Temporarily using simple rate limit error handler..

package.json

@@ -2,7 +2,7 @@
   "name": "claude-dev-unleashed",
   "displayName": "Cline Unleashed⛓️‍💥 (prev. Claude Dev)",
   "description": "Autonomous coding agent right in your IDE, capable of creating/editing files, running commands, using the browser, and more with your permission every step of the way.",
-  "version": "2.1.6",
+  "version": "2.1.6-ratelimit",
   "icon": "assets/icons/icon.png",
   "galleryBanner": {
     "color": "#617A91",
@@ -152,9 +152,9 @@
     "typescript": "^5.4.5"
   },
   "dependencies": {
-    "@anthropic-ai/bedrock-sdk": "^0.10.2",
-    "@anthropic-ai/sdk": "^0.26.0",
-    "@anthropic-ai/vertex-sdk": "^0.4.1",
+    "@anthropic-ai/bedrock-sdk": "^0.11.2",
+    "@anthropic-ai/sdk": "^0.32.1",
+    "@anthropic-ai/vertex-sdk": "^0.5.2",
     "@google/generative-ai": "^0.18.0",
     "@types/clone-deep": "^4.0.4",
     "@types/pdf-parse": "^1.1.4",

src/api/providers/anthropic.ts

@@ -40,34 +40,40 @@ export class AnthropicHandler implements ApiHandler {
 				)
 				const lastUserMsgIndex = userMsgIndices[userMsgIndices.length - 1] ?? -1
 				const secondLastMsgUserIndex = userMsgIndices[userMsgIndices.length - 2] ?? -1
+
+				const outgoingMessages:Array<any> =  messages.map((message, index) => {
+					if (index === lastUserMsgIndex || index === secondLastMsgUserIndex) {
+						return {
+							...message,
+							content:
+								typeof message.content === "string"
+									? [
+											{
+												type: "text",
+												text: message.content,
+												cache_control: { type: "ephemeral" },
+											},
+									  ]
+									: message.content.map((content, contentIndex) =>
+											contentIndex === message.content.length - 1
+												? { ...content, cache_control: { type: "ephemeral" } }
+												: content
+									  ),
+						}
+					}
+					return message
+				});
+
+				const currentTime = new Date().toLocaleTimeString([], { hour: '2-digit', minute: '2-digit', second: '2-digit' });
+				console.log(`[${currentTime}] 🤖 Sending ${outgoingMessages.length} messages to the prompt caching API`);
+
 				stream = await this.client.beta.promptCaching.messages.create(
 					{
 						model: modelId,
 						max_tokens: this.getModel().info.maxTokens || 8192,
 						temperature: 0,
 						system: [{ text: systemPrompt, type: "text", cache_control: { type: "ephemeral" } }], // setting cache breakpoint for system prompt so new tasks can reuse it
-						messages: messages.map((message, index) => {
-							if (index === lastUserMsgIndex || index === secondLastMsgUserIndex) {
-								return {
-									...message,
-									content:
-										typeof message.content === "string"
-											? [
-													{
-														type: "text",
-														text: message.content,
-														cache_control: { type: "ephemeral" },
-													},
-											  ]
-											: message.content.map((content, contentIndex) =>
-													contentIndex === message.content.length - 1
-														? { ...content, cache_control: { type: "ephemeral" } }
-														: content
-											  ),
-								}
-							}
-							return message
-						}),
+						messages: outgoingMessages,
 						// tools, // cache breakpoints go from tools > system > messages, and since tools dont change, we can just set the breakpoint at the end of system (this avoids having to set a breakpoint at the end of tools which by itself does not meet min requirements for haiku caching)
 						// tool_choice: { type: "auto" },
 						// tools: tools,
@@ -93,6 +99,8 @@ export class AnthropicHandler implements ApiHandler {
 				break
 			}
 			default: {
+				console.log(`🤖 Sending ${messages.length} to the API`, messages)
+
 				stream = (await this.client.messages.create({
 					model: modelId,
 					max_tokens: this.getModel().info.maxTokens || 8192,

src/core/Cline.ts

@@ -776,12 +776,26 @@ export class Cline {
 
 		const stream = this.api.createMessage(systemPrompt, this.apiConversationHistory)
 		const iterator = stream[Symbol.asyncIterator]()
+		let rateLimitsErrors = 0;
+		let rateLimitDelay = 30;
 
 		try {
 			// awaiting first chunk to see if it will throw an error
 			const firstChunk = await iterator.next()
 			yield firstChunk.value
 		} catch (error) {
+
+			if (error.status === 429 && rateLimitsErrors <= 3) {
+
+				const currentTime = new Date().toLocaleTimeString([], { hour: '2-digit', minute: '2-digit', second: '2-digit' });
+				console.log(`[${currentTime}] 🤖 Rate limited, waiting ${rateLimitDelay} seconds to retry`, error);
+				await delay(rateLimitDelay * 1000);
+				rateLimitsErrors++;
+				rateLimitDelay *= 2;
+				yield* this.attemptApiRequest(previousApiReqIndex)
+				return
+			}
+
 			// note that this api_req_failed ask is unique in that we only present this option if the api hasn't streamed any content yet (ie it fails on the first chunk due), as it would allow them to hit a retry button. However if the api failed mid-stream, it could be in any arbitrary state where some tools may have executed, so that error is handled differently and requires cancelling the task entirely.
 			const { response } = await this.ask(
 				"api_req_failed",