feat: add maxDelayMs setting to cap server-requested retry delays

When a provider (e.g., Google Gemini CLI) requests a retry delay longer
than maxDelayMs (default: 60s), the request fails immediately with an
informative error instead of waiting silently for hours.

The error is then handled by agent-level auto-retry, which shows the
delay to the user and allows aborting with Escape.

- Add maxRetryDelayMs to StreamOptions (packages/ai)
- Add maxRetryDelayMs to AgentOptions (packages/agent)
- Add retry.maxDelayMs to settings (packages/coding-agent)
- Update _isRetryableError to match 'retry delay' errors

fixes #1123

packages/ai/CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## [Unreleased]
 
+### Added
+
+- Added `maxRetryDelayMs` option to `StreamOptions` to cap server-requested retry delays. When a provider (e.g., Google Gemini CLI) requests a delay longer than this value, the request fails immediately with an informative error instead of waiting silently. Default: 60000ms (60 seconds). Set to 0 to disable the cap. ([#1123](https://github.com/badlogic/pi-mono/issues/1123))
+
 ## [0.50.7] - 2026-01-31
 
 ## [0.50.6] - 2026-01-30

packages/ai/src/providers/google-gemini-cli.ts

@@ -473,6 +473,16 @@ export const streamGoogleGeminiCli: StreamFunction<"google-gemini-cli", GoogleGe
 						// Use server-provided delay or exponential backoff
 						const serverDelay = extractRetryDelay(errorText, response);
 						const delayMs = serverDelay ?? BASE_DELAY_MS * 2 ** attempt;
+
+						// Check if server delay exceeds max allowed (default: 60s)
+						const maxDelayMs = options?.maxRetryDelayMs ?? 60000;
+						if (maxDelayMs > 0 && serverDelay && serverDelay > maxDelayMs) {
+							const delaySeconds = Math.ceil(serverDelay / 1000);
+							throw new Error(
+								`Server requested ${delaySeconds}s retry delay (max: ${Math.ceil(maxDelayMs / 1000)}s). ${extractErrorMessage(errorText)}`,
+							);
+						}
+
 						await sleep(delayMs, options?.signal);
 						continue;
 					}

packages/ai/src/providers/simple-options.ts

@@ -9,6 +9,7 @@ export function buildBaseOptions(model: Model<Api>, options?: SimpleStreamOption
 		sessionId: options?.sessionId,
 		headers: options?.headers,
 		onPayload: options?.onPayload,
+		maxRetryDelayMs: options?.maxRetryDelayMs,
 	};
 }
 

packages/ai/src/types.ts

@@ -72,6 +72,14 @@ export interface StreamOptions {
 	 * Not supported by all providers (e.g., AWS Bedrock uses SDK auth).
 	 */
 	headers?: Record<string, string>;
+	/**
+	 * Maximum delay in milliseconds to wait for a retry when the server requests a long wait.
+	 * If the server's requested delay exceeds this value, the request fails immediately
+	 * with an error containing the requested delay, allowing higher-level retry logic
+	 * to handle it with user visibility.
+	 * Default: 60000 (60 seconds). Set to 0 to disable the cap.
+	 */
+	maxRetryDelayMs?: number;
 }
 
 export type ProviderStreamOptions = StreamOptions & Record<string, unknown>;