered by the extension:

CommandDescription
Prompty: New PromptyCreate a new .prompty file
Prompty: Run PromptExecute the current .prompty file
Prompty: PreviewOpen live preview panel
Prompty: Add ConnectionAdd a new LLM provider connection
Prompty: Edit ConnectionModify an existing connection
Prompty: Delete ConnectionRemove a connection
Prompty: Test ConnectionTest connectivity to a provider
Prompty: Set Default ConnectionSet a connection as the default for its provider
Prompty: Refresh ConnectionsReload connections from disk
Prompty: Refresh ModelsRe-fetch available models for a connection
Prompty: View TraceOpen a trace file in the viewer
Prompty: Refresh TracesReload the traces list
Prompty: Sort Traces by DateSort traces newest first
Prompty: Sort Traces by NameSort traces alphabetically

Other VS Code extensions can integrate with Prompty by activating the extension and using the returned API surface.

const promptyExt = vscode.extensions.getExtension('ms-toolsai.prompty');
const api = await promptyExt?.activate();
MethodDescription
registerConnectionProvider(provider)Register a custom connection provider
registerExecutor(key, executor)Register a custom executor for a provider key
registerProcessor(key, processor)Register a custom processor for a provider key
getConnections()Query all configured connections
onConnectionsChanged(listener)Listen for connection changes
  • OpenAI / Anthropic — verify that your API key is correct and has not expired. Run Prompty: Test Connection to see the exact error.
  • Microsoft Foundry — ensure you are signed into Azure (Azure: Sign In in the Command Palette). Double-check that the endpoint URL matches your deployed resource.

The language server provides completions for frontmatter keys and values. If autocomplete stops working:

  1. Run Developer: Reload Window from the Command Palette.
  2. If the issue persists, the language server may need a restart — disable and re-enable the extension.
  1. Open the Output panel (Ctrl+Shift+U) and select the Prompty channel. Look for errors during execution.
  2. Verify that a .runs/ directory exists in your workspace — traces are written there as .tracy files.
  3. Click Refresh in the Traces view, or run Prompty: Refresh Traces.

If you see deprecation warnings when opening a .prompty file, the file uses the legacy v1 frontmatter format. See the migration guide for instructions on updating to the v2 format.

To suppress warnings without migrating, set:

{
"prompty.showV1DeprecationWarnings": false
}
  1. Set promptyLanguageServer.trace.server to "verbose" in settings.
  2. Open the Output panel and select Prompty Language Server.
  3. Look for startup errors or missing dependencies.
  4. If no output appears, try reinstalling the extension from the Marketplace.