Merge branch 'main' into rb/github-patch

This reverts commit 79492b6551.

Development.md

@@ -100,7 +100,7 @@ poetry run pytest ./tests/unit/test_*.py
 ### 9. Use existing Docker image
 To reduce build time (e.g., if no changes were made to the client-runtime component), you can use an existing Docker container image. Follow these steps:
 1. Set the SANDBOX_RUNTIME_CONTAINER_IMAGE environment variable to the desired Docker image.
-2. Example: export SANDBOX_RUNTIME_CONTAINER_IMAGE=ghcr.io/all-hands-ai/runtime:0.12-nikolaik
+2. Example: export SANDBOX_RUNTIME_CONTAINER_IMAGE=ghcr.io/all-hands-ai/runtime:0.13-nikolaik
 
 ## Develop inside Docker container
 

README.md

@@ -38,15 +38,16 @@ See the [Installation](https://docs.all-hands.dev/modules/usage/installation) gu
 system requirements and more information.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik
 
 docker run -it --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -p 3000:3000 \
+    -e LOG_ALL_EVENTS=true \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.12
+    docker.all-hands.dev/all-hands-ai/openhands:0.13
 ```
 
 You'll find OpenHands running at [http://localhost:3000](http://localhost:3000)!

compose.yml

@@ -7,7 +7,7 @@ services:
     image: openhands:latest
     container_name: openhands-app-${DATE:-}
     environment:
-      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.12-nikolaik}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.13-nikolaik}
       - SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234}
       - WORKSPACE_MOUNT_PATH=${WORKSPACE_BASE:-$PWD/workspace}
     ports:

containers/dev/compose.yml

@@ -11,7 +11,7 @@ services:
       - BACKEND_HOST=${BACKEND_HOST:-"0.0.0.0"}
       - SANDBOX_API_HOSTNAME=host.docker.internal
       #
-      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.12-nikolaik}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.13-nikolaik}
       - SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234}
       - WORKSPACE_MOUNT_PATH=${WORKSPACE_BASE:-$PWD/workspace}
     ports:

docs/modules/usage/how-to/cli-mode.md

@@ -50,7 +50,7 @@ LLM_API_KEY="sk_test_12345"
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -59,7 +59,7 @@ docker run -it \
     -v /var/run/docker.sock:/var/run/docker.sock \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.12 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.13 \
     python -m openhands.core.cli
 ```
 

docs/modules/usage/how-to/headless-mode.md

@@ -44,15 +44,16 @@ LLM_API_KEY="sk_test_12345"
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
     -e LLM_MODEL=$LLM_MODEL \
+    -e LOG_ALL_EVENTS=true \
     -v $WORKSPACE_BASE:/opt/workspace_base \
     -v /var/run/docker.sock:/var/run/docker.sock \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.12 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.13 \
     python -m openhands.core.main -t "write a bash script that prints hi"
 ```

docs/modules/usage/installation.mdx

@@ -11,15 +11,16 @@
 The easiest way to run OpenHands is in Docker.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -p 3000:3000 \
+    -e LOG_ALL_EVENTS=true \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.12
+    docker.all-hands.dev/all-hands-ai/openhands:0.13
 ```
 
 You can also run OpenHands in a scriptable [headless mode](https://docs.all-hands.dev/modules/usage/how-to/headless-mode), as an [interactive CLI](https://docs.all-hands.dev/modules/usage/how-to/cli-mode), or using the [OpenHands GitHub Action](https://docs.all-hands.dev/modules/usage/how-to/github-action).

docs/modules/usage/llms/llms.md

@@ -4,11 +4,11 @@ OpenHands can connect to any LLM supported by LiteLLM. However, it requires a po
 
 ## Model Recommendations
 
-Based on a recent evaluation of language models for coding tasks (using the SWE-bench dataset), we can provide some recommendations for model selection. The full analysis can be found in [this blog article](https://www.all-hands.dev/blog/evaluation-of-llms-as-coding-agents-on-swe-bench-at-30x-speed).
+Based on our evaluations of language models for coding tasks (using the SWE-bench dataset), we can provide some recommendations for model selection. Some analyses can be found in [this blog article comparing LLMs](https://www.all-hands.dev/blog/evaluation-of-llms-as-coding-agents-on-swe-bench-at-30x-speed) and [this blog article with some more recent results](https://www.all-hands.dev/blog/openhands-codeact-21-an-open-state-of-the-art-software-development-agent).
 
 When choosing a model, consider both the quality of outputs and the associated costs. Here's a summary of the findings:
 
-- Claude 3.5 Sonnet is the best by a fair amount, achieving a 27% resolve rate with the default agent in OpenHands.
+- Claude 3.5 Sonnet is the best by a fair amount, achieving a 53% resolve rate on SWE-Bench Verified with the default agent in OpenHands.
 - GPT-4o lags behind, and o1-mini actually performed somewhat worse than GPT-4o. We went in and analyzed the results a little, and briefly it seemed like o1 was sometimes "overthinking" things, performing extra environment configuration tasks when it could just go ahead and finish the task.
 - Finally, the strongest open models were Llama 3.1 405 B and deepseek-v2.5, and they performed reasonably, even besting some of the closed models.
 

evaluation/EDA/run_infer.py

@@ -35,7 +35,8 @@ def codeact_user_response_eda(state: State) -> str:
 
     # retrieve the latest model message from history
     if state.history:
-        model_guess = state.get_last_agent_message()
+        last_agent_message = state.get_last_agent_message()
+        model_guess = last_agent_message.content if last_agent_message else ''
 
     assert game is not None, 'Game is not initialized.'
     msg = game.generate_user_response(model_guess)
@@ -140,7 +141,8 @@ def process_instance(
     if state is None:
         raise ValueError('State should not be None.')
 
-    final_message = state.get_last_agent_message()
+    last_agent_message = state.get_last_agent_message()
+    final_message = last_agent_message.content if last_agent_message else ''
 
     logger.info(f'Final message: {final_message} | Ground truth: {instance["text"]}')
     test_result = game.reward()

evaluation/gorilla/run_infer.py

@@ -102,7 +102,8 @@ def process_instance(
         raise ValueError('State should not be None.')
 
     # retrieve the last message from the agent
-    model_answer_raw = state.get_last_agent_message()
+    last_agent_message = state.get_last_agent_message()
+    model_answer_raw = last_agent_message.content if last_agent_message else ''
 
     # attempt to parse model_answer
     ast_eval_fn = instance['ast_eval']

evaluation/swe_bench/eval_infer.py

@@ -83,6 +83,7 @@ def get_config(instance: pd.Series) -> AppConfig:
             timeout=1800,
             api_key=os.environ.get('ALLHANDS_API_KEY', None),
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
+            remote_runtime_init_timeout=1800,
         ),
         # do not mount workspace
         workspace_base=None,

evaluation/swe_bench/run_infer.py

@@ -146,6 +146,7 @@ def get_config(
             api_key=os.environ.get('ALLHANDS_API_KEY', None),
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
             keep_remote_runtime_alive=False,
+            remote_runtime_init_timeout=1800,
         ),
         # do not mount workspace
         workspace_base=None,

evaluation/toolqa/run_infer.py

@@ -127,7 +127,8 @@ def process_instance(instance: Any, metadata: EvalMetadata, reset_logger: bool =
         raise ValueError('State should not be None.')
 
     # retrieve the last message from the agent
-    model_answer_raw = state.get_last_agent_message()
+    last_agent_message = state.get_last_agent_message()
+    model_answer_raw = last_agent_message.content if last_agent_message else ''
 
     # attempt to parse model_answer
     correct = eval_answer(str(model_answer_raw), str(answer))

frontend/__tests__/components/chat/chat-interface.test.tsx

@@ -16,14 +16,14 @@ describe("Empty state", () => {
     send: vi.fn(),
   }));
 
-  const { useSocket: useSocketMock } = vi.hoisted(() => ({
+  const { useWsClient: useWsClientMock } = vi.hoisted(() => ({
-    useSocket: vi.fn(() => ({ send: sendMock, runtimeActive: true })),
+    useWsClient: vi.fn(() => ({ send: sendMock, runtimeActive: true })),
   }));
 
   beforeAll(() => {
     vi.mock("#/context/socket", async (importActual) => ({
-      ...(await importActual<typeof import("#/context/socket")>()),
+      ...(await importActual<typeof import("#/context/ws-client-provider")>()),
-      useSocket: useSocketMock,
+      useWsClient: useWsClientMock,
     }));
   });
 
@@ -77,7 +77,7 @@ describe("Empty state", () => {
     "should load the a user message to the input when selecting",
     async () => {
       // this is to test that the message is in the UI before the socket is called
-      useSocketMock.mockImplementation(() => ({
+      useWsClientMock.mockImplementation(() => ({
         send: sendMock,
         runtimeActive: false, // mock an inactive runtime setup
       }));
@@ -106,7 +106,7 @@ describe("Empty state", () => {
   it.fails(
     "should send the message to the socket only if the runtime is active",
     async () => {
-      useSocketMock.mockImplementation(() => ({
+      useWsClientMock.mockImplementation(() => ({
         send: sendMock,
         runtimeActive: false, // mock an inactive runtime setup
       }));
@@ -123,7 +123,7 @@ describe("Empty state", () => {
       await user.click(displayedSuggestions[0]);
       expect(sendMock).not.toHaveBeenCalled();
 
-      useSocketMock.mockImplementation(() => ({
+      useWsClientMock.mockImplementation(() => ({
         send: sendMock,
         runtimeActive: true, // mock an active runtime setup
       }));

frontend/__tests__/hooks/use-terminal.test.tsx

@@ -2,8 +2,9 @@ import { beforeAll, describe, expect, it, vi } from "vitest";
 import { render } from "@testing-library/react";
 import { afterEach } from "node:test";
 import { useTerminal } from "#/hooks/useTerminal";
-import { SocketProvider } from "#/context/socket";
 import { Command } from "#/state/commandSlice";
+import { WsClientProvider } from "#/context/ws-client-provider";
+import { ReactNode } from "react";
 
 interface TestTerminalComponentProps {
   commands: Command[];
@@ -18,6 +19,17 @@ function TestTerminalComponent({
   return <div ref={ref} />;
 }
 
+interface WrapperProps {
+  children: ReactNode;
+}
+
+
+function Wrapper({children}: WrapperProps) {
+  return (
+    <WsClientProvider enabled={true} token="NO_JWT" ghToken="NO_GITHUB" settings={null}>{children}</WsClientProvider>
+  )
+}
+
 describe("useTerminal", () => {
   const mockTerminal = vi.hoisted(() => ({
     loadAddon: vi.fn(),
@@ -50,7 +62,7 @@ describe("useTerminal", () => {
 
   it("should render", () => {
     render(<TestTerminalComponent commands={[]} secrets={[]} />, {
-      wrapper: SocketProvider,
+      wrapper: Wrapper,
     });
   });
 
@@ -61,7 +73,7 @@ describe("useTerminal", () => {
     ];
 
     render(<TestTerminalComponent commands={commands} secrets={[]} />, {
-      wrapper: SocketProvider,
+      wrapper: Wrapper,
     });
 
     expect(mockTerminal.writeln).toHaveBeenNthCalledWith(1, "echo hello");
@@ -85,7 +97,7 @@ describe("useTerminal", () => {
         secrets={[secret, anotherSecret]}
       />,
       {
-        wrapper: SocketProvider,
+        wrapper: Wrapper,
       },
     );
 

frontend/__tests__/utils/cache.test.ts

@@ -8,7 +8,6 @@ describe("Cache", () => {
   const testTTL = 1000; // 1 second
 
   beforeEach(() => {
-    localStorage.clear();
     vi.useFakeTimers();
   });
 
@@ -16,17 +15,7 @@ describe("Cache", () => {
     vi.useRealTimers();
   });
 
-  it("sets data in localStorage with expiration", () => {
+  it("gets data from memory if not expired", () => {
-    cache.set(testKey, testData, testTTL);
-    const cachedEntry = JSON.parse(
-      localStorage.getItem(`app_cache_${testKey}`) || "",
-    );
-
-    expect(cachedEntry.data).toEqual(testData);
-    expect(cachedEntry.expiration).toBeGreaterThan(Date.now());
-  });
-
-  it("gets data from localStorage if not expired", () => {
     cache.set(testKey, testData, testTTL);
 
     expect(cache.get(testKey)).toEqual(testData);
@@ -39,7 +28,6 @@ describe("Cache", () => {
     vi.advanceTimersByTime(5 * 60 * 1000 + 1);
 
     expect(cache.get(testKey)).toBeNull();
-    expect(localStorage.getItem(`app_cache_${testKey}`)).toBeNull();
   });
 
   it("returns null if cached data is expired", () => {
@@ -47,28 +35,19 @@ describe("Cache", () => {
 
     vi.advanceTimersByTime(testTTL + 1);
     expect(cache.get(testKey)).toBeNull();
-    expect(localStorage.getItem(`app_cache_${testKey}`)).toBeNull();
   });
 
-  it("deletes data from localStorage", () => {
+  it("deletes data from memory", () => {
     cache.set(testKey, testData, testTTL);
     cache.delete(testKey);
-
+    expect(cache.get(testKey)).toBeNull();
-    expect(localStorage.getItem(`app_cache_${testKey}`)).toBeNull();
   });
 
-  it("clears all data with the app prefix from localStorage", () => {
+  it("clears all data with the app prefix from memory", () => {
     cache.set(testKey, testData, testTTL);
     cache.set("anotherKey", { data: "More data" }, testTTL);
     cache.clearAll();
-
+    expect(cache.get(testKey)).toBeNull();
-    expect(localStorage.length).toBe(0);
+    expect(cache.get("anotherKey")).toBeNull();
-  });
-
-  it("does not retrieve non-prefixed data from localStorage when clearing", () => {
-    localStorage.setItem("nonPrefixedKey", "should remain");
-    cache.set(testKey, testData, testTTL);
-    cache.clearAll();
-    expect(localStorage.getItem("nonPrefixedKey")).toBe("should remain");
   });
 });

frontend/__tests__/utils/extractModelAndProvider.test.ts

@@ -59,9 +59,9 @@ describe("extractModelAndProvider", () => {
       separator: "/",
     });
 
-    expect(extractModelAndProvider("claude-3-5-sonnet-20241022")).toEqual({
+    expect(extractModelAndProvider("claude-3-5-sonnet-20240620")).toEqual({
       provider: "anthropic",
-      model: "claude-3-5-sonnet-20241022",
+      model: "claude-3-5-sonnet-20240620",
       separator: "/",
     });
 

frontend/__tests__/utils/organizeModelsAndProviders.test.ts

@@ -15,7 +15,7 @@ test("organizeModelsAndProviders", () => {
     "gpt-4o",
     "together-ai-21.1b-41b",
     "gpt-4o-mini",
-    "claude-3-5-sonnet-20241022",
+    "anthropic/claude-3-5-sonnet-20241022",
     "claude-3-haiku-20240307",
     "claude-2",
     "claude-2.1",

frontend/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "openhands-frontend",
-  "version": "0.12.3",
+  "version": "0.13.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "openhands-frontend",
-      "version": "0.12.3",
+      "version": "0.13.0",
       "dependencies": {
         "@monaco-editor/react": "^4.6.0",
         "@nextui-org/react": "^2.4.8",

frontend/package.json

@@ -1,6 +1,6 @@
 {
   "name": "openhands-frontend",
-  "version": "0.12.3",
+  "version": "0.13.0",
   "private": true,
   "type": "module",
   "engines": {
@@ -120,4 +120,4 @@
       "public"
     ]
   }
-}
+}

frontend/src/components/AgentControlBar.tsx

@@ -6,7 +6,7 @@ import PlayIcon from "#/assets/play";
 import { generateAgentStateChangeEvent } from "#/services/agentStateService";
 import { RootState } from "#/store";
 import AgentState from "#/types/AgentState";
-import { useSocket } from "#/context/socket";
+import { useWsClient } from "#/context/ws-client-provider";
 
 const IgnoreTaskStateMap: Record<string, AgentState[]> = {
   [AgentState.PAUSED]: [
@@ -72,7 +72,7 @@ function ActionButton({
 }
 
 function AgentControlBar() {
-  const { send } = useSocket();
+  const { send } = useWsClient();
   const { curAgentState } = useSelector((state: RootState) => state.agent);
 
   const handleAction = (action: AgentState) => {

frontend/src/components/attach-image-label.tsx

@@ -1,4 +1,4 @@
-import Clip from "#/assets/clip.svg?react";
+import Clip from "#/icons/clip.svg?react";
 
 export function AttachImageLabel() {
   return (

frontend/src/components/chat-input.tsx

@@ -1,6 +1,6 @@
 import React from "react";
 import TextareaAutosize from "react-textarea-autosize";
-import ArrowSendIcon from "#/assets/arrow-send.svg?react";
+import ArrowSendIcon from "#/icons/arrow-send.svg?react";
 import { cn } from "#/utils/utils";
 
 interface ChatInputProps {

frontend/src/components/chat-interface.tsx

@@ -1,7 +1,6 @@
 import { useDispatch, useSelector } from "react-redux";
 import React from "react";
 import posthog from "posthog-js";
-import { useSocket } from "#/context/socket";
 import { convertImageToBase64 } from "#/utils/convert-image-to-base-64";
 import { ChatMessage } from "./chat-message";
 import { FeedbackActions } from "./feedback-actions";
@@ -21,14 +20,15 @@ import { ContinueButton } from "./continue-button";
 import { ScrollToBottomButton } from "./scroll-to-bottom-button";
 import { Suggestions } from "./suggestions";
 import { SUGGESTIONS } from "#/utils/suggestions";
-import BuildIt from "#/assets/build-it.svg?react";
+import BuildIt from "#/icons/build-it.svg?react";
+import { useWsClient } from "#/context/ws-client-provider";
 
 const isErrorMessage = (
   message: Message | ErrorMessage,
 ): message is ErrorMessage => "error" in message;
 
 export function ChatInterface() {
-  const { send } = useSocket();
+  const { send } = useWsClient();
   const dispatch = useDispatch();
   const scrollRef = React.useRef<HTMLDivElement>(null);
   const { scrollDomToBottom, onChatBodyScroll, hitBottom } =

frontend/src/components/chat/ConfirmationButtons.tsx

@@ -5,7 +5,7 @@ import RejectIcon from "#/assets/reject";
 import { I18nKey } from "#/i18n/declaration";
 import AgentState from "#/types/AgentState";
 import { generateAgentStateChangeEvent } from "#/services/agentStateService";
-import { useSocket } from "#/context/socket";
+import { useWsClient } from "#/context/ws-client-provider";
 
 interface ActionTooltipProps {
   type: "confirm" | "reject";
@@ -37,7 +37,7 @@ function ActionTooltip({ type, onClick }: ActionTooltipProps) {
 
 function ConfirmationButtons() {
   const { t } = useTranslation();
-  const { send } = useSocket();
+  const { send } = useWsClient();
 
   const handleStateChange = (state: AgentState) => {
     const event = generateAgentStateChangeEvent(state);

frontend/src/components/event-handler.tsx

@@ -0,0 +1,188 @@
+import React from "react";
+import {
+  useFetcher,
+  useLoaderData,
+  useRouteLoaderData,
+} from "@remix-run/react";
+import { useDispatch, useSelector } from "react-redux";
+import toast from "react-hot-toast";
+
+import posthog from "posthog-js";
+import {
+  useWsClient,
+  WsClientProviderStatus,
+} from "#/context/ws-client-provider";
+import { ErrorObservation } from "#/types/core/observations";
+import { addErrorMessage, addUserMessage } from "#/state/chatSlice";
+import { handleAssistantMessage } from "#/services/actions";
+import {
+  getCloneRepoCommand,
+  getGitHubTokenCommand,
+} from "#/services/terminalService";
+import {
+  clearFiles,
+  clearSelectedRepository,
+  setImportedProjectZip,
+} from "#/state/initial-query-slice";
+import { clientLoader as appClientLoader } from "#/routes/_oh.app";
+import store, { RootState } from "#/store";
+import { createChatMessage } from "#/services/chatService";
+import { clientLoader as rootClientLoader } from "#/routes/_oh";
+import { isGitHubErrorReponse } from "#/api/github";
+import OpenHands from "#/api/open-hands";
+import { base64ToBlob } from "#/utils/base64-to-blob";
+import { setCurrentAgentState } from "#/state/agentSlice";
+import AgentState from "#/types/AgentState";
+import { getSettings } from "#/services/settings";
+
+interface ServerError {
+  error: boolean | string;
+  message: string;
+  [key: string]: unknown;
+}
+
+const isServerError = (data: object): data is ServerError => "error" in data;
+
+const isErrorObservation = (data: object): data is ErrorObservation =>
+  "observation" in data && data.observation === "error";
+
+export function EventHandler({ children }: React.PropsWithChildren) {
+  const { events, status, send } = useWsClient();
+  const statusRef = React.useRef<WsClientProviderStatus | null>(null);
+  const runtimeActive = status === WsClientProviderStatus.ACTIVE;
+  const fetcher = useFetcher();
+  const dispatch = useDispatch();
+  const { files, importedProjectZip } = useSelector(
+    (state: RootState) => state.initalQuery,
+  );
+  const { ghToken, repo } = useLoaderData<typeof appClientLoader>();
+  const initialQueryRef = React.useRef<string | null>(
+    store.getState().initalQuery.initialQuery,
+  );
+
+  const sendInitialQuery = (query: string, base64Files: string[]) => {
+    const timestamp = new Date().toISOString();
+    send(createChatMessage(query, base64Files, timestamp));
+  };
+  const data = useRouteLoaderData<typeof rootClientLoader>("routes/_oh");
+  const userId = React.useMemo(() => {
+    if (data?.user && !isGitHubErrorReponse(data.user)) return data.user.id;
+    return null;
+  }, [data?.user]);
+  const userSettings = getSettings();
+
+  React.useEffect(() => {
+    if (!events.length) {
+      return;
+    }
+    const event = events[events.length - 1];
+    if (event.token) {
+      fetcher.submit({ token: event.token as string }, { method: "post" });
+      return;
+    }
+
+    if (isServerError(event)) {
+      if (event.error_code === 401) {
+        toast.error("Session expired.");
+        fetcher.submit({}, { method: "POST", action: "/end-session" });
+        return;
+      }
+
+      if (typeof event.error === "string") {
+        toast.error(event.error);
+      } else {
+        toast.error(event.message);
+      }
+      return;
+    }
+
+    if (isErrorObservation(event)) {
+      dispatch(
+        addErrorMessage({
+          id: event.extras?.error_id,
+          message: event.message,
+        }),
+      );
+      return;
+    }
+    handleAssistantMessage(event);
+  }, [events.length]);
+
+  React.useEffect(() => {
+    if (statusRef.current === status) {
+      return; // This is a check because of strict mode - if the status did not change, don't do anything
+    }
+    statusRef.current = status;
+    const initialQuery = initialQueryRef.current;
+
+    if (status === WsClientProviderStatus.ACTIVE) {
+      let additionalInfo = "";
+      if (ghToken && repo) {
+        send(getCloneRepoCommand(ghToken, repo));
+        additionalInfo = `Repository ${repo} has been cloned to /workspace. Please check the /workspace for files.`;
+        dispatch(clearSelectedRepository()); // reset selected repository; maybe better to move this to '/'?
+      }
+      // if there's an uploaded project zip, add it to the chat
+      else if (importedProjectZip) {
+        additionalInfo = `Files have been uploaded. Please check the /workspace for files.`;
+      }
+
+      if (initialQuery) {
+        if (additionalInfo) {
+          sendInitialQuery(`${initialQuery}\n\n[${additionalInfo}]`, files);
+        } else {
+          sendInitialQuery(initialQuery, files);
+        }
+        dispatch(clearFiles()); // reset selected files
+        initialQueryRef.current = null;
+      }
+    }
+
+    if (status === WsClientProviderStatus.OPENING && initialQuery) {
+      dispatch(
+        addUserMessage({
+          content: initialQuery,
+          imageUrls: files,
+          timestamp: new Date().toISOString(),
+        }),
+      );
+    }
+
+    if (status === WsClientProviderStatus.STOPPED) {
+      store.dispatch(setCurrentAgentState(AgentState.STOPPED));
+    }
+  }, [status]);
+
+  React.useEffect(() => {
+    if (runtimeActive && userId && ghToken) {
+      // Export if the user valid, this could happen mid-session so it is handled here
+      send(getGitHubTokenCommand(ghToken));
+    }
+  }, [userId, ghToken, runtimeActive]);
+
+  React.useEffect(() => {
+    (async () => {
+      if (runtimeActive && importedProjectZip) {
+        // upload files action
+        try {
+          const blob = base64ToBlob(importedProjectZip);
+          const file = new File([blob], "imported-project.zip", {
+            type: blob.type,
+          });
+          await OpenHands.uploadFiles([file]);
+          dispatch(setImportedProjectZip(null));
+        } catch (error) {
+          toast.error("Failed to upload project files.");
+        }
+      }
+    })();
+  }, [runtimeActive, importedProjectZip]);
+
+  React.useEffect(() => {
+    if (userSettings.LLM_API_KEY) {
+      posthog.capture("user_activated");
+    }
+  }, [userSettings.LLM_API_KEY]);
+
+  return children;
+}

frontend/src/components/image-preview.tsx

@@ -1,4 +1,4 @@
-import CloseIcon from "#/assets/close.svg?react";
+import CloseIcon from "#/icons/close.svg?react";
 import { cn } from "#/utils/utils";
 
 interface ImagePreviewProps {

frontend/src/components/modals/LoadingProject.tsx

@@ -1,5 +1,5 @@
 import { useTranslation } from "react-i18next";
-import LoadingSpinnerOuter from "#/assets/loading-outer.svg?react";
+import LoadingSpinnerOuter from "#/icons/loading-outer.svg?react";
 import { cn } from "#/utils/utils";
 import ModalBody from "./ModalBody";
 import { I18nKey } from "#/i18n/declaration";

frontend/src/components/project-menu/ProjectMenuCard.tsx

@@ -2,17 +2,17 @@ import React from "react";
 import { useDispatch } from "react-redux";
 import toast from "react-hot-toast";
 import posthog from "posthog-js";
-import EllipsisH from "#/assets/ellipsis-h.svg?react";
+import EllipsisH from "#/icons/ellipsis-h.svg?react";
 import { ModalBackdrop } from "../modals/modal-backdrop";
 import { ConnectToGitHubModal } from "../modals/connect-to-github-modal";
 import { addUserMessage } from "#/state/chatSlice";
-import { useSocket } from "#/context/socket";
 import { createChatMessage } from "#/services/chatService";
 import { ProjectMenuCardContextMenu } from "./project.menu-card-context-menu";
 import { ProjectMenuDetailsPlaceholder } from "./project-menu-details-placeholder";
 import { ProjectMenuDetails } from "./project-menu-details";
 import { downloadWorkspace } from "#/utils/download-workspace";
 import { LoadingSpinner } from "../modals/LoadingProject";
+import { useWsClient } from "#/context/ws-client-provider";
 
 interface ProjectMenuCardProps {
   isConnectedToGitHub: boolean;
@@ -27,7 +27,7 @@ export function ProjectMenuCard({
   isConnectedToGitHub,
   githubData,
 }: ProjectMenuCardProps) {
-  const { send } = useSocket();
+  const { send } = useWsClient();
   const dispatch = useDispatch();
 
   const [contextMenuIsOpen, setContextMenuIsOpen] = React.useState(false);
@@ -43,10 +43,7 @@ export function ProjectMenuCard({
     posthog.capture("push_to_github_button_clicked");
     const rawEvent = {
       content: `
-Let's push the code to GitHub.
+Please push the changes to GitHub and open a pull request.
-If we're currently on the openhands-workspace branch, please create a new branch with a descriptive name.
-Commit any changes and push them to the remote repository.
-Finally, open up a pull request using the GitHub API and the token in the GITHUB_TOKEN environment variable, then show me the URL of the pull request.
 `,
       imageUrls: [],
       timestamp: new Date().toISOString(),

frontend/src/components/project-menu/project-menu-details-placeholder.tsx

@@ -1,6 +1,6 @@
 import { useTranslation } from "react-i18next";
 import { cn } from "#/utils/utils";
-import CloudConnection from "#/assets/cloud-connection.svg?react";
+import CloudConnection from "#/icons/cloud-connection.svg?react";
 import { I18nKey } from "#/i18n/declaration";
 
 interface ProjectMenuDetailsPlaceholderProps {

frontend/src/components/project-menu/project-menu-details.tsx

@@ -1,5 +1,5 @@
 import { useTranslation } from "react-i18next";
-import ExternalLinkIcon from "#/assets/external-link.svg?react";
+import ExternalLinkIcon from "#/icons/external-link.svg?react";
 import { formatTimeDelta } from "#/utils/format-time-delta";
 import { I18nKey } from "#/i18n/declaration";
 

frontend/src/components/scroll-to-bottom-button.tsx

@@ -1,4 +1,4 @@
-import ArrowSendIcon from "#/assets/arrow-send.svg?react";
+import ArrowSendIcon from "#/icons/arrow-send.svg?react";
 
 interface ScrollToBottomButtonProps {
   onClick: () => void;

frontend/src/components/suggestion-bubble.tsx

@@ -1,5 +1,5 @@
-import Lightbulb from "#/assets/lightbulb.svg?react";
+import Lightbulb from "#/icons/lightbulb.svg?react";
-import Refresh from "#/assets/refresh.svg?react";
+import Refresh from "#/icons/refresh.svg?react";
 
 interface SuggestionBubbleProps {
   suggestion: string;

frontend/src/components/upload-image-input.tsx

@@ -1,4 +1,4 @@
-import Clip from "#/assets/clip.svg?react";
+import Clip from "#/icons/clip.svg?react";
 
 interface UploadImageInputProps {
   onUpload: (files: File[]) => void;

frontend/src/components/user-avatar.tsx

@@ -1,5 +1,5 @@
 import { LoadingSpinner } from "./modals/LoadingProject";
-import DefaultUserAvatar from "#/assets/default-user.svg?react";
+import DefaultUserAvatar from "#/icons/default-user.svg?react";
 import { cn } from "#/utils/utils";
 
 interface UserAvatarProps {

frontend/src/context/socket.tsx

@@ -1,146 +0,0 @@
-import React from "react";
-import { Data } from "ws";
-import posthog from "posthog-js";
-import EventLogger from "#/utils/event-logger";
-
-interface WebSocketClientOptions {
-  token: string | null;
-  onOpen?: (event: Event) => void;
-  onMessage?: (event: MessageEvent<Data>) => void;
-  onError?: (event: Event) => void;
-  onClose?: (event: Event) => void;
-}
-
-interface WebSocketContextType {
-  send: (data: string | ArrayBufferLike | Blob | ArrayBufferView) => void;
-  start: (options?: WebSocketClientOptions) => void;
-  stop: () => void;
-  setRuntimeIsInitialized: () => void;
-  runtimeActive: boolean;
-  isConnected: boolean;
-  events: Record<string, unknown>[];
-}
-
-const SocketContext = React.createContext<WebSocketContextType | undefined>(
-  undefined,
-);
-
-interface SocketProviderProps {
-  children: React.ReactNode;
-}
-
-function SocketProvider({ children }: SocketProviderProps) {
-  const wsRef = React.useRef<WebSocket | null>(null);
-  const [isConnected, setIsConnected] = React.useState(false);
-  const [runtimeActive, setRuntimeActive] = React.useState(false);
-  const [events, setEvents] = React.useState<Record<string, unknown>[]>([]);
-
-  const setRuntimeIsInitialized = () => {
-    setRuntimeActive(true);
-  };
-
-  const start = React.useCallback((options?: WebSocketClientOptions): void => {
-    if (wsRef.current) {
-      EventLogger.warning(
-        "WebSocket connection is already established, but a new one is starting anyways.",
-      );
-    }
-
-    const baseUrl =
-      import.meta.env.VITE_BACKEND_BASE_URL || window?.location.host;
-    const protocol = window.location.protocol === "https:" ? "wss:" : "ws:";
-    const sessionToken = options?.token || "NO_JWT"; // not allowed to be empty or duplicated
-    const ghToken = localStorage.getItem("ghToken") || "NO_GITHUB";
-
-    const ws = new WebSocket(`${protocol}//${baseUrl}/ws`, [
-      "openhands",
-      sessionToken,
-      ghToken,
-    ]);
-
-    ws.addEventListener("open", (event) => {
-      posthog.capture("socket_opened");
-      setIsConnected(true);
-      options?.onOpen?.(event);
-    });
-
-    ws.addEventListener("message", (event) => {
-      EventLogger.message(event);
-
-      setEvents((prevEvents) => [...prevEvents, JSON.parse(event.data)]);
-      options?.onMessage?.(event);
-    });
-
-    ws.addEventListener("error", (event) => {
-      posthog.capture("socket_error");
-      EventLogger.event(event, "SOCKET ERROR");
-      options?.onError?.(event);
-    });
-
-    ws.addEventListener("close", (event) => {
-      posthog.capture("socket_closed");
-      EventLogger.event(event, "SOCKET CLOSE");
-
-      setIsConnected(false);
-      setRuntimeActive(false);
-      wsRef.current = null;
-      options?.onClose?.(event);
-    });
-
-    wsRef.current = ws;
-  }, []);
-
-  const stop = React.useCallback((): void => {
-    if (wsRef.current) {
-      wsRef.current.close();
-      wsRef.current = null;
-    }
-  }, []);
-
-  const send = React.useCallback(
-    (data: string | ArrayBufferLike | Blob | ArrayBufferView) => {
-      if (!wsRef.current) {
-        EventLogger.error("WebSocket is not connected.");
-        return;
-      }
-      setEvents((prevEvents) => [...prevEvents, JSON.parse(data.toString())]);
-      wsRef.current.send(data);
-    },
-    [],
-  );
-
-  const value = React.useMemo(
-    () => ({
-      send,
-      start,
-      stop,
-      setRuntimeIsInitialized,
-      runtimeActive,
-      isConnected,
-      events,
-    }),
-    [
-      send,
-      start,
-      stop,
-      setRuntimeIsInitialized,
-      runtimeActive,
-      isConnected,
-      events,
-    ],
-  );
-
-  return (
-    <SocketContext.Provider value={value}>{children}</SocketContext.Provider>
-  );
-}
-
-function useSocket() {
-  const context = React.useContext(SocketContext);
-  if (context === undefined) {
-    throw new Error("useSocket must be used within a SocketProvider");
-  }
-  return context;
-}
-
-export { SocketProvider, useSocket };

frontend/src/context/ws-client-provider.tsx

@@ -0,0 +1,175 @@
+import posthog from "posthog-js";
+import React from "react";
+import { Settings } from "#/services/settings";
+import ActionType from "#/types/ActionType";
+import EventLogger from "#/utils/event-logger";
+import AgentState from "#/types/AgentState";
+
+export enum WsClientProviderStatus {
+  STOPPED,
+  OPENING,
+  ACTIVE,
+  ERROR,
+}
+
+interface UseWsClient {
+  status: WsClientProviderStatus;
+  events: Record<string, unknown>[];
+  send: (event: Record<string, unknown>) => void;
+}
+
+const WsClientContext = React.createContext<UseWsClient>({
+  status: WsClientProviderStatus.STOPPED,
+  events: [],
+  send: () => {
+    throw new Error("not connected");
+  },
+});
+
+interface WsClientProviderProps {
+  enabled: boolean;
+  token: string | null;
+  ghToken: string | null;
+  settings: Settings | null;
+}
+
+export function WsClientProvider({
+  enabled,
+  token,
+  ghToken,
+  settings,
+  children,
+}: React.PropsWithChildren<WsClientProviderProps>) {
+  const wsRef = React.useRef<WebSocket | null>(null);
+  const tokenRef = React.useRef<string | null>(token);
+  const ghTokenRef = React.useRef<string | null>(ghToken);
+  const closeRef = React.useRef<ReturnType<typeof setTimeout> | null>(null);
+  const [status, setStatus] = React.useState(WsClientProviderStatus.STOPPED);
+  const [events, setEvents] = React.useState<Record<string, unknown>[]>([]);
+
+  function send(event: Record<string, unknown>) {
+    if (!wsRef.current) {
+      EventLogger.error("WebSocket is not connected.");
+      return;
+    }
+    wsRef.current.send(JSON.stringify(event));
+  }
+
+  function handleOpen() {
+    setStatus(WsClientProviderStatus.OPENING);
+    const initEvent = {
+      action: ActionType.INIT,
+      args: settings,
+    };
+    send(initEvent);
+  }
+
+  function handleMessage(messageEvent: MessageEvent) {
+    const event = JSON.parse(messageEvent.data);
+    setEvents((prevEvents) => [...prevEvents, event]);
+    if (event.extras?.agent_state === AgentState.INIT) {
+      setStatus(WsClientProviderStatus.ACTIVE);
+    }
+    if (
+      status !== WsClientProviderStatus.ACTIVE &&
+      event?.observation === "error"
+    ) {
+      setStatus(WsClientProviderStatus.ERROR);
+    }
+  }
+
+  function handleClose() {
+    setStatus(WsClientProviderStatus.STOPPED);
+    setEvents([]);
+    wsRef.current = null;
+  }
+
+  function handleError(event: Event) {
+    posthog.capture("socket_error");
+    EventLogger.event(event, "SOCKET ERROR");
+    setStatus(WsClientProviderStatus.ERROR);
+  }
+
+  // Connect websocket
+  React.useEffect(() => {
+    let ws = wsRef.current;
+
+    // If disabled close any existing websockets...
+    if (!enabled) {
+      if (ws) {
+        ws.close();
+      }
+      wsRef.current = null;
+      return () => {};
+    }
+
+    // If there is no websocket or the tokens have changed or the current websocket is closed,
+    // create a new one
+    if (
+      !ws ||
+      (tokenRef.current && token !== tokenRef.current) ||
+      ghToken !== ghTokenRef.current ||
+      ws.readyState === WebSocket.CLOSED ||
+      ws.readyState === WebSocket.CLOSING
+    ) {
+      ws?.close();
+      const baseUrl =
+        import.meta.env.VITE_BACKEND_BASE_URL || window?.location.host;
+      const protocol = window.location.protocol === "https:" ? "wss:" : "ws:";
+      ws = new WebSocket(`${protocol}//${baseUrl}/ws`, [
+        "openhands",
+        token || "NO_JWT",
+        ghToken || "NO_GITHUB",
+      ]);
+    }
+    ws.addEventListener("open", handleOpen);
+    ws.addEventListener("message", handleMessage);
+    ws.addEventListener("error", handleError);
+    ws.addEventListener("close", handleClose);
+    wsRef.current = ws;
+    tokenRef.current = token;
+    ghTokenRef.current = ghToken;
+
+    return () => {
+      ws.removeEventListener("open", handleOpen);
+      ws.removeEventListener("message", handleMessage);
+      ws.removeEventListener("error", handleError);
+      ws.removeEventListener("close", handleClose);
+    };
+  }, [enabled, token, ghToken]);
+
+  // Strict mode mounts and unmounts each component twice, so we have to wait in the destructor
+  // before actually closing the socket and cancel the operation if the component gets remounted.
+  React.useEffect(() => {
+    const timeout = closeRef.current;
+    if (timeout != null) {
+      clearTimeout(timeout);
+    }
+
+    return () => {
+      closeRef.current = setTimeout(() => {
+        wsRef.current?.close();
+      }, 100);
+    };
+  }, []);
+
+  const value = React.useMemo<UseWsClient>(
+    () => ({
+      status,
+      events,
+      send,
+    }),
+    [status, events],
+  );
+
+  return (
+    <WsClientContext.Provider value={value}>
+      {children}
+    </WsClientContext.Provider>
+  );
+}
+
+export function useWsClient() {
+  const context = React.useContext(WsClientContext);
+  return context;
+}

frontend/src/entry.client.tsx

@@ -10,7 +10,6 @@ import React, { startTransition, StrictMode } from "react";
 import { hydrateRoot } from "react-dom/client";
 import { Provider } from "react-redux";
 import posthog from "posthog-js";
-import { SocketProvider } from "./context/socket";
 import "./i18n";
 import store from "./store";
 
@@ -43,12 +42,10 @@ prepareApp().then(() =>
     hydrateRoot(
       document,
       <StrictMode>
-        <SocketProvider>
+        <Provider store={store}>
-          <Provider store={store}>
+          <RemixBrowser />
-            <RemixBrowser />
+          <PosthogInit />
-            <PosthogInit />
+        </Provider>
-          </Provider>
-        </SocketProvider>
       </StrictMode>,
     );
   }),

frontend/src/hooks/useTerminal.ts

@@ -4,7 +4,7 @@ import React from "react";
 import { Command } from "#/state/commandSlice";
 import { getTerminalCommand } from "#/services/terminalService";
 import { parseTerminalOutput } from "#/utils/parseTerminalOutput";
-import { useSocket } from "#/context/socket";
+import { useWsClient } from "#/context/ws-client-provider";
 
 /*
   NOTE: Tests for this hook are indirectly covered by the tests for the XTermTerminal component.
@@ -15,7 +15,7 @@ export const useTerminal = (
   commands: Command[] = [],
   secrets: string[] = [],
 ) => {
-  const { send } = useSocket();
+  const { send } = useWsClient();
   const terminal = React.useRef<Terminal | null>(null);
   const fitAddon = React.useRef<FitAddon | null>(null);
   const ref = React.useRef<HTMLDivElement>(null);

frontend/src/i18n/translation.json

@@ -535,7 +535,8 @@
     "pt": "Socket não inicializado",
     "ko-KR": "소켓이 초기화되지 않았습니다",
     "ar": "لم يتم تهيئة Socket",
-    "tr": "Soket başlatılmadı"
+    "tr": "Soket başlatılmadı",
+    "no": "Socket ikke initialisert"
   },
   "EXPLORER$UPLOAD_ERROR_MESSAGE": {
     "en": "Error uploading file",
@@ -548,7 +549,8 @@
     "pt": "Erro ao fazer upload do arquivo",
     "ko-KR": "파일 업로드 중 오류 발생",
     "ar": "خطأ في تحميل الملف",
-    "tr": "Dosya yüklenirken hata oluştu"
+    "tr": "Dosya yüklenirken hata oluştu",
+    "no": "Feil ved opplasting av fil"
   },
   "EXPLORER$LABEL_DROP_FILES": {
     "en": "Drop files here",
@@ -557,6 +559,7 @@
     "zh-TW": "將檔案拖曳至此",
     "es": "Suelta los archivos aquí",
     "fr": "Déposez les fichiers ici",
+    "no": "Slipp filer her",
     "it": "Trascina i file qui",
     "pt": "Solte os arquivos aqui",
     "ko-KR": "파일을 여기에 놓으세요",
@@ -574,7 +577,8 @@
     "pt": "Espaço de trabalho",
     "ko-KR": "작업 공간",
     "ar": "مساحة العمل",
-    "tr": "Çalışma alanı"
+    "tr": "Çalışma alanı",
+    "no": "Arbeidsområde"
   },
   "EXPLORER$EMPTY_WORKSPACE_MESSAGE": {
     "en": "No files in workspace",
@@ -587,7 +591,8 @@
     "pt": "Nenhum arquivo no espaço de trabalho",
     "ko-KR": "작업 공간에 파일이 없습니다",
     "ar": "لا توجد ملفات في مساحة العمل",
-    "tr": "Çalışma alanında dosya yok"
+    "tr": "Çalışma alanında dosya yok",
+    "no": "Ingen filer i arbeidsområdet"
   },
   "EXPLORER$LOADING_WORKSPACE_MESSAGE": {
     "en": "Loading workspace...",
@@ -600,7 +605,8 @@
     "pt": "Carregando espaço de trabalho...",
     "ko-KR": "작업 공간 로딩 중...",
     "ar": "جارٍ تحميل مساحة العمل...",
-    "tr": "Çalışma alanı yükleniyor..."
+    "tr": "Çalışma alanı yükleniyor...",
+    "no": "Laster arbeidsområde..."
   },
   "EXPLORER$REFRESH_ERROR_MESSAGE": {
     "en": "Error refreshing workspace",
@@ -613,7 +619,8 @@
     "pt": "Erro ao atualizar o espaço de trabalho",
     "ko-KR": "작업 공간 새로 고침 오류",
     "ar": "خطأ في تحديث مساحة العمل",
-    "tr": "Çalışma alanı yenilenirken hata oluştu"
+    "tr": "Çalışma alanı yenilenirken hata oluştu",
+    "no": "Feil ved oppdatering av arbeidsområde"
   },
   "EXPLORER$UPLOAD_SUCCESS_MESSAGE": {
     "en": "Successfully uploaded {{count}} file(s)",
@@ -626,7 +633,8 @@
     "pt": "{{count}} arquivo(s) carregado(s) com sucesso",
     "ko-KR": "{{count}}개의 파일을 성공적으로 업로드했습니다",
     "ar": "تم تحميل {{count}} ملف (ملفات) بنجاح",
-    "tr": "{{count}} dosya başarıyla yüklendi"
+    "tr": "{{count}} dosya başarıyla yüklendi",
+    "no": "Lastet opp {{count}} fil(er) vellykket"
   },
   "EXPLORER$NO_FILES_UPLOADED_MESSAGE": {
     "en": "No files were uploaded",
@@ -639,7 +647,8 @@
     "pt": "Nenhum arquivo foi carregado",
     "ko-KR": "업로드된 파일이 없습니다",
     "ar": "لم يتم تحميل أي ملفات",
-    "tr": "Hiçbir dosya yüklenmedi"
+    "tr": "Hiçbir dosya yüklenmedi",
+    "no": "Ingen filer ble lastet opp"
   },
   "EXPLORER$UPLOAD_PARTIAL_SUCCESS_MESSAGE": {
     "en": "{{count}} file(s) were skipped during upload",
@@ -652,7 +661,8 @@
     "pt": "{{count}} arquivo(s) foram ignorados durante o upload",
     "ko-KR": "업로드 중 {{count}}개의 파일이 건너뛰어졌습니다",
     "ar": "تم تخطي {{count}} ملف (ملفات) أثناء التحميل",
-    "tr": "Yükleme sırasında {{count}} dosya atlandı"
+    "tr": "Yükleme sırasında {{count}} dosya atlandı",
+    "no": "{{count}} fil(er) ble hoppet over under opplasting"
   },
   "EXPLORER$UPLOAD_UNEXPECTED_RESPONSE_MESSAGE": {
     "en": "Unexpected response structure from server",
@@ -665,7 +675,8 @@
     "pt": "Estrutura de resposta inesperada do servidor",
     "ko-KR": "서버로부터 예상치 못한 응답 구조",
     "ar": "بنية استجابة غير متوقعة من الخادم",
-    "tr": "Sunucudan beklenmeyen yanıt yapısı"
+    "tr": "Sunucudan beklenmeyen yanıt yapısı",
+    "no": "Uventet responsstruktur fra serveren"
   },
   "LOAD_SESSION$MODAL_TITLE": {
     "en": "Return to existing session?",
@@ -799,95 +810,325 @@
   },
   "FEEDBACK$EMAIL_PLACEHOLDER": {
     "en": "Enter your email address",
-    "es": "Ingresa tu correo electrónico"
+    "es": "Ingresa tu correo electrónico",
+    "zh-CN": "输入您的电子邮件地址",
+    "zh-TW": "輸入您的電子郵件地址",
+    "ko-KR": "이메일 주소를 입력하세요",
+    "no": "Skriv inn din e-postadresse",
+    "ar": "أدخل عنوان بريدك الإلكتروني",
+    "de": "Geben Sie Ihre E-Mail-Adresse ein",
+    "fr": "Entrez votre adresse e-mail",
+    "it": "Inserisci il tuo indirizzo email",
+    "pt": "Digite seu endereço de e-mail",
+    "tr": "E-posta adresinizi girin"
   },
   "FEEDBACK$PASSWORD_COPIED_MESSAGE": {
     "en": "Password copied to clipboard.",
-    "es": "Contraseña copiada al portapapeles."
+    "es": "Contraseña copiada al portapapeles.",
+    "zh-CN": "密码已复制到剪贴板。",
+    "zh-TW": "密碼已複製到剪貼板。",
+    "ko-KR": "비밀번호가 클립보드에 복사되었습니다.",
+    "no": "Passord kopiert til utklippstavlen.",
+    "ar": "تم نسخ كلمة المرور إلى الحافظة.",
+    "de": "Passwort in die Zwischenablage kopiert.",
+    "fr": "Mot de passe copié dans le presse-papiers.",
+    "it": "Password copiata negli appunti.",
+    "pt": "Senha copiada para a área de transferência.",
+    "tr": "Parola panoya kopyalandı."
   },
   "FEEDBACK$GO_TO_FEEDBACK": {
     "en": "Go to shared feedback",
-    "es": "Ir a feedback compartido"
+    "es": "Ir a feedback compartido",
+    "zh-CN": "转到共享反馈",
+    "zh-TW": "前往共享反饋",
+    "ko-KR": "공유된 피드백으로 이동",
+    "no": "Gå til delt tilbakemelding",
+    "ar": "الذهاب إلى التعليقات المشتركة",
+    "de": "Zum geteilten Feedback gehen",
+    "fr": "Aller aux commentaires partagés",
+    "it": "Vai al feedback condiviso",
+    "pt": "Ir para feedback compartilhado",
+    "tr": "Paylaşılan geri bildirimlere git"
   },
   "FEEDBACK$PASSWORD": {
     "en": "Password:",
-    "es": "Contraseña:"
+    "es": "Contraseña:",
+    "zh-CN": "密码：",
+    "zh-TW": "密碼：",
+    "ko-KR": "비밀번호:",
+    "no": "Passord:",
+    "ar": "كلمة المرور:",
+    "de": "Passwort:",
+    "fr": "Mot de passe :",
+    "it": "Password:",
+    "pt": "Senha:",
+    "tr": "Parola:"
   },
   "FEEDBACK$INVALID_EMAIL_FORMAT": {
     "en": "Invalid email format",
-    "es": "Formato de correo inválido"
+    "es": "Formato de correo inválido",
+    "zh-CN": "无效的电子邮件格式",
+    "zh-TW": "無效的電子郵件格式",
+    "ko-KR": "잘못된 이메일 형식",
+    "no": "Ugyldig e-postformat",
+    "ar": "تنسيق البريد الإلكتروني غير صالح",
+    "de": "Ungültiges E-Mail-Format",
+    "fr": "Format d'e-mail invalide",
+    "it": "Formato email non valido",
+    "pt": "Formato de e-mail inválido",
+    "tr": "Geçersiz e-posta biçimi"
   },
   "FEEDBACK$FAILED_TO_SHARE": {
     "en": "Failed to share, please contact the developers:",
-    "es": "Error al compartir, por favor contacta con los desarrolladores:"
+    "es": "Error al compartir, por favor contacta con los desarrolladores:",
+    "zh-CN": "分享失败，请联系开发人员：",
+    "zh-TW": "分享失敗，請聯繫開發人員：",
+    "ko-KR": "공유 실패, 개발자에게 문의하세요:",
+    "no": "Deling mislyktes, vennligst kontakt utviklerne:",
+    "ar": "فشل المشاركة، يرجى الاتصال بالمطورين:",
+    "de": "Teilen fehlgeschlagen, bitte kontaktieren Sie die Entwickler:",
+    "fr": "Échec du partage, veuillez contacter les développeurs :",
+    "it": "Condivisione fallita, contattare gli sviluppatori:",
+    "pt": "Falha ao compartilhar, entre em contato com os desenvolvedores:",
+    "tr": "Paylaşım başarısız, lütfen geliştiricilerle iletişime geçin:"
   },
   "FEEDBACK$COPY_LABEL": {
     "en": "Copy",
-    "es": "Copiar"
+    "es": "Copiar",
+    "zh-CN": "复制",
+    "zh-TW": "複製",
+    "ko-KR": "복사",
+    "no": "Kopier",
+    "ar": "نسخ",
+    "de": "Kopieren",
+    "fr": "Copier",
+    "it": "Copia",
+    "pt": "Copiar",
+    "tr": "Kopyala"
   },
   "FEEDBACK$SHARING_SETTINGS_LABEL": {
     "en": "Sharing settings",
-    "es": "Configuración de compartir"
+    "es": "Configuración de compartir",
+    "zh-CN": "共享设置",
+    "zh-TW": "共享設定",
+    "ko-KR": "공유 설정",
+    "no": "Delingsinnstillinger",
+    "ar": "إعدادات المشاركة",
+    "de": "Freigabeeinstellungen",
+    "fr": "Paramètres de partage",
+    "it": "Impostazioni di condivisione",
+    "pt": "Configurações de compartilhamento",
+    "tr": "Paylaşım ayarları"
   },
   "SECURITY$UNKNOWN_ANALYZER_LABEL":{
     "en": "Unknown security analyzer chosen",
-    "es": "Analizador de seguridad desconocido"
+    "es": "Analizador de seguridad desconocido",
+    "zh-CN": "选择了未知的安全分析器",
+    "zh-TW": "選擇了未知的安全分析器",
+    "ko-KR": "알 수 없는 보안 분석기가 선택되었습니다",
+    "no": "Ukjent sikkerhetsanalysator valgt",
+    "ar": "تم اختيار محلل أمان غير معروف",
+    "de": "Unbekannter Sicherheitsanalysator ausgewählt",
+    "fr": "Analyseur de sécurité inconnu choisi",
+    "it": "Analizzatore di sicurezza sconosciuto selezionato",
+    "pt": "Analisador de segurança desconhecido escolhido",
+    "tr": "Bilinmeyen güvenlik analizörü seçildi"
   },
   "INVARIANT$UPDATE_POLICY_LABEL": {
     "en": "Update Policy",
-    "es": "Actualizar política"
+    "es": "Actualizar política",
+    "zh-CN": "更新策略",
+    "zh-TW": "更新策略",
+    "ko-KR": "정책 업데이트",
+    "no": "Oppdater policy",
+    "ar": "تحديث السياسة",
+    "de": "Richtlinie aktualisieren",
+    "fr": "Mettre à jour la politique",
+    "it": "Aggiorna policy",
+    "pt": "Atualizar política",
+    "tr": "İlkeyi güncelle"
   },
   "INVARIANT$UPDATE_SETTINGS_LABEL": {
     "en": "Update Settings",
-    "es": "Actualizar configuración"
+    "es": "Actualizar configuración",
+    "zh-CN": "更新设置",
+    "zh-TW": "更新設定",
+    "ko-KR": "설정 업데이트",
+    "no": "Oppdater innstillinger",
+    "ar": "تحديث الإعدادات",
+    "de": "Einstellungen aktualisieren",
+    "fr": "Mettre à jour les paramètres",
+    "it": "Aggiorna impostazioni",
+    "pt": "Atualizar configurações",
+    "tr": "Ayarları güncelle"
   },
   "INVARIANT$SETTINGS_LABEL": {
     "en": "Settings",
-    "es": "Configuración"
+    "es": "Configuración",
+    "zh-CN": "设置",
+    "zh-TW": "設定",
+    "ko-KR": "설정",
+    "no": "Innstillinger",
+    "ar": "الإعدادات",
+    "de": "Einstellungen",
+    "fr": "Paramètres",
+    "it": "Impostazioni",
+    "pt": "Configurações",
+    "tr": "Ayarlar"
   },
   "INVARIANT$ASK_CONFIRMATION_RISK_SEVERITY_LABEL": {
     "en": "Ask for user confirmation on risk severity:",
-    "es": "Preguntar por confirmación del usuario sobre severidad del riesgo:"
+    "es": "Preguntar por confirmación del usuario sobre severidad del riesgo:",
+    "zh-CN": "询问用户确认风险等级：",
+    "zh-TW": "詢問用戶確認風險等級：",
+    "ko-KR": "위험 심각도에 대한 사용자 확인 요청:",
+    "no": "Be om brukerbekreftelse på risikoalvorlighet:",
+    "ar": "اطلب تأكيد المستخدم على مستوى الخطورة:",
+    "de": "Nach Benutzerbestätigung für Risikoschweregrad fragen:",
+    "fr": "Demander la confirmation de l'utilisateur sur la gravité du risque :",
+    "it": "Chiedi conferma all'utente sulla gravità del rischio:",
+    "pt": "Solicitar confirmação do usuário sobre a gravidade do risco:",
+    "tr": "Risk şiddeti için kullanıcı onayı iste:"
   },
   "INVARIANT$DONT_ASK_FOR_CONFIRMATION_LABEL": {
     "en": "Don't ask for confirmation",
-    "es": "No solicitar confirmación"
+    "es": "No solicitar confirmación",
+    "zh-CN": "不要请求确认",
+    "zh-TW": "不要請求確認",
+    "ko-KR": "확인 요청하지 않음",
+    "no": "Ikke spør om bekreftelse",
+    "ar": "لا تطلب التأكيد",
+    "de": "Nicht nach Bestätigung fragen",
+    "fr": "Ne pas demander de confirmation",
+    "it": "Non chiedere conferma",
+    "pt": "Não solicitar confirmação",
+    "tr": "Onay isteme"
   },
   "INVARIANT$INVARIANT_ANALYZER_LABEL": {
     "en": "Invariant Analyzer",
-    "es": "Analizador de invariantes"
+    "es": "Analizador de invariantes",
+    "zh-CN": "不变量分析器",
+    "zh-TW": "不變量分析器",
+    "ko-KR": "불변성 분석기",
+    "no": "Invariant-analysator",
+    "ar": "محلل الثوابت",
+    "de": "Invarianten-Analysator",
+    "fr": "Analyseur d'invariants",
+    "it": "Analizzatore di invarianti",
+    "pt": "Analisador de invariantes",
+    "tr": "Değişmez Analizörü"
   },
   "INVARIANT$INVARIANT_ANALYZER_MESSAGE": {
     "en": "Invariant Analyzer continuously monitors your OpenHands agent for security issues.",
-    "es": "Analizador de invariantes continuamente monitorea tu agente de OpenHands por problemas de seguridad."
+    "es": "Analizador de invariantes continuamente monitorea tu agente de OpenHands por problemas de seguridad.",
+    "zh-CN": "不变量分析器持续监控您的 OpenHands 代理的安全问题。",
+    "zh-TW": "不變量分析器持續監控您的 OpenHands 代理的安全問題。",
+    "ko-KR": "불변성 분석기는 OpenHands 에이전트의 보안 문제를 지속적으로 모니터링합니다.",
+    "no": "Invariant-analysatoren overvåker kontinuerlig OpenHands-agenten din for sikkerhetsproblemer.",
+    "ar": "يراقب محلل الثوابت وكيل OpenHands الخاص بك باستمرار للتحقق من المشاكل الأمنية.",
+    "de": "Der Invarianten-Analysator überwacht kontinuierlich Ihren OpenHands-Agenten auf Sicherheitsprobleme.",
+    "fr": "L'analyseur d'invariants surveille en permanence votre agent OpenHands pour détecter les problèmes de sécurité.",
+    "it": "L'analizzatore di invarianti monitora continuamente il tuo agente OpenHands per problemi di sicurezza.",
+    "pt": "O analisador de invariantes monitora continuamente seu agente OpenHands em busca de problemas de segurança.",
+    "tr": "Değişmez Analizörü, OpenHands ajanınızı güvenlik sorunları için sürekli olarak izler."
   },
   "INVARIANT$CLICK_TO_LEARN_MORE_LABEL": {
     "en": "Click to learn more",
-    "es": "Clic para aprender más"
+    "es": "Clic para aprender más",
+    "zh-CN": "点击了解更多",
+    "zh-TW": "點擊了解更多",
+    "ko-KR": "자세히 알아보기",
+    "no": "Klikk for å lære mer",
+    "ar": "انقر لمعرفة المزيد",
+    "de": "Klicken Sie, um mehr zu erfahren",
+    "fr": "Cliquez pour en savoir plus",
+    "it": "Clicca per saperne di più",
+    "pt": "Clique para saber mais",
+    "tr": "Daha fazla bilgi için tıklayın"
   },
   "INVARIANT$POLICY_LABEL": {
     "en": "Policy",
-    "es": "Política"
+    "es": "Política",
+    "zh-CN": "策略",
+    "zh-TW": "策略",
+    "ko-KR": "정책",
+    "no": "Policy",
+    "ar": "السياسة",
+    "de": "Richtlinie",
+    "fr": "Politique",
+    "it": "Policy",
+    "pt": "Política",
+    "tr": "İlke"
   },
   "INVARIANT$LOG_LABEL": {
     "en": "Logs",
-    "es": "Logs"
+    "es": "Logs",
+    "zh-CN": "日志",
+    "zh-TW": "日誌",
+    "ko-KR": "로그",
+    "no": "Logger",
+    "ar": "السجلات",
+    "de": "Protokolle",
+    "fr": "Journaux",
+    "it": "Log",
+    "pt": "Logs",
+    "tr": "Günlükler"
   },
   "INVARIANT$EXPORT_TRACE_LABEL": {
     "en": "Export Trace",
-    "es": "Exportar traza"
+    "es": "Exportar traza",
+    "zh-CN": "导出跟踪",
+    "zh-TW": "匯出追蹤",
+    "ko-KR": "추적 내보내기",
+    "no": "Eksporter sporing",
+    "ar": "تصدير التتبع",
+    "de": "Ablaufverfolgung exportieren",
+    "fr": "Exporter la trace",
+    "it": "Esporta traccia",
+    "pt": "Exportar rastreamento",
+    "tr": "İzlemeyi dışa aktar"
   },
   "INVARIANT$TRACE_EXPORTED_MESSAGE": {
     "en": "Trace exported",
-    "es": "Traza exportada"
+    "es": "Traza exportada",
+    "zh-CN": "跟踪已导出",
+    "zh-TW": "追蹤已匯出",
+    "ko-KR": "추적 내보내기 완료",
+    "no": "Sporing eksportert",
+    "ar": "تم تصدير التتبع",
+    "de": "Ablaufverfolgung exportiert",
+    "fr": "Trace exportée",
+    "it": "Traccia esportata",
+    "pt": "Rastreamento exportado",
+    "tr": "İzleme dışa aktarıldı"
   },
   "INVARIANT$POLICY_UPDATED_MESSAGE": {
     "en": "Policy updated",
-    "es": "Política actualizada"
+    "es": "Política actualizada",
+    "zh-CN": "策略已更新",
+    "zh-TW": "策略已更新",
+    "ko-KR": "정책이 업데이트되었습니다",
+    "no": "Policy oppdatert",
+    "ar": "تم تحديث السياسة",
+    "de": "Richtlinie aktualisiert",
+    "fr": "Politique mise à jour",
+    "it": "Policy aggiornata",
+    "pt": "Política atualizada",
+    "tr": "İlke güncellendi"
   },
   "INVARIANT$SETTINGS_UPDATED_MESSAGE": {
     "en": "Settings updated",
-    "es": "Configuración actualizada"
+    "es": "Configuración actualizada",
+    "zh-CN": "设置已更新",
+    "zh-TW": "設定已更新",
+    "ko-KR": "설정이 업데이트되었습니다",
+    "no": "Innstillinger oppdatert",
+    "ar": "تم تحديث الإعدادات",
+    "de": "Einstellungen aktualisiert",
+    "fr": "Paramètres mis à jour",
+    "it": "Impostazioni aggiornate",
+    "pt": "Configurações atualizadas",
+    "tr": "Ayarlar güncellendi"
   },
   "CHAT_INTERFACE$INITIALIZING_AGENT_LOADING_MESSAGE": {
     "en": "Starting up!",
@@ -1276,7 +1517,8 @@
     "pt": "Conversa de chat",
     "es": "Conversación de chat",
     "ar": "محادثة تلقيم",
-    "fr": "Conversation de chat"
+    "fr": "Conversation de chat",
+    "tr": "Sohbet Konuşması"
   },
   "CHAT_INTERFACE$UNKNOWN_SENDER": {
     "en": "Unknown",
@@ -1531,10 +1773,12 @@
     "tr": "Özel"
   },
   "ERROR_MESSAGE$SHOW_DETAILS": {
-    "en": "Show details"
+    "en": "Show details",
+    "es": "Mostrar detalles"
   },
   "ERROR_MESSAGE$HIDE_DETAILS": {
-    "en": "Hide details"
+    "en": "Hide details",
+    "es": "Ocultar detalles"
   },
   "STATUS$STARTING_RUNTIME": {
     "en": "Starting Runtime...",
@@ -1616,7 +1860,7 @@
   },
   "ACCOUNT_SETTINGS_MODAL$CLOSE":{
     "en": "Close",
-    "es": ""
+    "es": "Cerrar"
   },
   "ACCOUNT_SETTINGS_MODAL$GITHUB_TOKEN_INVALID":{
     "en": "GitHub token is invalid. Please try again.",
@@ -1735,7 +1979,8 @@
     "es":"atrás"
   },
   "STATUS$ERROR_LLM_AUTHENTICATION": {
-    "en": "Error authenticating with the LLM provider. Please check your API key"
+    "en": "Error authenticating with the LLM provider. Please check your API key",
+    "es": "Error autenticando con el proveedor de LLM. Por favor revisa tu API key"
   },
   "STATUS$ERROR_RUNTIME_DISCONNECTED": {
     "en": "There was an error while connecting to the runtime. Please refresh the page."

frontend/src/routes/_oh._index/hero-heading.tsx

@@ -1,4 +1,4 @@
-import BuildIt from "#/assets/build-it.svg?react";
+import BuildIt from "#/icons/build-it.svg?react";
 
 export function HeroHeading() {
   return (

frontend/src/routes/_oh.app._index/code-editor-component.tsx

@@ -29,6 +29,10 @@ function CodeEditorCompoonent({
     if (selectedPath && value) modifyFileContent(selectedPath, value);
   };
 
+  const isBase64Image = (content: string) => content.startsWith("data:image/");
+  const isPDF = (content: string) => content.startsWith("data:application/pdf");
+  const isVideo = (content: string) => content.startsWith("data:video/");
+
   React.useEffect(() => {
     const handleSave = async (event: KeyboardEvent) => {
       if (selectedPath && event.metaKey && event.key === "s") {
@@ -62,16 +66,40 @@ function CodeEditorCompoonent({
     );
   }
 
+  const fileContent = modifiedFiles[selectedPath] || files[selectedPath];
+
+  if (isBase64Image(fileContent)) {
+    return (
+      <section className="flex flex-col relative items-center overflow-auto h-[90%]">
+        <img src={fileContent} alt={selectedPath} className="object-contain" />
+      </section>
+    );
+  }
+
+  if (isPDF(fileContent)) {
+    return (
+      <iframe
+        src={fileContent}
+        title={selectedPath}
+        width="100%"
+        height="100%"
+      />
+    );
+  }
+
+  if (isVideo(fileContent)) {
+    return (
+      <video controls src={fileContent} width="100%" height="100%">
+        <track kind="captions" label="English captions" />
+      </video>
+    );
+  }
   return (
     <Editor
       data-testid="code-editor"
       path={selectedPath ?? undefined}
       defaultValue=""
-      value={
+      value={selectedPath ? fileContent : undefined}
-        selectedPath
-          ? modifiedFiles[selectedPath] || files[selectedPath]
-          : undefined
-      }
       onMount={onMount}
       onChange={handleEditorChange}
       options={{ readOnly: isReadOnly }}

frontend/src/routes/_oh.app.tsx

@@ -2,71 +2,29 @@ import { useDisclosure } from "@nextui-org/react";
 import React from "react";
 import {
   Outlet,
-  useFetcher,
   useLoaderData,
   json,
   ClientActionFunctionArgs,
-  useRouteLoaderData,
 } from "@remix-run/react";
-import { useDispatch, useSelector } from "react-redux";
+import { useDispatch } from "react-redux";
-import WebSocket from "ws";
-import toast from "react-hot-toast";
 import { getSettings } from "#/services/settings";
 import Security from "../components/modals/security/Security";
 import { Controls } from "#/components/controls";
-import store, { RootState } from "#/store";
+import store from "#/store";
 import { Container } from "#/components/container";
-import ActionType from "#/types/ActionType";
+import { clearMessages } from "#/state/chatSlice";
-import { handleAssistantMessage } from "#/services/actions";
-import {
-  addErrorMessage,
-  addUserMessage,
-  clearMessages,
-} from "#/state/chatSlice";
-import { useSocket } from "#/context/socket";
-import {
-  getGitHubTokenCommand,
-  getCloneRepoCommand,
-} from "#/services/terminalService";
 import { clearTerminal } from "#/state/commandSlice";
 import { useEffectOnce } from "#/utils/use-effect-once";
-import CodeIcon from "#/assets/code.svg?react";
+import CodeIcon from "#/icons/code.svg?react";
-import GlobeIcon from "#/assets/globe.svg?react";
+import GlobeIcon from "#/icons/globe.svg?react";
-import ListIcon from "#/assets/list-type-number.svg?react";
+import ListIcon from "#/icons/list-type-number.svg?react";
-import { createChatMessage } from "#/services/chatService";
+import { clearInitialQuery } from "#/state/initial-query-slice";
-import {
-  clearFiles,
-  clearInitialQuery,
-  clearSelectedRepository,
-  setImportedProjectZip,
-} from "#/state/initial-query-slice";
 import { isGitHubErrorReponse, retrieveLatestGitHubCommit } from "#/api/github";
-import OpenHands from "#/api/open-hands";
-import AgentState from "#/types/AgentState";
-import { base64ToBlob } from "#/utils/base64-to-blob";
-import { clientLoader as rootClientLoader } from "#/routes/_oh";
 import { clearJupyter } from "#/state/jupyterSlice";
 import { FilesProvider } from "#/context/files";
-import { ErrorObservation } from "#/types/core/observations";
 import { ChatInterface } from "#/components/chat-interface";
-
+import { WsClientProvider } from "#/context/ws-client-provider";
-interface ServerError {
+import { EventHandler } from "#/components/event-handler";
-  error: boolean | string;
-  message: string;
-  [key: string]: unknown;
-}
-
-const isServerError = (data: object): data is ServerError => "error" in data;
-
-const isErrorObservation = (data: object): data is ErrorObservation =>
-  "observation" in data && data.observation === "error";
-
-const isAgentStateChange = (
-  data: object,
-): data is { extras: { agent_state: AgentState } } =>
-  "extras" in data &&
-  data.extras instanceof Object &&
-  "agent_state" in data.extras;
 
 export const clientLoader = async () => {
   const ghToken = localStorage.getItem("ghToken");
@@ -116,174 +74,26 @@ export const clientAction = async ({ request }: ClientActionFunctionArgs) => {
 
 function App() {
   const dispatch = useDispatch();
-  const { files, importedProjectZip } = useSelector(
+  const { settings, token, ghToken, lastCommit } =
-    (state: RootState) => state.initalQuery,
-  );
-  const { start, send, setRuntimeIsInitialized, runtimeActive } = useSocket();
-  const { settings, token, ghToken, repo, q, lastCommit } =
     useLoaderData<typeof clientLoader>();
-  const fetcher = useFetcher();
-  const data = useRouteLoaderData<typeof rootClientLoader>("routes/_oh");
 
   const secrets = React.useMemo(
     () => [ghToken, token].filter((secret) => secret !== null),
     [ghToken, token],
   );
 
-  // To avoid re-rendering the component when the user object changes, we memoize the user ID.
-  // We use this to ensure the github token is valid before exporting it to the terminal.
-  const userId = React.useMemo(() => {
-    if (data?.user && !isGitHubErrorReponse(data.user)) return data.user.id;
-    return null;
-  }, [data?.user]);
-
   const Terminal = React.useMemo(
     () => React.lazy(() => import("../components/terminal/Terminal")),
     [],
   );
 
-  const addIntialQueryToChat = (
-    query: string,
-    base64Files: string[],
-    timestamp = new Date().toISOString(),
-  ) => {
-    dispatch(
-      addUserMessage({
-        content: query,
-        imageUrls: base64Files,
-        timestamp,
-      }),
-    );
-  };
-
-  const sendInitialQuery = (query: string, base64Files: string[]) => {
-    const timestamp = new Date().toISOString();
-    send(createChatMessage(query, base64Files, timestamp));
-  };
-
-  const handleOpen = React.useCallback(() => {
-    const initEvent = {
-      action: ActionType.INIT,
-      args: settings,
-    };
-    send(JSON.stringify(initEvent));
-
-    // display query in UI, but don't send it to the server
-    if (q) addIntialQueryToChat(q, files);
-  }, [settings]);
-
-  const handleMessage = React.useCallback(
-    (message: MessageEvent<WebSocket.Data>) => {
-      // set token received from the server
-      const parsed = JSON.parse(message.data.toString());
-      if ("token" in parsed) {
-        fetcher.submit({ token: parsed.token }, { method: "post" });
-        return;
-      }
-
-      if (isServerError(parsed)) {
-        if (parsed.error_code === 401) {
-          toast.error("Session expired.");
-          fetcher.submit({}, { method: "POST", action: "/end-session" });
-          return;
-        }
-
-        if (typeof parsed.error === "string") {
-          toast.error(parsed.error);
-        } else {
-          toast.error(parsed.message);
-        }
-
-        return;
-      }
-      if (isErrorObservation(parsed)) {
-        dispatch(
-          addErrorMessage({
-            id: parsed.extras?.error_id,
-            message: parsed.message,
-          }),
-        );
-        return;
-      }
-
-      handleAssistantMessage(message.data.toString());
-
-      // handle first time connection
-      if (
-        isAgentStateChange(parsed) &&
-        parsed.extras.agent_state === AgentState.INIT
-      ) {
-        setRuntimeIsInitialized();
-
-        // handle new session
-        if (!token) {
-          let additionalInfo = "";
-          if (ghToken && repo) {
-            send(getCloneRepoCommand(ghToken, repo));
-            additionalInfo = `Repository ${repo} has been cloned to /workspace. Please check the /workspace for files.`;
-            dispatch(clearSelectedRepository()); // reset selected repository; maybe better to move this to '/'?
-          }
-          // if there's an uploaded project zip, add it to the chat
-          else if (importedProjectZip) {
-            additionalInfo = `Files have been uploaded. Please check the /workspace for files.`;
-          }
-
-          if (q) {
-            if (additionalInfo) {
-              sendInitialQuery(`${q}\n\n[${additionalInfo}]`, files);
-            } else {
-              sendInitialQuery(q, files);
-            }
-            dispatch(clearFiles()); // reset selected files
-          }
-        }
-      }
-    },
-    [token, ghToken, repo, q, files],
-  );
-
-  const startSocketConnection = React.useCallback(() => {
-    start({
-      token,
-      onOpen: handleOpen,
-      onMessage: handleMessage,
-    });
-  }, [token, handleOpen, handleMessage]);
-
   useEffectOnce(() => {
-    // clear and restart the socket connection
     dispatch(clearMessages());
     dispatch(clearTerminal());
     dispatch(clearJupyter());
     dispatch(clearInitialQuery()); // Clear initial query when navigating to /app
-    startSocketConnection();
   });
 
-  React.useEffect(() => {
-    if (runtimeActive && userId && ghToken) {
-      // Export if the user valid, this could happen mid-session so it is handled here
-      send(getGitHubTokenCommand(ghToken));
-    }
-  }, [userId, ghToken, runtimeActive]);
-
-  React.useEffect(() => {
-    (async () => {
-      if (runtimeActive && importedProjectZip) {
-        // upload files action
-        try {
-          const blob = base64ToBlob(importedProjectZip);
-          const file = new File([blob], "imported-project.zip", {
-            type: blob.type,
-          });
-          await OpenHands.uploadFiles([file]);
-          dispatch(setImportedProjectZip(null));
-        } catch (error) {
-          toast.error("Failed to upload project files.");
-        }
-      }
-    })();
-  }, [runtimeActive, importedProjectZip]);
-
   const {
     isOpen: securityModalIsOpen,
     onOpen: onSecurityModalOpen,
@@ -291,53 +101,62 @@ function App() {
   } = useDisclosure();
 
   return (
-    <div className="flex flex-col h-full gap-3">
+    <WsClientProvider
-      <div className="flex h-full overflow-auto gap-3">
+      enabled
-        <Container className="w-[390px] max-h-full relative">
+      token={token}
-          <ChatInterface />
+      ghToken={ghToken}
-        </Container>
+      settings={settings}
+    >
+      <EventHandler>
+        <div className="flex flex-col h-full gap-3">
+          <div className="flex h-full overflow-auto gap-3">
+            <Container className="w-[390px] max-h-full relative">
+              <ChatInterface />
+            </Container>
 
-        <div className="flex flex-col grow gap-3">
+            <div className="flex flex-col grow gap-3">
-          <Container
+              <Container
-            className="h-2/3"
+                className="h-2/3"
-            labels={[
+                labels={[
-              { label: "Workspace", to: "", icon: <CodeIcon /> },
+                  { label: "Workspace", to: "", icon: <CodeIcon /> },
-              { label: "Jupyter", to: "jupyter", icon: <ListIcon /> },
+                  { label: "Jupyter", to: "jupyter", icon: <ListIcon /> },
-              {
+                  {
-                label: "Browser",
+                    label: "Browser",
-                to: "browser",
+                    to: "browser",
-                icon: <GlobeIcon />,
+                    icon: <GlobeIcon />,
-                isBeta: true,
+                    isBeta: true,
-              },
+                  },
-            ]}
+                ]}
-          >
+              >
-            <FilesProvider>
+                <FilesProvider>
-              <Outlet />
+                  <Outlet />
-            </FilesProvider>
+                </FilesProvider>
-          </Container>
+              </Container>
-          {/* Terminal uses some API that is not compatible in a server-environment. For this reason, we lazy load it to ensure
+              {/* Terminal uses some API that is not compatible in a server-environment. For this reason, we lazy load it to ensure
-           * that it loads only in the client-side. */}
+               * that it loads only in the client-side. */}
-          <Container className="h-1/3 overflow-scroll" label="Terminal">
+              <Container className="h-1/3 overflow-scroll" label="Terminal">
-            <React.Suspense fallback={<div className="h-full" />}>
+                <React.Suspense fallback={<div className="h-full" />}>
-              <Terminal secrets={secrets} />
+                  <Terminal secrets={secrets} />
-            </React.Suspense>
+                </React.Suspense>
-          </Container>
+              </Container>
+            </div>
+          </div>
+
+          <div className="h-[60px]">
+            <Controls
+              setSecurityOpen={onSecurityModalOpen}
+              showSecurityLock={!!settings.SECURITY_ANALYZER}
+              lastCommitData={lastCommit}
+            />
+          </div>
+          <Security
+            isOpen={securityModalIsOpen}
+            onOpenChange={onSecurityModalOpenChange}
+            securityAnalyzer={settings.SECURITY_ANALYZER}
+          />
         </div>
-      </div>
+      </EventHandler>
-
+    </WsClientProvider>
-      <div className="h-[60px]">
-        <Controls
-          setSecurityOpen={onSecurityModalOpen}
-          showSecurityLock={!!settings.SECURITY_ANALYZER}
-          lastCommitData={lastCommit}
-        />
-      </div>
-      <Security
-        isOpen={securityModalIsOpen}
-        onOpenChange={onSecurityModalOpenChange}
-        securityAnalyzer={settings.SECURITY_ANALYZER}
-      />
-    </div>
   );
 }
 

frontend/src/routes/_oh.tsx

@@ -21,12 +21,11 @@ import { DangerModal } from "#/components/modals/confirmation-modals/danger-moda
 import { LoadingSpinner } from "#/components/modals/LoadingProject";
 import { ModalBackdrop } from "#/components/modals/modal-backdrop";
 import { UserActions } from "#/components/user-actions";
-import { useSocket } from "#/context/socket";
 import i18n from "#/i18n";
 import { getSettings, settingsAreUpToDate } from "#/services/settings";
 import AllHandsLogo from "#/assets/branding/all-hands-logo.svg?react";
-import NewProjectIcon from "#/assets/new-project.svg?react";
+import NewProjectIcon from "#/icons/new-project.svg?react";
-import DocsIcon from "#/assets/docs.svg?react";
+import DocsIcon from "#/icons/docs.svg?react";
 import { userIsAuthenticated } from "#/utils/user-is-authenticated";
 import { generateGitHubAuthUrl } from "#/utils/generate-github-auth-url";
 import { WaitlistModal } from "#/components/waitlist-modal";
@@ -135,7 +134,6 @@ type SettingsFormData = {
 };
 
 export default function MainApp() {
-  const { stop, isConnected } = useSocket();
   const navigation = useNavigation();
   const location = useLocation();
   const {
@@ -202,14 +200,6 @@ export default function MainApp() {
     }
   }, [user]);
 
-  React.useEffect(() => {
-    if (location.pathname === "/") {
-      // If the user is on the home page, we should stop the socket connection.
-      // This is relevant when the user redirects here for whatever reason.
-      if (isConnected) stop();
-    }
-  }, [location.pathname]);
-
   const handleUserLogout = () => {
     logoutFetcher.submit(
       {},
@@ -313,11 +303,9 @@ export default function MainApp() {
             <p className="text-xs text-[#A3A3A3]">
               To continue, connect an OpenAI, Anthropic, or other LLM account
             </p>
-            {isConnected && (
+            <p className="text-xs text-danger">
-              <p className="text-xs text-danger">
+              Changing settings during an active session will end the session
-                Changing settings during an active session will end the session
+            </p>
-              </p>
-            )}
             <SettingsForm
               settings={settings}
               models={settingsFormData.models}

frontend/src/services/actions.ts

@@ -12,8 +12,11 @@ import {
 import { setCurStatusMessage } from "#/state/statusSlice";
 import store from "#/store";
 import ActionType from "#/types/ActionType";
-import { ActionMessage, StatusMessage } from "#/types/Message";
+import {
-import { SocketMessage } from "#/types/ResponseType";
+  ActionMessage,
+  ObservationMessage,
+  StatusMessage,
+} from "#/types/Message";
 import { handleObservationMessage } from "./observations";
 
 const messageActions = {
@@ -138,22 +141,14 @@ export function handleStatusMessage(message: StatusMessage) {
   }
 }
 
-export function handleAssistantMessage(data: string | SocketMessage) {
+export function handleAssistantMessage(message: Record<string, unknown>) {
-  let socketMessage: SocketMessage;
+  if (message.action) {
-
+    handleActionMessage(message as unknown as ActionMessage);
-  if (typeof data === "string") {
+  } else if (message.observation) {
-    socketMessage = JSON.parse(data) as SocketMessage;
+    handleObservationMessage(message as unknown as ObservationMessage);
+  } else if (message.status_update) {
+    handleStatusMessage(message as unknown as StatusMessage);
   } else {
-    socketMessage = data;
+    console.error("Unknown message type", message);
-  }
-
-  if ("action" in socketMessage) {
-    handleActionMessage(socketMessage);
-  } else if ("observation" in socketMessage) {
-    handleObservationMessage(socketMessage);
-  } else if ("status_update" in socketMessage) {
-    handleStatusMessage(socketMessage);
-  } else {
-    console.error("Unknown message type", socketMessage);
   }
 }

frontend/src/services/agentStateService.ts

@@ -1,8 +1,7 @@
 import ActionType from "#/types/ActionType";
 import AgentState from "#/types/AgentState";
 
-export const generateAgentStateChangeEvent = (state: AgentState) =>
+export const generateAgentStateChangeEvent = (state: AgentState) => ({
-  JSON.stringify({
+  action: ActionType.CHANGE_AGENT_STATE,
-    action: ActionType.CHANGE_AGENT_STATE,
+  args: { agent_state: state },
-    args: { agent_state: state },
+});
-  });

frontend/src/services/api.ts

@@ -63,6 +63,16 @@ export async function request(
   } catch (e) {
     onFail(`Error fetching ${url}`);
   }
+  if (response?.status === 401 && !url.startsWith("/api/authenticate")) {
+    await request(
+      "/api/authenticate",
+      {
+        method: "POST",
+      },
+      true,
+    );
+    return request(url, options, disableToast, returnResponse, maxRetries - 1);
+  }
   if (response?.status && response?.status >= 400) {
     onFail(
       `${response.status} error while fetching ${url}: ${response?.statusText}`,

frontend/src/services/chatService.ts

@@ -9,5 +9,5 @@ export function createChatMessage(
     action: ActionType.MESSAGE,
     args: { content: message, images_urls, timestamp },
   };
-  return JSON.stringify(event);
+  return event;
 }

frontend/src/services/terminalService.ts

@@ -2,7 +2,7 @@ import ActionType from "#/types/ActionType";
 
 export function getTerminalCommand(command: string, hidden: boolean = false) {
   const event = { action: ActionType.RUN, args: { command, hidden } };
-  return JSON.stringify(event);
+  return event;
 }
 
 export function getGitHubTokenCommand(gitHubToken: string) {

frontend/src/utils/cache.ts

@@ -5,26 +5,17 @@ type CacheEntry<T> = {
 };
 
 class Cache {
-  private prefix = "app_cache_";
-
   private defaultTTL = 5 * 60 * 1000; // 5 minutes
 
-  /**
+  private cacheMemory: Record<string, string> = {};
-   * Generate a unique key with prefix for local storage
-   * @param key The key to be stored in local storage
-   * @returns The unique key with prefix
-   */
-  private getKey(key: CacheKey): string {
-    return `${this.prefix}${key}`;
-  }
 
   /**
-   * Retrieve the cached data from local storage
+   * Retrieve the cached data from memory
-   * @param key The key to be retrieved from local storage
+   * @param key The key to be retrieved from memory
-   * @returns The data stored in local storage
+   * @returns The data stored in memory
    */
   public get<T>(key: CacheKey): T | null {
-    const cachedEntry = localStorage.getItem(this.getKey(key));
+    const cachedEntry = this.cacheMemory[key];
     if (cachedEntry) {
       const { data, expiration } = JSON.parse(cachedEntry) as CacheEntry<T>;
       if (Date.now() < expiration) return data;
@@ -35,34 +26,34 @@ class Cache {
   }
 
   /**
-   * Store the data in local storage with expiration
+   * Store the data in memory with expiration
-   * @param key The key to be stored in local storage
+   * @param key The key to be stored in memory
-   * @param data The data to be stored in local storage
+   * @param data The data to be stored in memory
    * @param ttl The time to live for the data in milliseconds
    * @returns void
    */
   public set<T>(key: CacheKey, data: T, ttl = this.defaultTTL): void {
     const expiration = Date.now() + ttl;
     const entry: CacheEntry<T> = { data, expiration };
-    localStorage.setItem(this.getKey(key), JSON.stringify(entry));
+    this.cacheMemory[key] = JSON.stringify(entry);
   }
 
   /**
-   * Remove the data from local storage
+   * Remove the data from memory
-   * @param key The key to be removed from local storage
+   * @param key The key to be removed from memory
    * @returns void
    */
   public delete(key: CacheKey): void {
-    localStorage.removeItem(this.getKey(key));
+    delete this.cacheMemory[key];
   }
 
   /**
-   * Clear all data with the app prefix from local storage
+   * Clear all data
    * @returns void
    */
   public clearAll(): void {
-    Object.keys(localStorage).forEach((key) => {
+    Object.keys(this.cacheMemory).forEach((key) => {
-      if (key.startsWith(this.prefix)) localStorage.removeItem(key);
+      delete this.cacheMemory[key];
     });
   }
 }

frontend/src/utils/organizeModelsAndProviders.ts

@@ -26,6 +26,7 @@ import { extractModelAndProvider } from "./extractModelAndProvider";
  */
 export const organizeModelsAndProviders = (models: string[]) => {
   const object: Record<string, { separator: string; models: string[] }> = {};
+
   models.forEach((model) => {
     const {
       separator,
@@ -45,5 +46,6 @@ export const organizeModelsAndProviders = (models: string[]) => {
     }
     object[key].models.push(modelId);
   });
+
   return object;
 };

frontend/src/utils/verified-models.ts

@@ -1,10 +1,6 @@
 // Here are the list of verified models and providers that we know work well with OpenHands.
 export const VERIFIED_PROVIDERS = ["openai", "azure", "anthropic"];
-export const VERIFIED_MODELS = [
+export const VERIFIED_MODELS = ["gpt-4o", "claude-3-5-sonnet-20241022"];
-  "gpt-4o",
-  "claude-3-5-sonnet-20240620",
-  "claude-3-5-sonnet-20241022",
-];
 
 // LiteLLM does not return OpenAI models with the provider, so we list them here to set them ourselves for consistency
 // (e.g., they return `gpt-4o` instead of `openai/gpt-4o`)
@@ -23,11 +19,9 @@ export const VERIFIED_OPENAI_MODELS = [
 export const VERIFIED_ANTHROPIC_MODELS = [
   "claude-2",
   "claude-2.1",
-  "claude-3-5-sonnet-20241022",
   "claude-3-5-sonnet-20240620",
+  "claude-3-5-sonnet-20241022",
   "claude-3-haiku-20240307",
   "claude-3-opus-20240229",
   "claude-3-sonnet-20240229",
-  "claude-instant-1",
-  "claude-instant-1.2",
 ];

frontend/test-utils.tsx

@@ -6,7 +6,7 @@ import { configureStore } from "@reduxjs/toolkit";
 // eslint-disable-next-line import/no-extraneous-dependencies
 import { RenderOptions, render } from "@testing-library/react";
 import { AppStore, RootState, rootReducer } from "./src/store";
-import { SocketProvider } from "#/context/socket";
+import { WsClientProvider } from "#/context/ws-client-provider";
 
 const setupStore = (preloadedState?: Partial<RootState>): AppStore =>
   configureStore({
@@ -35,7 +35,7 @@ export function renderWithProviders(
   function Wrapper({ children }: PropsWithChildren<object>): JSX.Element {
     return (
       <Provider store={store}>
-        <SocketProvider>{children}</SocketProvider>
+        <WsClientProvider enabled={true} token={null} ghToken={null} settings={null}>{children}</WsClientProvider>
       </Provider>
     );
   }

openhands/agenthub/codeact_agent/codeact_agent.py

@@ -39,7 +39,6 @@ from openhands.runtime.plugins import (
     JupyterRequirement,
     PluginRequirement,
 )
-from openhands.utils.microagent import MicroAgent
 from openhands.utils.prompt import PromptManager
 
 
@@ -86,16 +85,6 @@ class CodeActAgent(Agent):
         super().__init__(llm, config)
         self.reset()
 
-        self.micro_agent = (
-            MicroAgent(
-                os.path.join(
-                    os.path.dirname(__file__), 'micro', f'{config.micro_agent_name}.md'
-                )
-            )
-            if config.micro_agent_name
-            else None
-        )
-
         self.function_calling_active = self.config.function_calling
         if self.function_calling_active and not self.llm.is_function_calling_active():
             logger.warning(
@@ -105,7 +94,6 @@ class CodeActAgent(Agent):
             self.function_calling_active = False
 
         if self.function_calling_active:
-            # Function calling mode
             self.tools = codeact_function_calling.get_tools(
                 codeact_enable_browsing=self.config.codeact_enable_browsing,
                 codeact_enable_jupyter=self.config.codeact_enable_jupyter,
@@ -114,18 +102,17 @@ class CodeActAgent(Agent):
             logger.debug(
                 f'TOOLS loaded for CodeActAgent: {json.dumps(self.tools, indent=2)}'
             )
-            self.system_prompt = codeact_function_calling.SYSTEM_PROMPT
+            self.prompt_manager = PromptManager(
-            self.initial_user_message = None
+                microagent_dir=os.path.join(os.path.dirname(__file__), 'micro'),
+                prompt_dir=os.path.join(os.path.dirname(__file__), 'prompts', 'tools'),
+            )
         else:
-            # Non-function-calling mode
             self.action_parser = CodeActResponseParser()
             self.prompt_manager = PromptManager(
-                prompt_dir=os.path.join(os.path.dirname(__file__)),
+                microagent_dir=os.path.join(os.path.dirname(__file__), 'micro'),
+                prompt_dir=os.path.join(os.path.dirname(__file__), 'prompts', 'default'),
                 agent_skills_docs=AgentSkillsRequirement.documentation,
-                micro_agent=self.micro_agent,
             )
-            self.system_prompt = self.prompt_manager.system_message
-            self.initial_user_message = self.prompt_manager.initial_user_message
 
         self.pending_actions: deque[Action] = deque()
 
@@ -337,8 +324,8 @@ class CodeActAgent(Agent):
             return self.pending_actions.popleft()
 
         # if we're done, go back
-        last_user_message = state.get_last_user_message()
+        latest_user_message = state.get_last_user_message()
-        if last_user_message and last_user_message.strip() == '/exit':
+        if latest_user_message and latest_user_message.content.strip() == '/exit':
             return AgentFinishAction()
 
         # prepare what we want to send to the LLM
@@ -403,17 +390,19 @@ class CodeActAgent(Agent):
                 role='system',
                 content=[
                     TextContent(
-                        text=self.system_prompt,
+                        text=self.prompt_manager.get_system_message(),
-                        cache_prompt=self.llm.is_caching_prompt_active(),  # Cache system prompt
+                        cache_prompt=self.llm.is_caching_prompt_active(),
                     )
                 ],
             )
         ]
-        if self.initial_user_message:
+        example_message = self.prompt_manager.get_example_user_message()
+        if example_message:
             messages.append(
                 Message(
                     role='user',
-                    content=[TextContent(text=self.initial_user_message)],
+                    content=[TextContent(text=example_message)],
+                    cache_prompt=self.llm.is_caching_prompt_active(),
                 )
             )
 
@@ -462,8 +451,9 @@ class CodeActAgent(Agent):
                 pending_tool_call_action_messages.pop(response_id)
 
             for message in messages_to_add:
-                # add regular message
                 if message:
+                    if message.role == 'user':
+                        self.prompt_manager.enhance_message(message)
                     # handle error if the message is the SAME role as the previous message
                     # litellm.exceptions.BadRequestError: litellm.BadRequestError: OpenAIException - Error code: 400 - {'detail': 'Only supports u/a/u/a/u...'}
                     # there shouldn't be two consecutive messages from the same role
@@ -493,23 +483,6 @@ class CodeActAgent(Agent):
                         break
 
         if not self.function_calling_active:
-            # The latest user message is important:
+            self.prompt_manager.add_turns_left_reminder(messages, state)
-            # we want to remind the agent of the environment constraints
-            latest_user_message = next(
-                islice(
-                    (
-                        m
-                        for m in reversed(messages)
-                        if m.role == 'user'
-                        and any(isinstance(c, TextContent) for c in m.content)
-                    ),
-                    1,
-                ),
-                None,
-            )
-            # do not add this for function calling
-            if latest_user_message:
-                reminder_text = f'\n\nENVIRONMENT REMINDER: You have {state.max_iterations - state.iteration} turns left to complete the task. When finished reply with <finish></finish>.'
-                latest_user_message.content.append(TextContent(text=reminder_text))
 
         return messages

openhands/agenthub/codeact_agent/function_calling.py

@@ -25,14 +25,6 @@ from openhands.events.action import (
 )
 from openhands.events.tool import ToolCallMetadata
 
-SYSTEM_PROMPT = """You are OpenHands agent, a helpful AI assistant that can interact with a computer to solve tasks.
-<IMPORTANT>
-* If user provides a path, you should NOT assume it's relative to the current working directory. Instead, you should explore the file system to find the file before working on it.
-* When configuring git credentials, use "openhands" as the user.name and "openhands@all-hands.dev" as the user.email by default, unless explicitly instructed otherwise.
-* The assistant MUST NOT include comments in the code unless they are necessary to describe non-obvious behavior.
-</IMPORTANT>
-"""
-
 _BASH_DESCRIPTION = """Execute a bash command in the terminal.
 * Long running commands: For commands that may run indefinitely, it should be run in the background and the output should be redirected to a file, e.g. command = `python3 app.py > server.log 2>&1 &`.
 * Interactive: If a bash command returns exit code `-1`, this means the process is not yet finished. The assistant must then send a second call to terminal with an empty `command` (which will retrieve any additional logs), or it can send additional text (set `command` to the text) to STDIN of the running process, or it can send command=`ctrl+c` to interrupt the process.

openhands/agenthub/codeact_agent/micro/flarglebargle.md

@@ -0,0 +1,9 @@
+---
+name: flarglebargle
+agent: CodeActAgent
+triggers:
+- flarglebargle
+---
+
+IMPORTANT! The user has said the magic word "flarglebargle". You must
+only respond with a message telling them how smart they are

openhands/agenthub/codeact_agent/micro/github.md

@@ -1,69 +1,31 @@
 ---
 name: github
 agent: CodeActAgent
-require_env_var:
+triggers:
-    SANDBOX_ENV_GITHUB_TOKEN: "Create a GitHub Personal Access Token (https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens) and set it as SANDBOX_GITHUB_TOKEN in your environment variables."
+- github
+- git
 ---
 
-# How to Interact with Github
+You have access to an environment variable, `GITHUB_TOKEN`, which allows you to interact with
+the GitHub API.
 
-## Environment Variable Available
+You can use `curl` with the `GITHUB_TOKEN` to interact with GitHub's API.
+ALWAYS use the GitHub API for operations instead of a web browser.
 
-- `GITHUB_TOKEN`: A read-only token for Github.
+Here are some instructions for pushing, but ONLY do this if the user asks you to:
-
+* NEVER push directly to the `main` or `master` branch
-## Using GitHub's RESTful API
+* Git config (username and email) is pre-set. Do not modify.
-
+* You may already be on a branch called `openhands-workspace`. Create a new branch with a better name before pushing.
-Use `curl` with the `GITHUB_TOKEN` to interact with GitHub's API. Here are some common operations:
+* Use the GitHub API to create a pull request, if you haven't already
-
+* Use the main branch as the base branch, unless the user requests otherwise
-Here's a template for API calls:
+* After opening or updating a pull request, send the user a short message with a link to the pull request.
-
+* Do all of the above in as few steps as possible. E.g. you could open a PR with one step by running the following bash commands:
-```sh
+```bash
-curl -H "Authorization: token $GITHUB_TOKEN" \
+git checkout -b create-widget
-    "https://api.github.com/{endpoint}"
+git add .
+git commit -m "Create widget"
+git push origin create-widget
+curl -X POST "https://api.github.com/repos/CodeActOrg/openhands/pulls" \
+    -H "Authorization: Bearer $GITHUB_TOKEN" \
+    -d '{"title":"Create widget","head":"create-widget","base":"openhands-workspace"}'
 ```
-
-First replace `{endpoint}` with the specific API path. Common operations:
-
-1. View an issue or pull request:
-   - Issues: `/repos/{owner}/{repo}/issues/{issue_number}`
-   - Pull requests: `/repos/{owner}/{repo}/pulls/{pull_request_number}`
-
-2. List repository issues or pull requests:
-   - Issues: `/repos/{owner}/{repo}/issues`
-   - Pull requests: `/repos/{owner}/{repo}/pulls`
-
-3. Search issues or pull requests:
-   - `/search/issues?q=repo:{owner}/{repo}+is:{type}+{search_term}+state:{state}`
-   - Replace `{type}` with `issue` or `pr`
-
-4. List repository branches:
-   `/repos/{owner}/{repo}/branches`
-
-5. Get commit details:
-   `/repos/{owner}/{repo}/commits/{commit_sha}`
-
-6. Get repository details:
-   `/repos/{owner}/{repo}`
-
-7. Get user information:
-   `/user`
-
-8. Search repositories:
-   `/search/repositories?q={query}`
-
-9. Get rate limit status:
-   `/rate_limit`
-
-Replace `{owner}`, `{repo}`, `{commit_sha}`, `{issue_number}`, `{pull_request_number}`,
-`{search_term}`, `{state}`, and `{query}` with appropriate values.
-
-## Important Notes
-
-1. Always use the GitHub API for operations instead of a web browser.
-2. The `GITHUB_TOKEN` is read-only. Avoid operations that require write access.
-3. Git config (username and email) is pre-set. Do not modify.
-4. Edit and test code locally. Never push directly to remote.
-5. Verify correct branch before committing.
-6. Commit changes frequently.
-7. If the issue or task is ambiguous or lacks sufficient detail, always request clarification from the user before proceeding.
-8. You should avoid using command line tools like `sed` for file editing.

openhands/agenthub/codeact_agent/prompts/default/user_prompt.j2

@@ -215,12 +215,5 @@ The server is running on port 5000 with PID 126. You can access the list of numb
 {% endset %}
 Here is an example of how you can interact with the environment for task solving:
 {{ DEFAULT_EXAMPLE }}
-{% if micro_agent %}
---- BEGIN OF GUIDELINE ---
-The following information may assist you in completing your task:
-
-{{ micro_agent }}
---- END OF GUIDELINE ---
-{% endif %}
 
 NOW, LET'S START!

openhands/agenthub/codeact_agent/prompts/tools/system_prompt.j2

@@ -0,0 +1,7 @@
+You are OpenHands agent, a helpful AI assistant that can interact with a computer to solve tasks.
+<IMPORTANT>
+* If user provides a path, you should NOT assume it's relative to the current working directory. Instead, you should explore the file system to find the file before working on it.
+* When configuring git credentials, use "openhands" as the user.name and "openhands@all-hands.dev" as the user.email by default, unless explicitly instructed otherwise.
+* The assistant MUST NOT include comments in the code unless they are necessary to describe non-obvious behavior.
+</IMPORTANT>
+

openhands/agenthub/codeact_swe_agent/codeact_swe_agent.py

@@ -155,7 +155,7 @@ class CodeActSWEAgent(Agent):
         """
         # if we're done, go back
         last_user_message = state.get_last_user_message()
-        if last_user_message and last_user_message.strip() == '/exit':
+        if last_user_message and last_user_message.content.strip() == '/exit':
             return AgentFinishAction()
 
         # prepare what we want to send to the LLM

openhands/controller/agent_controller.py

@@ -1,5 +1,6 @@
 import asyncio
 import copy
+import os
 import traceback
 from typing import Callable, ClassVar, Type
 
@@ -259,7 +260,11 @@ class AgentController:
             observation_to_print.content = truncate_content(
                 observation_to_print.content, self.agent.llm.config.max_message_chars
             )
-        self.log('debug', str(observation_to_print), extra={'msg_type': 'OBSERVATION'})
+        # Use info level if LOG_ALL_EVENTS is set
+        log_level = 'info' if os.getenv('LOG_ALL_EVENTS') in ('true', '1') else 'debug'
+        self.log(
+            log_level, str(observation_to_print), extra={'msg_type': 'OBSERVATION'}
+        )
 
         if observation.llm_metrics is not None:
             self.agent.llm.metrics.merge(observation.llm_metrics)
@@ -282,8 +287,12 @@ class AgentController:
             action (MessageAction): The message action to handle.
         """
         if action.source == EventSource.USER:
+            # Use info level if LOG_ALL_EVENTS is set
+            log_level = (
+                'info' if os.getenv('LOG_ALL_EVENTS') in ('true', '1') else 'debug'
+            )
             self.log(
-                'debug',
+                log_level,
                 str(action),
                 extra={'msg_type': 'ACTION', 'event_source': EventSource.USER},
             )
@@ -497,7 +506,9 @@ class AgentController:
 
         await self.update_state_after_step()
 
-        self.log('debug', str(action), extra={'msg_type': 'ACTION'})
+        # Use info level if LOG_ALL_EVENTS is set
+        log_level = 'info' if os.getenv('LOG_ALL_EVENTS') in ('true', '1') else 'debug'
+        self.log(log_level, str(action), extra={'msg_type': 'ACTION'})
 
     async def _delegate_step(self):
         """Executes a single step of the delegate agent."""
@@ -663,7 +674,7 @@ class AgentController:
         # sanity check
         if start_id > end_id + 1:
             self.log(
-                'debug',
+                'warning',
                 f'start_id {start_id} is greater than end_id + 1 ({end_id + 1}). History will be empty.',
             )
             self.state.history = []
@@ -694,7 +705,7 @@ class AgentController:
                 # Match with most recent unmatched delegate action
                 if not delegate_action_ids:
                     self.log(
-                        'error',
+                        'warning',
                         f'Found AgentDelegateObservation without matching action at id={event.id}',
                     )
                     continue

openhands/controller/state/state.py

@@ -156,14 +156,14 @@ class State:
 
         return last_user_message, last_user_message_image_urls
 
-    def get_last_agent_message(self) -> str | None:
+    def get_last_agent_message(self) -> MessageAction | None:
         for event in reversed(self.history):
             if isinstance(event, MessageAction) and event.source == EventSource.AGENT:
-                return event.content
+                return event
         return None
 
-    def get_last_user_message(self) -> str | None:
+    def get_last_user_message(self) -> MessageAction | None:
         for event in reversed(self.history):
             if isinstance(event, MessageAction) and event.source == EventSource.USER:
-                return event.content
+                return event
         return None

openhands/core/config/app_config.py

@@ -69,6 +69,7 @@ class AppConfig:
     file_uploads_max_file_size_mb: int = 0
     file_uploads_restrict_file_types: bool = False
     file_uploads_allowed_extensions: list[str] = field(default_factory=lambda: ['.*'])
+    runloop_api_key: str | None = None
 
     defaults_dict: ClassVar[dict] = {}
 
@@ -139,6 +140,7 @@ class AppConfig:
                 'jwt_secret',
                 'modal_api_token_id',
                 'modal_api_token_secret',
+                'runloop_api_key',
             ]:
                 attr_value = '******' if attr_value else None
 

openhands/core/config/sandbox_config.py

@@ -14,7 +14,8 @@ class SandboxConfig:
         base_container_image: The base container image from which to build the runtime image.
         runtime_container_image: The runtime container image to use.
         user_id: The user ID for the sandbox.
-        timeout: The timeout for the sandbox.
+        timeout: The timeout for the default sandbox action execution.
+        remote_runtime_init_timeout: The timeout for the remote runtime to start.
         enable_auto_lint: Whether to enable auto-lint.
         use_host_network: Whether to use the host network.
         initialize_plugins: Whether to initialize plugins.
@@ -41,6 +42,7 @@ class SandboxConfig:
     runtime_container_image: str | None = None
     user_id: int = os.getuid() if hasattr(os, 'getuid') else 1000
     timeout: int = 120
+    remote_runtime_init_timeout: int = 180
     enable_auto_lint: bool = (
         False  # once enabled, OpenHands would lint files after editing
     )

openhands/core/logger.py

@@ -177,7 +177,7 @@ class SensitiveDataFilter(logging.Filter):
         return True
 
 
-def get_console_handler(log_level=logging.INFO, extra_info: str | None = None):
+def get_console_handler(log_level: int = logging.INFO, extra_info: str | None = None):
     """Returns a console handler for logging."""
     console_handler = logging.StreamHandler()
     console_handler.setLevel(log_level)
@@ -188,7 +188,7 @@ def get_console_handler(log_level=logging.INFO, extra_info: str | None = None):
     return console_handler
 
 
-def get_file_handler(log_dir, log_level=logging.INFO):
+def get_file_handler(log_dir: str, log_level: int = logging.INFO):
     """Returns a file handler for logging."""
     os.makedirs(log_dir, exist_ok=True)
     timestamp = datetime.now().strftime('%Y-%m-%d')

openhands/events/observation/browse.py

@@ -64,7 +64,7 @@ class BrowserOutputObservation(Observation):
             # We do not filter visible only here because we want to show the full content
             # of the web page to the agent for simplicity.
             # FIXME: handle the case when the web page is too large
-            cur_axtree_txt = self.get_axtree_str(filter_visible_only=True)
+            cur_axtree_txt = self.get_axtree_str(filter_visible_only=False)
             text += (
                 f'============== BEGIN accessibility tree ==============\n'
                 f'{cur_axtree_txt}\n'
@@ -74,12 +74,12 @@ class BrowserOutputObservation(Observation):
             text += f'\n[Error encountered when processing the accessibility tree: {e}]'
         return text
 
-    def get_axtree_str(self, filter_visible_only: bool = True) -> str:
+    def get_axtree_str(self, filter_visible_only: bool = False) -> str:
         cur_axtree_txt = flatten_axtree_to_str(
             self.axtree_object,
             extra_properties=self.extra_element_properties,
             with_clickable=True,
-            skip_generic=True,
+            skip_generic=False,
             filter_visible_only=filter_visible_only,
         )
         self._axtree_str = cur_axtree_txt

openhands/runtime/__init__.py

@@ -23,6 +23,10 @@ def get_runtime_cls(name: str):
         from openhands.runtime.impl.modal.modal_runtime import ModalRuntime
 
         return ModalRuntime
+    elif name == 'runloop':
+        from openhands.runtime.impl.runloop.runloop_runtime import RunloopRuntime
+
+        return RunloopRuntime
     else:
         raise ValueError(f'Runtime {name} not supported')
 

openhands/runtime/action_execution_server.py

@@ -7,7 +7,9 @@ NOTE: this will be executed inside the docker sandbox.
 
 import argparse
 import asyncio
+import base64
 import io
+import mimetypes
 import os
 import shutil
 import tempfile
@@ -217,6 +219,33 @@ class ActionExecutor:
         working_dir = self.bash_session.workdir
         filepath = self._resolve_path(action.path, working_dir)
         try:
+            if filepath.lower().endswith(('.png', '.jpg', '.jpeg', '.bmp', '.gif')):
+                with open(filepath, 'rb') as file:
+                    image_data = file.read()
+                    encoded_image = base64.b64encode(image_data).decode('utf-8')
+                    mime_type, _ = mimetypes.guess_type(filepath)
+                    if mime_type is None:
+                        mime_type = 'image/png'  # default to PNG if mime type cannot be determined
+                    encoded_image = f'data:{mime_type};base64,{encoded_image}'
+
+                return FileReadObservation(path=filepath, content=encoded_image)
+            elif filepath.lower().endswith('.pdf'):
+                with open(filepath, 'rb') as file:
+                    pdf_data = file.read()
+                    encoded_pdf = base64.b64encode(pdf_data).decode('utf-8')
+                    encoded_pdf = f'data:application/pdf;base64,{encoded_pdf}'
+                return FileReadObservation(path=filepath, content=encoded_pdf)
+            elif filepath.lower().endswith(('.mp4', '.webm', '.ogg')):
+                with open(filepath, 'rb') as file:
+                    video_data = file.read()
+                    encoded_video = base64.b64encode(video_data).decode('utf-8')
+                    mime_type, _ = mimetypes.guess_type(filepath)
+                    if mime_type is None:
+                        mime_type = 'video/mp4'  # default to MP4 if MIME type cannot be determined
+                    encoded_video = f'data:{mime_type};base64,{encoded_video}'
+
+                return FileReadObservation(path=filepath, content=encoded_video)
+
             with open(filepath, 'r', encoding='utf-8') as file:
                 lines = read_lines(file.readlines(), action.start, action.end)
         except FileNotFoundError:

openhands/runtime/impl/remote/remote_runtime.py

@@ -1,7 +1,7 @@
 import os
-from pathlib import Path
 import tempfile
 import threading
+from pathlib import Path
 from typing import Callable, Optional
 from zipfile import ZipFile
 
@@ -137,7 +137,7 @@ class RemoteRuntime(Runtime):
         try:
             response = self._send_request(
                 'GET',
-                f'{self.config.sandbox.remote_runtime_api_url}/runtime/{self.sid}',
+                f'{self.config.sandbox.remote_runtime_api_url}/sessions/{self.sid}',
                 timeout=5,
             )
         except requests.HTTPError as e:
@@ -227,7 +227,7 @@ class RemoteRuntime(Runtime):
             'command': command,
             'working_dir': '/openhands/code/',
             'environment': {'DEBUG': 'true'} if self.config.debug else {},
-            'runtime_id': self.sid,
+            'session_id': self.sid,
         }
 
         # Start the sandbox using the /start endpoint
@@ -260,17 +260,23 @@ class RemoteRuntime(Runtime):
                 {'X-Session-API-Key': start_response['session_api_key']}
             )
 
-    @tenacity.retry(
-        stop=tenacity.stop_after_delay(180) | stop_if_should_exit(),
-        reraise=True,
-        retry=tenacity.retry_if_exception_type(RuntimeNotReadyError),
-        wait=tenacity.wait_fixed(2),
-    )
     def _wait_until_alive(self):
+        retry_decorator = tenacity.retry(
+            stop=tenacity.stop_after_delay(
+                self.config.sandbox.remote_runtime_init_timeout
+            )
+            | stop_if_should_exit(),
+            reraise=True,
+            retry=tenacity.retry_if_exception_type(RuntimeNotReadyError),
+            wait=tenacity.wait_fixed(2),
+        )
+        return retry_decorator(self._wait_until_alive_impl)()
+
+    def _wait_until_alive_impl(self):
         self.log('debug', f'Waiting for runtime to be alive at url: {self.runtime_url}')
         runtime_info_response = self._send_request(
             'GET',
-            f'{self.config.sandbox.remote_runtime_api_url}/runtime/{self.runtime_id}',
+            f'{self.config.sandbox.remote_runtime_api_url}/sessions/{self.sid}',
         )
         runtime_data = runtime_info_response.json()
         assert 'runtime_id' in runtime_data

openhands/runtime/impl/runloop/README.md

@@ -0,0 +1,31 @@
+# Runloop Runtime
+Runloop provides a fast, secure and scalable AI sandbox (Devbox).
+ Check out the [runloop docs](https://docs.runloop.ai/overview/what-is-runloop)
+for more detail
+
+## Access
+Runloop is currently available in a closed beta. For early access, or
+just to say hello, sign up at https://www.runloop.ai/hello
+
+## Set up
+With your runloop API,
+```bash
+export RUNLOOP_API_KEY=<your-api-key>
+```
+
+Configure the runtime
+```bash
+export RUNTIME="runloop"
+```
+
+## Interact with your devbox
+Runloop provides additional tools to interact with your Devbox based
+runtime environment. See the [docs](https://docs.runloop.ai/tools) for an up
+to date list of tools.
+
+### Dashboard
+View logs, ssh into, or view your Devbox status from the [dashboard](https://platform.runloop.ai)
+
+### CLI
+Use the Runloop CLI to view logs, execute commands, and more.
+See the setup instructions [here](https://docs.runloop.ai/tools/cli)

openhands/runtime/impl/runloop/runloop_runtime.py

@@ -0,0 +1,272 @@
+import logging
+import threading
+import time
+from typing import Callable
+
+import requests
+import tenacity
+from runloop_api_client import Runloop
+from runloop_api_client.types import DevboxView
+from runloop_api_client.types.shared_params import LaunchParameters
+
+from openhands.core.config import AppConfig
+from openhands.core.logger import openhands_logger as logger
+from openhands.events import EventStream
+from openhands.runtime.impl.eventstream.eventstream_runtime import (
+    EventStreamRuntime,
+    LogBuffer,
+)
+from openhands.runtime.plugins import PluginRequirement
+from openhands.runtime.utils.command import get_remote_startup_command
+from openhands.runtime.utils.request import send_request
+from openhands.utils.tenacity_stop import stop_if_should_exit
+
+
+class RunloopLogBuffer(LogBuffer):
+    """Synchronous buffer for Runloop devbox logs.
+
+    This class provides a thread-safe way to collect, store, and retrieve logs
+    from a Docker container. It uses a list to store log lines and provides methods
+    for appending, retrieving, and clearing logs.
+    """
+
+    def __init__(self, runloop_api_client: Runloop, devbox_id: str):
+        self.client_ready = False
+        self.init_msg = 'Runtime client initialized.'
+
+        self.buffer: list[str] = []
+        self.lock = threading.Lock()
+        self._stop_event = threading.Event()
+        self.runloop_api_client = runloop_api_client
+        self.devbox_id = devbox_id
+        self.log_index = 0
+        self.log_stream_thread = threading.Thread(target=self.stream_logs)
+        self.log_stream_thread.daemon = True
+        self.log_stream_thread.start()
+
+    def stream_logs(self):
+        """Stream logs from the Docker container in a separate thread.
+
+        This method runs in its own thread to handle the blocking
+        operation of reading log lines from the Docker SDK's synchronous generator.
+        """
+
+        try:
+            # TODO(Runloop) Replace with stream
+            while True:
+                raw_logs = self.runloop_api_client.devboxes.logs.list(
+                    self.devbox_id
+                ).logs[self.log_index :]
+                logs = [
+                    log.message
+                    for log in raw_logs
+                    if log.message and log.cmd_id is None
+                ]
+
+                self.log_index += len(raw_logs)
+                if self._stop_event.is_set():
+                    break
+                if logs:
+                    for log_line in logs:
+                        self.append(log_line)
+                        if self.init_msg in log_line:
+                            self.client_ready = True
+
+                time.sleep(1)
+        except Exception as e:
+            logger.error(f'Error streaming runloop logs: {e}')
+
+    # NB: Match LogBuffer behavior on below methods
+
+    def get_and_clear(self) -> list[str]:
+        with self.lock:
+            logs = list(self.buffer)
+            self.buffer.clear()
+            return logs
+
+    def append(self, log_line: str):
+        with self.lock:
+            self.buffer.append(log_line)
+
+    def close(self, timeout: float = 5.0):
+        self._stop_event.set()
+        self.log_stream_thread.join(timeout)
+
+
+class RunloopRuntime(EventStreamRuntime):
+    """The RunloopRuntime class is an EventStreamRuntime that utilizes Runloop Devbox as a runtime environment."""
+
+    _sandbox_port: int = 4444
+
+    def __init__(
+        self,
+        config: AppConfig,
+        event_stream: EventStream,
+        sid: str = 'default',
+        plugins: list[PluginRequirement] | None = None,
+        env_vars: dict[str, str] | None = None,
+        status_callback: Callable | None = None,
+        attach_to_existing: bool = False,
+    ):
+        assert config.runloop_api_key is not None, 'Runloop API key is required'
+        self.devbox: DevboxView | None = None
+        self.config = config
+        self.runloop_api_client = Runloop(
+            bearer_token=config.runloop_api_key,
+        )
+        self.session = requests.Session()
+        self.container_name = self.container_name_prefix + sid
+        self.action_semaphore = threading.Semaphore(1)  # Ensure one action at a time
+        self.init_base_runtime(
+            config,
+            event_stream,
+            sid,
+            plugins,
+            env_vars,
+            status_callback,
+            attach_to_existing,
+        )
+        # Buffer for container logs
+        self.log_buffer: LogBuffer | None = None
+
+    @tenacity.retry(
+        stop=tenacity.stop_after_attempt(120),
+        wait=tenacity.wait_fixed(1),
+    )
+    def _wait_for_devbox(self, devbox: DevboxView) -> DevboxView:
+        """Pull devbox status until it is running"""
+        if devbox == 'running':
+            return devbox
+
+        devbox = self.runloop_api_client.devboxes.retrieve(id=devbox.id)
+        if devbox.status != 'running':
+            raise ConnectionRefusedError('Devbox is not running')
+
+        # Devbox is connected and running
+        logging.debug(f'devbox.id={devbox.id} is running')
+        return devbox
+
+    def _create_new_devbox(self) -> DevboxView:
+        # Note: Runloop connect
+        sandbox_workspace_dir = self.config.workspace_mount_path_in_sandbox
+        plugin_args = []
+        if self.plugins is not None and len(self.plugins) > 0:
+            plugin_args.append('--plugins')
+            plugin_args.extend([plugin.name for plugin in self.plugins])
+
+        browsergym_args = []
+        if self.config.sandbox.browsergym_eval_env is not None:
+            browsergym_args = [
+                '-browsergym-eval-env',
+                self.config.sandbox.browsergym_eval_env,
+            ]
+
+        # Copied from EventstreamRuntime
+        start_command = get_remote_startup_command(
+            self._sandbox_port,
+            sandbox_workspace_dir,
+            'openhands' if self.config.run_as_openhands else 'root',
+            self.config.sandbox.user_id,
+            plugin_args,
+            browsergym_args,
+        )
+
+        # Add some additional commands based on our image
+        # NB: start off as root, action_execution_server will ultimately choose user but expects all context
+        # (ie browser) to be installed as root
+        start_command = (
+            'export MAMBA_ROOT_PREFIX=/openhands/micromamba && '
+            'cd /openhands/code && '
+            + '/openhands/micromamba/bin/micromamba run -n openhands poetry config virtualenvs.path /openhands/poetry && '
+            + ' '.join(start_command)
+        )
+        entrypoint = f"sudo bash -c '{start_command}'"
+
+        devbox = self.runloop_api_client.devboxes.create(
+            entrypoint=entrypoint,
+            setup_commands=[f'mkdir -p {self.config.workspace_mount_path_in_sandbox}'],
+            name=self.sid,
+            environment_variables={'DEBUG': 'true'} if self.config.debug else {},
+            prebuilt='openhands',
+            launch_parameters=LaunchParameters(
+                available_ports=[self._sandbox_port],
+                resource_size_request="LARGE",
+            ),
+            metadata={'container-name': self.container_name},
+        )
+        return self._wait_for_devbox(devbox)
+
+    async def connect(self):
+        self.send_status_message('STATUS$STARTING_RUNTIME')
+
+        if self.attach_to_existing:
+            active_devboxes = self.runloop_api_client.devboxes.list(
+                status='running'
+            ).devboxes
+            self.devbox = next(
+                (devbox for devbox in active_devboxes if devbox.name == self.sid), None
+            )
+
+        if self.devbox is None:
+            self.devbox = self._create_new_devbox()
+
+        # Create tunnel - this will return a stable url, so is safe to call if we are attaching to existing
+        tunnel = self.runloop_api_client.devboxes.create_tunnel(
+            id=self.devbox.id,
+            port=self._sandbox_port,
+        )
+
+        # Hook up logs
+        self.log_buffer = RunloopLogBuffer(self.runloop_api_client, self.devbox.id)
+        self.api_url = f'https://{tunnel.url}'
+        logger.info(f'Container started. Server url: {self.api_url}')
+
+        # End Runloop connect
+        # NOTE: Copied from EventStreamRuntime
+        logger.info('Waiting for client to become ready...')
+        self.send_status_message('STATUS$WAITING_FOR_CLIENT')
+        self._wait_until_alive()
+
+        if not self.attach_to_existing:
+            self.setup_initial_env()
+
+        logger.info(
+            f'Container initialized with plugins: {[plugin.name for plugin in self.plugins]}'
+        )
+        self.send_status_message(' ')
+
+    @tenacity.retry(
+        stop=tenacity.stop_after_delay(120) | stop_if_should_exit(),
+        wait=tenacity.wait_fixed(1),
+        reraise=(ConnectionRefusedError,),
+    )
+    def _wait_until_alive(self):
+        # NB(Runloop): Remote logs are not guaranteed realtime, removing client_ready check from logs
+        self._refresh_logs()
+        if not self.log_buffer:
+            raise RuntimeError('Runtime client is not ready.')
+        response = send_request(
+            self.session,
+            'GET',
+            f'{self.api_url}/alive',
+            timeout=5,
+        )
+        if response.status_code == 200:
+            return
+        else:
+            msg = f'Action execution API is not alive. Response: {response}'
+            logger.error(msg)
+            raise RuntimeError(msg)
+
+    def close(self, rm_all_containers: bool = True):
+        if self.log_buffer:
+            self.log_buffer.close()
+
+        if self.session:
+            self.session.close()
+
+        if self.attach_to_existing:
+            return
+
+        if self.devbox:
+            self.runloop_api_client.devboxes.shutdown(self.devbox.id)

openhands/server/github.py

@@ -1,10 +1,12 @@
 import os
 
-import httpx
+from github import Github
+from github.GithubException import GithubException
 from tenacity import retry, stop_after_attempt, wait_exponential
 
 from openhands.core.logger import openhands_logger as logger
 from openhands.server.sheets_client import GoogleSheetsClient
+from openhands.utils.async_utils import call_sync_from_async
 
 GITHUB_CLIENT_ID = os.getenv('GITHUB_CLIENT_ID', '').strip()
 GITHUB_CLIENT_SECRET = os.getenv('GITHUB_CLIENT_SECRET', '').strip()
@@ -12,7 +14,7 @@ GITHUB_CLIENT_SECRET = os.getenv('GITHUB_CLIENT_SECRET', '').strip()
 
 class UserVerifier:
     def __init__(self) -> None:
-        logger.info('Initializing UserVerifier')
+        logger.debug('Initializing UserVerifier')
         self.file_users: list[str] | None = None
         self.sheets_client: GoogleSheetsClient | None = None
         self.spreadsheet_id: str | None = None
@@ -25,7 +27,7 @@ class UserVerifier:
         """Load users from text file if configured"""
         waitlist = os.getenv('GITHUB_USER_LIST_FILE')
         if not waitlist:
-            logger.info('GITHUB_USER_LIST_FILE not configured')
+            logger.debug('GITHUB_USER_LIST_FILE not configured')
             return
 
         if not os.path.exists(waitlist):
@@ -46,10 +48,10 @@ class UserVerifier:
         sheet_id = os.getenv('GITHUB_USERS_SHEET_ID')
 
         if not sheet_id:
-            logger.info('GITHUB_USERS_SHEET_ID not configured')
+            logger.debug('GITHUB_USERS_SHEET_ID not configured')
             return
 
-        logger.info('Initializing Google Sheets integration')
+        logger.debug('Initializing Google Sheets integration')
         self.sheets_client = GoogleSheetsClient()
         self.spreadsheet_id = sheet_id
 
@@ -61,21 +63,21 @@ class UserVerifier:
         if not self.is_active():
             return True
 
-        logger.info(f'Checking if GitHub user {username} is allowed')
+        logger.debug(f'Checking if GitHub user {username} is allowed')
         if self.file_users:
             if username in self.file_users:
-                logger.info(f'User {username} found in text file allowlist')
+                logger.debug(f'User {username} found in text file allowlist')
                 return True
             logger.debug(f'User {username} not found in text file allowlist')
 
         if self.sheets_client and self.spreadsheet_id:
             sheet_users = self.sheets_client.get_usernames(self.spreadsheet_id)
             if username in sheet_users:
-                logger.info(f'User {username} found in Google Sheets allowlist')
+                logger.debug(f'User {username} found in Google Sheets allowlist')
                 return True
             logger.debug(f'User {username} not found in Google Sheets allowlist')
 
-        logger.info(f'User {username} not found in any allowlist')
+        logger.debug(f'User {username} not found in any allowlist')
         return False
 
 
@@ -83,10 +85,10 @@ async def authenticate_github_user(auth_token) -> bool:
     user_verifier = UserVerifier()
 
     if not user_verifier.is_active():
-        logger.info('No user verification sources configured - allowing all users')
+        logger.debug('No user verification sources configured - allowing all users')
         return True
 
-    logger.info('Checking GitHub token')
+    logger.debug('Checking GitHub token')
 
     if not auth_token:
         logger.warning('No GitHub token provided')
@@ -112,25 +114,14 @@ async def get_github_user(token: str) -> str:
     Returns:
         github handle of the user
     """
-    logger.info('Fetching GitHub user info from token')
+    logger.debug('Fetching GitHub user info from token')
-    headers = {
+    try:
-        'Accept': 'application/vnd.github+json',
+        g = Github(token)
-        'Authorization': f'Bearer {token}',
+        user = await call_sync_from_async(g.get_user)
-    }
+        login = user.login
-    async with httpx.AsyncClient(
-        timeout=httpx.Timeout(connect=5.0, read=5.0, write=5.0, pool=5.0)
-    ) as client:
-        try:
-            response = await client.get('https://api.github.com/user', headers=headers)
-        except httpx.RequestError as e:
-            logger.error(f'Error making request to GitHub API: {str(e)}')
-            logger.error(e)
-            raise
-
-        logger.info('Received response from GitHub API')
-        logger.debug(f'Response status code: {response.status_code}')
-        response.raise_for_status()
-        user_data = response.json()
-        login = user_data.get('login')
         logger.info(f'Successfully retrieved GitHub user: {login}')
         return login
+    except GithubException as e:
+        logger.error(f'Error making request to GitHub API: {str(e)}')
+        logger.error(e)
+        raise

openhands/server/listen.py

@@ -2,10 +2,12 @@ import asyncio
 import os
 import re
 import tempfile
+import time
 import uuid
 import warnings
 from contextlib import asynccontextmanager
 
+import jwt
 import requests
 from pathspec import PathSpec
 from pathspec.patterns import GitWildMatchPattern
@@ -15,6 +17,7 @@ from openhands.server.data_models.feedback import FeedbackDataModel, store_feedb
 from openhands.server.github import (
     GITHUB_CLIENT_ID,
     GITHUB_CLIENT_SECRET,
+    UserVerifier,
     authenticate_github_user,
 )
 from openhands.storage import get_file_store
@@ -60,7 +63,7 @@ from openhands.events.serialization import event_to_dict
 from openhands.events.stream import AsyncEventStreamWrapper
 from openhands.llm import bedrock
 from openhands.runtime.base import Runtime
-from openhands.server.auth import get_sid_from_token, sign_token
+from openhands.server.auth.auth import get_sid_from_token, sign_token
 from openhands.server.middleware import LocalhostCORSMiddleware, NoCacheMiddleware
 from openhands.server.session import SessionManager
 
@@ -204,12 +207,22 @@ async def attach_session(request: Request, call_next):
         response = await call_next(request)
         return response
 
-    github_token = request.headers.get('X-GitHub-Token')
+    user_verifier = UserVerifier()
-    if not await authenticate_github_user(github_token):
+    if user_verifier.is_active():
-        return JSONResponse(
+        signed_token = request.cookies.get('github_auth')
-            status_code=status.HTTP_401_UNAUTHORIZED,
+        if not signed_token:
-            content={'error': 'Not authenticated'},
+            return JSONResponse(
-        )
+                status_code=status.HTTP_401_UNAUTHORIZED,
+                content={'error': 'Not authenticated'},
+            )
+        try:
+            jwt.decode(signed_token, config.jwt_secret, algorithms=['HS256'])
+        except Exception as e:
+            logger.warning(f'Invalid token: {e}')
+            return JSONResponse(
+                status_code=status.HTTP_401_UNAUTHORIZED,
+                content={'error': 'Invalid token'},
+            )
 
     if not request.headers.get('Authorization'):
         logger.warning('Missing Authorization header')
@@ -864,10 +877,26 @@ async def authenticate(request: Request):
             content={'error': 'Not authorized via GitHub waitlist'},
         )
 
+    # Create a signed JWT token with 1-hour expiration
+    cookie_data = {
+        'github_token': token,
+        'exp': int(time.time()) + 3600,  # 1 hour expiration
+    }
+    signed_token = sign_token(cookie_data, config.jwt_secret)
+
     response = JSONResponse(
         status_code=status.HTTP_200_OK, content={'message': 'User authenticated'}
     )
 
+    # Set secure cookie with signed token
+    response.set_cookie(
+        key='github_auth',
+        value=signed_token,
+        max_age=3600,  # 1 hour in seconds
+        httponly=True,
+        secure=True,
+        samesite='strict',
+    )
     return response
 
 

openhands/utils/microagent.py

@@ -3,15 +3,11 @@ import os
 import frontmatter
 import pydantic
 
-from openhands.controller.agent import Agent
-from openhands.core.exceptions import MicroAgentValidationError
-from openhands.core.logger import openhands_logger as logger
-
 
 class MicroAgentMetadata(pydantic.BaseModel):
     name: str
     agent: str
-    require_env_var: dict[str, str]
+    triggers: list[str] = []
 
 
 class MicroAgent:
@@ -23,22 +19,30 @@ class MicroAgent:
             self._loaded = frontmatter.load(file)
             self._content = self._loaded.content
             self._metadata = MicroAgentMetadata(**self._loaded.metadata)
-        self._validate_micro_agent()
+
+    def get_trigger(self, message: str) -> str | None:
+        message = message.lower()
+        for trigger in self.triggers:
+            if trigger.lower() in message:
+                return trigger
+        return None
 
     @property
     def content(self) -> str:
         return self._content
 
-    def _validate_micro_agent(self):
+    @property
-        logger.debug(
+    def metadata(self) -> MicroAgentMetadata:
-            f'Loading and validating micro agent [{self._metadata.name}] based on [{self._metadata.agent}]'
+        return self._metadata
-        )
+
-        # Make sure the agent is registered
+    @property
-        agent_cls = Agent.get_cls(self._metadata.agent)
+    def name(self) -> str:
-        assert agent_cls is not None
+        return self._metadata.name
-        # Make sure the environment variables are set
+
-        for env_var, instruction in self._metadata.require_env_var.items():
+    @property
-            if env_var not in os.environ:
+    def triggers(self) -> list[str]:
-                raise MicroAgentValidationError(
+        return self._metadata.triggers
-                    f'Environment variable [{env_var}] is required by micro agent [{self._metadata.name}] but not set. {instruction}'
+
-                )
+    @property
+    def agent(self) -> str:
+        return self._metadata.agent

openhands/utils/prompt.py

@@ -1,7 +1,10 @@
 import os
+from itertools import islice
 
 from jinja2 import Template
 
+from openhands.controller.state.state import State
+from openhands.core.message import Message, TextContent
 from openhands.utils.microagent import MicroAgent
 
 
@@ -16,21 +19,31 @@ class PromptManager:
     Attributes:
         prompt_dir (str): Directory containing prompt templates.
         agent_skills_docs (str): Documentation of agent skills.
-        micro_agent (MicroAgent | None): Micro-agent, if specified.
     """
 
     def __init__(
         self,
         prompt_dir: str,
-        agent_skills_docs: str,
+        microagent_dir: str = '',
-        micro_agent: MicroAgent | None = None,
+        agent_skills_docs: str = '',
     ):
         self.prompt_dir: str = prompt_dir
         self.agent_skills_docs: str = agent_skills_docs
 
         self.system_template: Template = self._load_template('system_prompt')
         self.user_template: Template = self._load_template('user_prompt')
-        self.micro_agent: MicroAgent | None = micro_agent
+        self.microagents: dict = {}
+
+        microagent_files = []
+        if microagent_dir:
+            microagent_files = [
+                os.path.join(microagent_dir, f)
+                for f in os.listdir(microagent_dir)
+                if f.endswith('.md')
+            ]
+        for microagent_file in microagent_files:
+            microagent = MicroAgent(microagent_file)
+            self.microagents[microagent.name] = microagent
 
     def _load_template(self, template_name: str) -> Template:
         template_path = os.path.join(self.prompt_dir, f'{template_name}.j2')
@@ -39,15 +52,13 @@ class PromptManager:
         with open(template_path, 'r') as file:
             return Template(file.read())
 
-    @property
+    def get_system_message(self) -> str:
-    def system_message(self) -> str:
         rendered = self.system_template.render(
             agent_skills_docs=self.agent_skills_docs,
         ).strip()
         return rendered
 
-    @property
+    def get_example_user_message(self) -> str:
-    def initial_user_message(self) -> str:
         """This is the initial user message provided to the agent
         before *actual* user instructions are provided.
 
@@ -57,7 +68,39 @@ class PromptManager:
         These additional context will convert the current generic agent
         into a more specialized agent that is tailored to the user's task.
         """
-        rendered = self.user_template.render(
+        return self.user_template.render().strip()
-            micro_agent=self.micro_agent.content if self.micro_agent else None
+
+    def enhance_message(self, message: Message) -> None:
+        """Enhance the user message with additional context.
+
+        This method is used to enhance the user message with additional context
+        about the user's task. The additional context will convert the current
+        generic agent into a more specialized agent that is tailored to the user's task.
+        """
+        if not message.content:
+            return
+        message_content = message.content[0].text
+        for microagent in self.microagents.values():
+            trigger = microagent.get_trigger(message_content)
+            if trigger:
+                micro_text = f'<extra_info>\nThe following information has been included based on a keyword match for "{trigger}". It may or may not be relevant to the user\'s request.'
+                micro_text += '\n\n' + microagent.content
+                micro_text += '\n</extra_info>'
+                message.content.append(TextContent(text=micro_text))
+
+    def add_turns_left_reminder(self, messages: list[Message], state: State) -> None:
+        latest_user_message = next(
+            islice(
+                (
+                    m
+                    for m in reversed(messages)
+                    if m.role == 'user'
+                    and any(isinstance(c, TextContent) for c in m.content)
+                ),
+                1,
+            ),
+            None,
         )
-        return rendered.strip()
+        if latest_user_message:
+            reminder_text = f'\n\nENVIRONMENT REMINDER: You have {state.max_iterations - state.iteration} turns left to complete the task. When finished reply with <finish></finish>.'
+            latest_user_message.content.append(TextContent(text=reminder_text))

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "openhands-ai"
-version = "0.12.3"
+version = "0.13.0"
 description = "OpenHands: Code Less, Make More"
 authors = ["OpenHands"]
 license = "MIT"
@@ -61,6 +61,8 @@ protobuf = "^4.21.6,<5.0.0" # chromadb currently fails on 5.0+
 opentelemetry-api = "1.25.0"
 opentelemetry-exporter-otlp-proto-grpc = "1.25.0"
 modal = "^0.64.145"
+runloop-api-client = "0.7.0"
+pygithub = "^2.5.0"
 
 [tool.poetry.group.llama-index.dependencies]
 llama-index = "*"

tests/runtime/conftest.py

@@ -14,6 +14,7 @@ from openhands.events import EventStream
 from openhands.runtime.base import Runtime
 from openhands.runtime.impl.eventstream.eventstream_runtime import EventStreamRuntime
 from openhands.runtime.impl.remote.remote_runtime import RemoteRuntime
+from openhands.runtime.impl.runloop.runloop_runtime import RunloopRuntime
 from openhands.runtime.plugins import AgentSkillsRequirement, JupyterRequirement
 from openhands.storage import get_file_store
 from openhands.utils.async_utils import call_async_from_sync
@@ -131,6 +132,8 @@ def get_runtime_classes():
         return [EventStreamRuntime]
     elif runtime.lower() == 'remote':
         return [RemoteRuntime]
+    elif runtime.lower() == 'runloop':
+        return [RunloopRuntime]
     else:
         raise ValueError(f'Invalid runtime: {runtime}')