Merge branch 'main' into rb/fix-client-loader

.github/workflows/openhands-resolver.yml

@@ -11,5 +11,5 @@ jobs:
     uses: All-Hands-AI/openhands-resolver/.github/workflows/openhands-resolver.yml@main
     if: github.event.label.name == 'fix-me'
     with:
-      max_iterations: 50
+      issue_number: ${{ github.event.issue.number || github.event.pull_request.number }}
     secrets: inherit

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:0.13-nikolaik
+2. Example: export SANDBOX_RUNTIME_CONTAINER_IMAGE=ghcr.io/all-hands-ai/runtime:0.12-nikolaik
 
 ## Develop inside Docker container
 

README.md

@@ -38,15 +38,15 @@ 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:0.13-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik
 
 docker run -it --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -p 3000:3000 \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.13
+    docker.all-hands.dev/all-hands-ai/openhands:0.12
 ```
 
 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:0.13-nikolaik}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.12-nikolaik}
       - SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234}
       - WORKSPACE_MOUNT_PATH=${WORKSPACE_BASE:-$PWD/workspace}
     ports:

config.template.toml

@@ -32,8 +32,7 @@ workspace_base = "./workspace"
 # Enable saving and restoring the session when run from CLI
 #enable_cli_session = false
 
-# Path to store trajectories, can be a folder or a file
-# If it's a folder, the session id will be used as the file name
+# Path to store trajectories
 #trajectories_path="./trajectories"
 
 # File store path

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:0.13-nikolaik}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.12-nikolaik}
       - SANDBOX_USER_ID=${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:0.13-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-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:0.13 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.12 \
     python -m openhands.core.cli
 ```
 

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

@@ -44,7 +44,7 @@ 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:0.13-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -53,6 +53,6 @@ docker run -it \
     -v /var/run/docker.sock:/var/run/docker.sock \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.13 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.12 \
     python -m openhands.core.main -t "write a bash script that prints hi"
 ```

docs/modules/usage/installation.mdx

@@ -11,15 +11,15 @@
 The easiest way to run OpenHands is in Docker.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.13-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.12-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -p 3000:3000 \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.13
+    docker.all-hands.dev/all-hands-ai/openhands:0.12
 ```
 
 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).

evaluation/EDA/run_infer.py

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

evaluation/gorilla/run_infer.py

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

evaluation/swe_bench/eval_infer.py

@@ -83,7 +83,6 @@ 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

@@ -146,7 +146,6 @@ 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

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

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

@@ -1,5 +1,5 @@
 import userEvent from "@testing-library/user-event";
-import { fireEvent, render, screen } from "@testing-library/react";
+import { render, screen } from "@testing-library/react";
 import { describe, afterEach, vi, it, expect } from "vitest";
 import { ChatInput } from "#/components/chat-input";
 
@@ -158,46 +158,4 @@ describe("ChatInput", () => {
     await user.tab();
     expect(onBlurMock).toHaveBeenCalledOnce();
   });
-
-  it("should handle text paste correctly", () => {
-    const onSubmit = vi.fn();
-    const onChange = vi.fn();
-
-    render(<ChatInput onSubmit={onSubmit} onChange={onChange} />);
-
-    const input = screen.getByTestId("chat-input").querySelector("textarea");
-    expect(input).toBeTruthy();
-
-    // Fire paste event with text data
-    fireEvent.paste(input!, {
-      clipboardData: {
-        getData: (type: string) => type === 'text/plain' ? 'test paste' : '',
-        files: []
-      }
-    });
-  });
-
-  it("should handle image paste correctly", () => {
-    const onSubmit = vi.fn();
-    const onImagePaste = vi.fn();
-
-    render(<ChatInput onSubmit={onSubmit} onImagePaste={onImagePaste} />);
-
-    const input = screen.getByTestId("chat-input").querySelector("textarea");
-    expect(input).toBeTruthy();
-
-    // Create a paste event with an image file
-    const file = new File(["dummy content"], "image.png", { type: "image/png" });
-
-    // Fire paste event with image data
-    fireEvent.paste(input!, {
-      clipboardData: {
-        getData: () => '',
-        files: [file]
-      }
-    });
-
-    // Verify image paste was handled
-    expect(onImagePaste).toHaveBeenCalledWith([file]);
-  });
 });

frontend/__tests__/utils/cache.test.ts

@@ -8,6 +8,7 @@ describe("Cache", () => {
   const testTTL = 1000; // 1 second
 
   beforeEach(() => {
+    localStorage.clear();
     vi.useFakeTimers();
   });
 
@@ -15,7 +16,17 @@ describe("Cache", () => {
     vi.useRealTimers();
   });
 
-  it("gets data from memory if not expired", () => {
+  it("sets data in localStorage with expiration", () => {
+    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);
@@ -28,6 +39,7 @@ 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", () => {
@@ -35,19 +47,28 @@ describe("Cache", () => {
 
     vi.advanceTimersByTime(testTTL + 1);
     expect(cache.get(testKey)).toBeNull();
+    expect(localStorage.getItem(`app_cache_${testKey}`)).toBeNull();
   });
 
-  it("deletes data from memory", () => {
+  it("deletes data from localStorage", () => {
     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 memory", () => {
+  it("clears all data with the app prefix from localStorage", () => {
     cache.set(testKey, testData, testTTL);
     cache.set("anotherKey", { data: "More data" }, testTTL);
     cache.clearAll();
-    expect(cache.get(testKey)).toBeNull();
-    expect(cache.get("anotherKey")).toBeNull();
+
+    expect(localStorage.length).toBe(0);
+  });
+
+  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/package-lock.json

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

frontend/package.json

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

frontend/src/components/chat-input.tsx

@@ -40,18 +40,13 @@ export function ChatInput({
   const [isDraggingOver, setIsDraggingOver] = React.useState(false);
 
   const handlePaste = (event: React.ClipboardEvent<HTMLTextAreaElement>) => {
-    // Only handle paste if we have an image paste handler and there are files
+    event.preventDefault();
     if (onImagePaste && event.clipboardData.files.length > 0) {
       const files = Array.from(event.clipboardData.files).filter((file) =>
         file.type.startsWith("image/"),
       );
-      // Only prevent default if we found image files to handle
-      if (files.length > 0) {
-        event.preventDefault();
-        onImagePaste(files);
-      }
+      if (files.length > 0) onImagePaste(files);
     }
-    // For text paste, let the default behavior handle it
   };
 
   const handleDragOver = (event: React.DragEvent<HTMLTextAreaElement>) => {

frontend/src/components/modals/AccountSettingsModal.tsx

@@ -1,4 +1,4 @@
-import { useFetcher, useRouteLoaderData } from "@remix-run/react";
+import { useFetcher } from "@remix-run/react";
 import React from "react";
 import { useTranslation } from "react-i18next";
 import { BaseModalTitle } from "./confirmation-modals/BaseModal";
@@ -6,7 +6,6 @@ import ModalBody from "./ModalBody";
 import ModalButton from "../buttons/ModalButton";
 import FormFieldset from "../form/FormFieldset";
 import { CustomInput } from "../form/custom-input";
-import { clientLoader } from "#/routes/_oh";
 import { clientAction as settingsClientAction } from "#/routes/settings";
 import { clientAction as loginClientAction } from "#/routes/login";
 import { AvailableLanguages } from "#/i18n";
@@ -25,8 +24,8 @@ function AccountSettingsModal({
   gitHubError,
   analyticsConsent,
 }: AccountSettingsModalProps) {
+  const ghToken = localStorage.getItem("ghToken");
   const { t } = useTranslation();
-  const data = useRouteLoaderData<typeof clientLoader>("routes/_oh");
   const settingsFetcher = useFetcher<typeof settingsClientAction>({
     key: "settings",
   });
@@ -36,7 +35,7 @@ function AccountSettingsModal({
     event.preventDefault();
     const formData = new FormData(event.currentTarget);
     const language = formData.get("language")?.toString();
-    const ghToken = formData.get("ghToken")?.toString();
+    const newGHToken = formData.get("ghToken")?.toString();
     const analytics = formData.get("analytics")?.toString() === "on";
 
     const accountForm = new FormData();
@@ -49,7 +48,7 @@ function AccountSettingsModal({
       )?.value;
       accountForm.append("language", languageKey ?? "en");
     }
-    if (ghToken) loginForm.append("ghToken", ghToken);
+    if (newGHToken) loginForm.append("ghToken", newGHToken);
     accountForm.append("analytics", analytics.toString());
 
     settingsFetcher.submit(accountForm, {
@@ -85,14 +84,14 @@ function AccountSettingsModal({
             name="ghToken"
             label="GitHub Token"
             type="password"
-            defaultValue={data?.ghToken ?? ""}
+            defaultValue={ghToken ?? ""}
           />
           {gitHubError && (
             <p className="text-danger text-xs">
               {t(I18nKey.ACCOUNT_SETTINGS_MODAL$GITHUB_TOKEN_INVALID)}
             </p>
           )}
-          {data?.ghToken && !gitHubError && (
+          {ghToken && !gitHubError && (
             <ModalButton
               variant="text-like"
               text={t(I18nKey.ACCOUNT_SETTINGS_MODAL$DISCONNECT)}

frontend/src/components/modals/connect-to-github-modal.tsx

@@ -1,4 +1,4 @@
-import { useFetcher, useRouteLoaderData } from "@remix-run/react";
+import { useFetcher } from "@remix-run/react";
 import { useTranslation } from "react-i18next";
 import ModalBody from "./ModalBody";
 import { CustomInput } from "../form/custom-input";
@@ -7,7 +7,6 @@ import {
   BaseModalDescription,
   BaseModalTitle,
 } from "./confirmation-modals/BaseModal";
-import { clientLoader } from "#/routes/_oh";
 import { clientAction } from "#/routes/login";
 import { I18nKey } from "#/i18n/declaration";
 
@@ -16,7 +15,7 @@ interface ConnectToGitHubModalProps {
 }
 
 export function ConnectToGitHubModal({ onClose }: ConnectToGitHubModalProps) {
-  const data = useRouteLoaderData<typeof clientLoader>("routes/_oh");
+  const ghToken = localStorage.getItem("ghToken");
   const fetcher = useFetcher<typeof clientAction>({ key: "login" });
   const { t } = useTranslation();
 
@@ -51,7 +50,7 @@ export function ConnectToGitHubModal({ onClose }: ConnectToGitHubModalProps) {
           name="ghToken"
           required
           type="password"
-          defaultValue={data?.ghToken ?? ""}
+          defaultValue={ghToken ?? ""}
         />
 
         <div className="flex flex-col gap-2 w-full">

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

@@ -12,7 +12,6 @@ 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";
 
 interface ProjectMenuCardProps {
   isConnectedToGitHub: boolean;
@@ -33,7 +32,6 @@ export function ProjectMenuCard({
   const [contextMenuIsOpen, setContextMenuIsOpen] = React.useState(false);
   const [connectToGitHubModalOpen, setConnectToGitHubModalOpen] =
     React.useState(false);
-  const [working, setWorking] = React.useState(false);
 
   const toggleMenuVisibility = () => {
     setContextMenuIsOpen((prev) => !prev);
@@ -43,7 +41,10 @@ export function ProjectMenuCard({
     posthog.capture("push_to_github_button_clicked");
     const rawEvent = {
       content: `
-Please push the changes to GitHub and open a pull request.
+Let's push the code to GitHub.
+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(),
@@ -62,11 +63,7 @@ Please push the changes to GitHub and open a pull request.
   const handleDownloadWorkspace = () => {
     posthog.capture("download_workspace_button_clicked");
     try {
-      setWorking(true);
-      downloadWorkspace().then(
-        () => setWorking(false),
-        () => setWorking(false),
-      );
+      downloadWorkspace();
     } catch (error) {
       toast.error("Failed to download workspace");
     }
@@ -74,7 +71,7 @@ Please push the changes to GitHub and open a pull request.
 
   return (
     <div className="px-4 py-[10px] w-[337px] rounded-xl border border-[#525252] flex justify-between items-center relative">
-      {!working && contextMenuIsOpen && (
+      {contextMenuIsOpen && (
         <ProjectMenuCardContextMenu
           isConnectedToGitHub={isConnectedToGitHub}
           onConnectToGitHub={() => setConnectToGitHubModalOpen(true)}
@@ -101,11 +98,7 @@ Please push the changes to GitHub and open a pull request.
         onClick={toggleMenuVisibility}
         aria-label="Open project menu"
       >
-        {working ? (
-          <LoadingSpinner size="small" />
-        ) : (
-          <EllipsisH width={36} height={36} />
-        )}
+        <EllipsisH width={36} height={36} />
       </button>
       {connectToGitHubModalOpen && (
         <ModalBackdrop onClose={() => setConnectToGitHubModalOpen(false)}>

frontend/src/components/project-menu/project.menu-card-context-menu.tsx

@@ -1,8 +1,6 @@
-import { useTranslation } from "react-i18next";
 import { useClickOutsideElement } from "#/hooks/useClickOutsideElement";
 import { ContextMenu } from "../context-menu/context-menu";
 import { ContextMenuListItem } from "../context-menu/context-menu-list-item";
-import { I18nKey } from "#/i18n/declaration";
 
 interface ProjectMenuCardContextMenuProps {
   isConnectedToGitHub: boolean;
@@ -20,7 +18,7 @@ export function ProjectMenuCardContextMenu({
   onClose,
 }: ProjectMenuCardContextMenuProps) {
   const menuRef = useClickOutsideElement<HTMLUListElement>(onClose);
-  const { t } = useTranslation();
+
   return (
     <ContextMenu
       ref={menuRef}
@@ -28,16 +26,16 @@ export function ProjectMenuCardContextMenu({
     >
       {!isConnectedToGitHub && (
         <ContextMenuListItem onClick={onConnectToGitHub}>
-          {t(I18nKey.PROJECT_MENU_CARD_CONTEXT_MENU$CONNECT_TO_GITHUB_LABEL)}
+          Connect to GitHub
         </ContextMenuListItem>
       )}
       {isConnectedToGitHub && (
         <ContextMenuListItem onClick={onPushToGitHub}>
-          {t(I18nKey.PROJECT_MENU_CARD_CONTEXT_MENU$PUSH_TO_GITHUB_LABEL)}
+          Push to GitHub
         </ContextMenuListItem>
       )}
       <ContextMenuListItem onClick={onDownloadWorkspace}>
-        {t(I18nKey.PROJECT_MENU_CARD_CONTEXT_MENU$DOWNLOAD_AS_ZIP_LABEL)}
+        Download as .zip
       </ContextMenuListItem>
     </ContextMenu>
   );

frontend/src/i18n/translation.json

@@ -1531,12 +1531,10 @@
     "tr": "Özel"
   },
   "ERROR_MESSAGE$SHOW_DETAILS": {
-    "en": "Show details",
-    "es": "Mostrar detalles"
+    "en": "Show details"
   },
   "ERROR_MESSAGE$HIDE_DETAILS": {
-    "en": "Hide details",
-    "es": "Ocultar detalles"
+    "en": "Hide details"
   },
   "STATUS$STARTING_RUNTIME": {
     "en": "Starting Runtime...",
@@ -1618,7 +1616,7 @@
   },
   "ACCOUNT_SETTINGS_MODAL$CLOSE":{
     "en": "Close",
-    "es": "Cerrar"
+    "es": ""
   },
   "ACCOUNT_SETTINGS_MODAL$GITHUB_TOKEN_INVALID":{
     "en": "GitHub token is invalid. Please try again.",
@@ -1737,8 +1735,7 @@
     "es":"atrás"
   },
   "STATUS$ERROR_LLM_AUTHENTICATION": {
-    "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"
+    "en": "Error authenticating with the LLM provider. Please check your API key"
   },
   "STATUS$ERROR_RUNTIME_DISCONNECTED": {
     "en": "There was an error while connecting to the runtime. Please refresh the page."
@@ -1748,17 +1745,5 @@
   },
   "AGENT_ERROR$ACTION_TIMEOUT": {
     "en": "Action timed out."
-  },
-  "PROJECT_MENU_CARD_CONTEXT_MENU$CONNECT_TO_GITHUB_LABEL": {
-    "en": "Connect to GitHub",
-    "es": "Conectar a GitHub"
-  },
-  "PROJECT_MENU_CARD_CONTEXT_MENU$PUSH_TO_GITHUB_LABEL": {
-    "en": "Push to GitHub",
-    "es": "Subir a GitHub"
-  },
-  "PROJECT_MENU_CARD_CONTEXT_MENU$DOWNLOAD_AS_ZIP_LABEL": {
-    "en": "Download as .zip",
-    "es": "Descargar como .zip"
   }
 }

frontend/src/routes/_oh.app.tsx

@@ -49,6 +49,7 @@ import { clearJupyter } from "#/state/jupyterSlice";
 import { FilesProvider } from "#/context/files";
 import { ErrorObservation } from "#/types/core/observations";
 import { ChatInterface } from "#/components/chat-interface";
+import { cn } from "#/utils/utils";
 
 interface ServerError {
   error: boolean | string;
@@ -294,6 +295,15 @@ function App() {
     <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">
+          <div
+            className={cn(
+              "w-2 h-2 rounded-full border",
+              "absolute left-3 top-3",
+              runtimeActive
+                ? "bg-green-800 border-green-500"
+                : "bg-red-800 border-red-500",
+            )}
+          />
           <ChatInterface />
         </Container>
 

frontend/src/services/api.ts

@@ -63,16 +63,6 @@ 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/utils/cache.ts

@@ -5,17 +5,26 @@ 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 memory
-   * @param key The key to be retrieved from memory
-   * @returns The data stored in memory
+   * Retrieve the cached data from local storage
+   * @param key The key to be retrieved from local storage
+   * @returns The data stored in local storage
    */
   public get<T>(key: CacheKey): T | null {
-    const cachedEntry = this.cacheMemory[key];
+    const cachedEntry = localStorage.getItem(this.getKey(key));
     if (cachedEntry) {
       const { data, expiration } = JSON.parse(cachedEntry) as CacheEntry<T>;
       if (Date.now() < expiration) return data;
@@ -26,34 +35,34 @@ class Cache {
   }
 
   /**
-   * Store the data in memory with expiration
-   * @param key The key to be stored in memory
-   * @param data The data to be stored in memory
+   * Store the data in local storage with expiration
+   * @param key The key to be stored in local storage
+   * @param data The data to be stored in local storage
    * @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 };
-    this.cacheMemory[key] = JSON.stringify(entry);
+    localStorage.setItem(this.getKey(key), JSON.stringify(entry));
   }
 
   /**
-   * Remove the data from memory
-   * @param key The key to be removed from memory
+   * Remove the data from local storage
+   * @param key The key to be removed from local storage
    * @returns void
    */
   public delete(key: CacheKey): void {
-    delete this.cacheMemory[key];
+    localStorage.removeItem(this.getKey(key));
   }
 
   /**
-   * Clear all data
+   * Clear all data with the app prefix from local storage
    * @returns void
    */
   public clearAll(): void {
-    Object.keys(this.cacheMemory).forEach((key) => {
-      delete this.cacheMemory[key];
+    Object.keys(localStorage).forEach((key) => {
+      if (key.startsWith(this.prefix)) localStorage.removeItem(key);
     });
   }
 }

frontend/src/utils/verified-models.ts

@@ -1,6 +1,10 @@
 // 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 = ["gpt-4o", "claude-3-5-sonnet-20241022"];
+export const VERIFIED_MODELS = [
+  "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`)
@@ -19,8 +23,11 @@ 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-haiku-20240307",
   "claude-3-opus-20240229",
   "claude-3-sonnet-20240229",
+  "claude-instant-1",
+  "claude-instant-1.2",
 ];

openhands/agenthub/codeact_agent/codeact_agent.py

@@ -39,6 +39,7 @@ from openhands.runtime.plugins import (
     JupyterRequirement,
     PluginRequirement,
 )
+from openhands.utils.microagent import MicroAgent
 from openhands.utils.prompt import PromptManager
 
 
@@ -85,6 +86,16 @@ 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(
@@ -94,6 +105,7 @@ 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,
@@ -102,17 +114,18 @@ class CodeActAgent(Agent):
             logger.debug(
                 f'TOOLS loaded for CodeActAgent: {json.dumps(self.tools, indent=2)}'
             )
-            self.prompt_manager = PromptManager(
-                microagent_dir=os.path.join(os.path.dirname(__file__), 'micro'),
-                prompt_dir=os.path.join(os.path.dirname(__file__), 'prompts', 'tools'),
-            )
+            self.system_prompt = codeact_function_calling.SYSTEM_PROMPT
+            self.initial_user_message = None
         else:
+            # Non-function-calling mode
             self.action_parser = CodeActResponseParser()
             self.prompt_manager = PromptManager(
-                microagent_dir=os.path.join(os.path.dirname(__file__), 'micro'),
-                prompt_dir=os.path.join(os.path.dirname(__file__), 'prompts', 'default'),
+                prompt_dir=os.path.join(os.path.dirname(__file__)),
                 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()
 
@@ -324,8 +337,8 @@ class CodeActAgent(Agent):
             return self.pending_actions.popleft()
 
         # if we're done, go back
-        latest_user_message = state.get_last_user_message()
-        if latest_user_message and latest_user_message.content.strip() == '/exit':
+        last_user_message = state.get_last_user_message()
+        if last_user_message and last_user_message.strip() == '/exit':
             return AgentFinishAction()
 
         # prepare what we want to send to the LLM
@@ -390,19 +403,17 @@ class CodeActAgent(Agent):
                 role='system',
                 content=[
                     TextContent(
-                        text=self.prompt_manager.get_system_message(),
-                        cache_prompt=self.llm.is_caching_prompt_active(),
+                        text=self.system_prompt,
+                        cache_prompt=self.llm.is_caching_prompt_active(),  # Cache system prompt
                     )
                 ],
             )
         ]
-        example_message = self.prompt_manager.get_example_user_message()
-        if example_message:
+        if self.initial_user_message:
             messages.append(
                 Message(
                     role='user',
-                    content=[TextContent(text=example_message)],
-                    cache_prompt=self.llm.is_caching_prompt_active(),
+                    content=[TextContent(text=self.initial_user_message)],
                 )
             )
 
@@ -451,9 +462,8 @@ 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
@@ -483,6 +493,23 @@ class CodeActAgent(Agent):
                         break
 
         if not self.function_calling_active:
-            self.prompt_manager.add_turns_left_reminder(messages, state)
+            # The latest user message is important:
+            # 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

@@ -25,6 +25,13 @@ 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.
+</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

@@ -1,9 +0,0 @@
----
-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

@@ -1,31 +1,69 @@
 ---
 name: github
 agent: CodeActAgent
-triggers:
-- github
-- git
+require_env_var:
+    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."
 ---
 
-You have access to an environment variable, `GITHUB_TOKEN`, which allows you to interact with
-the GitHub API.
+# How to Interact with Github
 
-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.
+## Environment Variable Available
 
-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
-* 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 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
-* 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:
-```bash
-git checkout -b create-widget
-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"}'
+- `GITHUB_TOKEN`: A read-only token for Github.
+
+## Using GitHub's RESTful API
+
+Use `curl` with the `GITHUB_TOKEN` to interact with GitHub's API. Here are some common operations:
+
+Here's a template for API calls:
+
+```sh
+curl -H "Authorization: token $GITHUB_TOKEN" \
+    "https://api.github.com/{endpoint}"
 ```
+
+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/tools/system_prompt.j2

@@ -1,7 +0,0 @@
-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_agent/system_prompt.j2

@@ -163,9 +163,6 @@ IMPORTANT: Execute code using <execute_ipython>, <execute_bash>, or <execute_bro
 The assistant should utilize full file paths and the `pwd` command to prevent path-related errors.
 The assistant MUST NOT apologize to the user or thank the user after running commands or editing files. It should only address the user in response to an explicit message from the user, or to ask for more information.
 The assistant MUST NOT push any changes to GitHub unless explicitly requested to do so.
-The assistant MUST NOT include comments in the code unless they are necessary to describe non-obvious behavior, or
-to describe precisely how to apply proposed edits. Comments about applying edits should always have blank lines above
-and below.
 
 {% endset %}
 {# Combine all parts without newlines between them #}

openhands/agenthub/codeact_agent/user_prompt.j2

@@ -215,5 +215,12 @@ 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_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.content.strip() == '/exit':
+        if last_user_message and last_user_message.strip() == '/exit':
             return AgentFinishAction()
 
         # prepare what we want to send to the LLM

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

openhands/core/config/app_config.py

@@ -69,7 +69,6 @@ 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] = {}
 
@@ -140,7 +139,6 @@ 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

@@ -14,8 +14,7 @@ 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 default sandbox action execution.
-        remote_runtime_init_timeout: The timeout for the remote runtime to start.
+        timeout: The timeout for the sandbox.
         enable_auto_lint: Whether to enable auto-lint.
         use_host_network: Whether to use the host network.
         initialize_plugins: Whether to initialize plugins.
@@ -42,7 +41,6 @@ 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/main.py

@@ -123,7 +123,6 @@ async def run_controller(
 
     if runtime is None:
         runtime = create_runtime(config, sid=sid)
-        await runtime.connect()
 
     event_stream = runtime.event_stream
 
@@ -189,6 +188,8 @@ async def run_controller(
 
     event_stream.subscribe(EventStreamSubscriber.MAIN, on_event, sid)
 
+    await runtime.connect()
+
     end_states = [
         AgentState.FINISHED,
         AgentState.REJECTED,
@@ -212,11 +213,7 @@ async def run_controller(
 
     # save trajectories if applicable
     if config.trajectories_path is not None:
-        # if trajectories_path is a folder, use session id as file name
-        if os.path.isdir(config.trajectories_path):
-            file_path = os.path.join(config.trajectories_path, sid + '.json')
-        else:
-            file_path = config.trajectories_path
+        file_path = os.path.join(config.trajectories_path, sid + '.json')
         os.makedirs(os.path.dirname(file_path), exist_ok=True)
         histories = [event_to_trajectory(event) for event in state.history]
         with open(file_path, 'w') as f:

openhands/runtime/__init__.py

@@ -23,10 +23,6 @@ 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/base.py

@@ -3,7 +3,6 @@ import copy
 import json
 import os
 from abc import abstractmethod
-from pathlib import Path
 from typing import Callable
 
 from requests.exceptions import ConnectionError
@@ -275,6 +274,6 @@ class Runtime(FileEditRuntimeMixin):
         raise NotImplementedError('This method is not implemented in the base class.')
 
     @abstractmethod
-    def copy_from(self, path: str) -> Path:
-        """Zip all files in the sandbox and return a path in the local filesystem."""
+    def copy_from(self, path: str) -> bytes:
+        """Zip all files in the sandbox and return as a stream of bytes."""
         raise NotImplementedError('This method is not implemented in the base class.')

openhands/runtime/impl/eventstream/eventstream_runtime.py

@@ -1,5 +1,4 @@
 import os
-from pathlib import Path
 import tempfile
 import threading
 from functools import lru_cache
@@ -605,7 +604,7 @@ class EventStreamRuntime(Runtime):
         except requests.Timeout:
             raise TimeoutError('List files operation timed out')
 
-    def copy_from(self, path: str) -> Path:
+    def copy_from(self, path: str) -> bytes:
         """Zip all files in the sandbox and return as a stream of bytes."""
         self._refresh_logs()
         try:
@@ -618,11 +617,8 @@ class EventStreamRuntime(Runtime):
                 stream=True,
                 timeout=30,
             )
-            temp_file = tempfile.NamedTemporaryFile(delete=False)
-            for chunk in response.iter_content(chunk_size=8192):
-                if chunk:  # filter out keep-alive new chunks
-                    temp_file.write(chunk)
-            return Path(temp_file.name)
+            data = response.content
+            return data
         except requests.Timeout:
             raise TimeoutError('Copy operation timed out')
 

openhands/runtime/impl/remote/remote_runtime.py

@@ -1,7 +1,6 @@
 import os
 import tempfile
 import threading
-from pathlib import Path
 from typing import Callable, Optional
 from zipfile import ZipFile
 
@@ -260,19 +259,13 @@ 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',
@@ -474,18 +467,13 @@ class RemoteRuntime(Runtime):
         assert isinstance(response_json, list)
         return response_json
 
-    def copy_from(self, path: str) -> Path:
+    def copy_from(self, path: str) -> bytes:
         """Zip all files in the sandbox and return as a stream of bytes."""
         params = {'path': path}
         response = self._send_request(
             'GET',
             f'{self.runtime_url}/download_files',
             params=params,
-            stream=True,
             timeout=30,
         )
-        temp_file = tempfile.NamedTemporaryFile(delete=False)
-        for chunk in response.iter_content(chunk_size=8192):
-            if chunk:  # filter out keep-alive new chunks
-                temp_file.write(chunk)
-        return Path(temp_file.name)
+        return response.content

openhands/runtime/impl/runloop/README.md

@@ -1,31 +0,0 @@
-# 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

@@ -1,272 +0,0 @@
-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

@@ -1,12 +1,9 @@
 import os
 
-from github import Github
-from github.GithubException import GithubException
-from tenacity import retry, stop_after_attempt, wait_exponential
+import httpx
 
 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()
@@ -104,7 +101,6 @@ async def authenticate_github_user(auth_token) -> bool:
     return True
 
 
-@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=1, max=5))
 async def get_github_user(token: str) -> str:
     """Get GitHub user info from token.
 
@@ -112,16 +108,21 @@ async def get_github_user(token: str) -> str:
         token: GitHub access token
 
     Returns:
-        github handle of the user
+        Tuple of (login, error_message)
+        If successful, error_message is None
+        If failed, login is None and error_message contains the error
     """
     logger.info('Fetching GitHub user info from token')
-    try:
-        g = Github(token)
-        user = await call_sync_from_async(g.get_user)
-        login = user.login
+    headers = {
+        'Accept': 'application/vnd.github+json',
+        'Authorization': f'Bearer {token}',
+        'X-GitHub-Api-Version': '2022-11-28',
+    }
+    async with httpx.AsyncClient() as client:
+        logger.debug('Making request to GitHub API')
+        response = await client.get('https://api.github.com/user', headers=headers)
+        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

@@ -1,13 +1,12 @@
 import asyncio
+import io
 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
@@ -17,7 +16,6 @@ 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
@@ -29,7 +27,6 @@ with warnings.catch_warnings():
 
 from dotenv import load_dotenv
 from fastapi import (
-    BackgroundTasks,
     FastAPI,
     HTTPException,
     Request,
@@ -37,7 +34,7 @@ from fastapi import (
     WebSocket,
     status,
 )
-from fastapi.responses import FileResponse, JSONResponse
+from fastapi.responses import JSONResponse, StreamingResponse
 from fastapi.security import HTTPBearer
 from fastapi.staticfiles import StaticFiles
 from pydantic import BaseModel
@@ -63,7 +60,7 @@ 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.auth import get_sid_from_token, sign_token
+from openhands.server.auth import get_sid_from_token, sign_token
 from openhands.server.middleware import LocalhostCORSMiddleware, NoCacheMiddleware
 from openhands.server.session import SessionManager
 
@@ -207,22 +204,12 @@ async def attach_session(request: Request, call_next):
         response = await call_next(request)
         return response
 
-    user_verifier = UserVerifier()
-    if user_verifier.is_active():
-        signed_token = request.cookies.get('github_auth')
-        if not signed_token:
-            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'},
-            )
+    github_token = request.headers.get('X-GitHub-Token')
+    if not await authenticate_github_user(github_token):
+        return JSONResponse(
+            status_code=status.HTTP_401_UNAUTHORIZED,
+            content={'error': 'Not authenticated'},
+        )
 
     if not request.headers.get('Authorization'):
         logger.warning('Missing Authorization header')
@@ -803,20 +790,19 @@ async def security_api(request: Request):
 
 
 @app.get('/api/zip-directory')
-async def zip_current_workspace(request: Request, background_tasks: BackgroundTasks):
+async def zip_current_workspace(request: Request):
     try:
         logger.debug('Zipping workspace')
         runtime: Runtime = request.state.conversation.runtime
-        path = runtime.config.workspace_mount_path_in_sandbox
-        zip_file = await call_sync_from_async(runtime.copy_from, path)
-        response = FileResponse(
-            path=zip_file,
-            filename='workspace.zip',
-            media_type='application/x-zip-compressed',
-        )
 
-        # This will execute after the response is sent (So the file is not deleted before being sent)
-        background_tasks.add_task(zip_file.unlink)
+        path = runtime.config.workspace_mount_path_in_sandbox
+        zip_file_bytes = await call_sync_from_async(runtime.copy_from, path)
+        zip_stream = io.BytesIO(zip_file_bytes)  # Wrap to behave like a file stream
+        response = StreamingResponse(
+            zip_stream,
+            media_type='application/x-zip-compressed',
+            headers={'Content-Disposition': 'attachment; filename=workspace.zip'},
+        )
 
         return response
     except Exception as e:
@@ -877,26 +863,10 @@ 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

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

openhands/utils/prompt.py

@@ -1,10 +1,7 @@
 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
 
 
@@ -19,31 +16,21 @@ 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,
-        microagent_dir: str = '',
-        agent_skills_docs: str = '',
+        agent_skills_docs: str,
+        micro_agent: MicroAgent | None = None,
     ):
         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.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
+        self.micro_agent: MicroAgent | None = micro_agent
 
     def _load_template(self, template_name: str) -> Template:
         template_path = os.path.join(self.prompt_dir, f'{template_name}.j2')
@@ -52,13 +39,15 @@ class PromptManager:
         with open(template_path, 'r') as file:
             return Template(file.read())
 
-    def get_system_message(self) -> str:
+    @property
+    def system_message(self) -> str:
         rendered = self.system_template.render(
             agent_skills_docs=self.agent_skills_docs,
         ).strip()
         return rendered
 
-    def get_example_user_message(self) -> str:
+    @property
+    def initial_user_message(self) -> str:
         """This is the initial user message provided to the agent
         before *actual* user instructions are provided.
 
@@ -68,39 +57,7 @@ class PromptManager:
         These additional context will convert the current generic agent
         into a more specialized agent that is tailored to the user's task.
         """
-        return self.user_template.render().strip()
-
-    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,
+        rendered = self.user_template.render(
+            micro_agent=self.micro_agent.content if self.micro_agent else None
         )
-        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 rendered.strip()

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "openhands-ai"
-version = "0.13.0"
+version = "0.12.3"
 description = "OpenHands: Code Less, Make More"
 authors = ["OpenHands"]
 license = "MIT"
@@ -13,7 +13,6 @@ packages = [
 [tool.poetry.dependencies]
 python = "^3.12"
 datasets = "*"
-PyGithub = "*"
 pandas = "*"
 litellm = "^1.51.1"
 google-generativeai = "*" # To use litellm with Gemini Pro API
@@ -62,7 +61,6 @@ 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"
 
 [tool.poetry.group.llama-index.dependencies]
 llama-index = "*"

tests/runtime/conftest.py

@@ -14,7 +14,6 @@ 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
@@ -132,8 +131,6 @@ 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}')
 

tests/runtime/test_bash.py

@@ -1,7 +1,6 @@
 """Bash-related tests for the EventStreamRuntime, which connects to the ActionExecutor running in the sandbox."""
 
 import os
-from pathlib import Path
 
 import pytest
 from conftest import (
@@ -587,10 +586,8 @@ def test_copy_from_directory(temp_dir, runtime_cls):
         path_to_copy_from = f'{sandbox_dir}/test_dir'
         result = runtime.copy_from(path=path_to_copy_from)
 
-        # Result is returned as a path
-        assert isinstance(result, Path)
-
-        result.unlink()
+        # Result is returned in bytes
+        assert isinstance(result, bytes)
     finally:
         _close_test_runtime(runtime)
 

tests/unit/test_config.py

@@ -461,7 +461,6 @@ def test_api_keys_repr_str():
         jwt_secret='my_jwt_secret',
         modal_api_token_id='my_modal_api_token_id',
         modal_api_token_secret='my_modal_api_token_secret',
-        runloop_api_key='my_runloop_api_key',
     )
     assert "e2b_api_key='******'" in repr(app_config)
     assert "e2b_api_key='******'" in str(app_config)
@@ -471,8 +470,6 @@ def test_api_keys_repr_str():
     assert "modal_api_token_id='******'" in str(app_config)
     assert "modal_api_token_secret='******'" in repr(app_config)
     assert "modal_api_token_secret='******'" in str(app_config)
-    assert "runloop_api_key='******'" in repr(app_config)
-    assert "runloop_api_key='******'" in str(app_config)
 
     # Check that no other attrs in AppConfig have 'key' or 'token' in their name
     # This will fail when new attrs are added, and attract attention
@@ -480,7 +477,6 @@ def test_api_keys_repr_str():
         'e2b_api_key',
         'modal_api_token_id',
         'modal_api_token_secret',
-        'runloop_api_key',
     ]
     for attr_name in dir(AppConfig):
         if (

tests/unit/test_microagent_utils.py

@@ -1,8 +1,13 @@
 import os
 
+import pytest
 from pytest import MonkeyPatch
 
 import openhands.agenthub  # noqa: F401
+from openhands.core.exceptions import (
+    AgentNotRegisteredError,
+    MicroAgentValidationError,
+)
 from openhands.utils.microagent import MicroAgent
 
 CONTENT = (
@@ -29,3 +34,40 @@ def test_micro_agent_load(tmp_path, monkeypatch: MonkeyPatch):
     micro_agent = MicroAgent(os.path.join(tmp_path, 'dummy.md'))
     assert micro_agent is not None
     assert micro_agent.content == CONTENT.strip()
+
+
+def test_not_existing_agent(tmp_path, monkeypatch: MonkeyPatch):
+    with open(os.path.join(tmp_path, 'dummy.md'), 'w') as f:
+        f.write(
+            (
+                '---\n'
+                'name: dummy\n'
+                'agent: NotExistingAgent\n'
+                'require_env_var:\n'
+                '  SANDBOX_OPENHANDS_TEST_ENV_VAR: "Set this environment variable for testing purposes"\n'
+                '---\n' + CONTENT
+            )
+        )
+    monkeypatch.setenv('SANDBOX_OPENHANDS_TEST_ENV_VAR', 'dummy_value')
+
+    with pytest.raises(AgentNotRegisteredError):
+        MicroAgent(os.path.join(tmp_path, 'dummy.md'))
+
+
+def test_not_existing_env_var(tmp_path):
+    with open(os.path.join(tmp_path, 'dummy.md'), 'w') as f:
+        f.write(
+            (
+                '---\n'
+                'name: dummy\n'
+                'agent: CodeActAgent\n'
+                'require_env_var:\n'
+                '  SANDBOX_OPENHANDS_TEST_ENV_VAR: "Set this environment variable for testing purposes"\n'
+                '---\n' + CONTENT
+            )
+        )
+
+    with pytest.raises(MicroAgentValidationError) as excinfo:
+        MicroAgent(os.path.join(tmp_path, 'dummy.md'))
+
+    assert 'Set this environment variable for testing purposes' in str(excinfo.value)

tests/unit/test_prompt_manager.py

@@ -1,9 +1,9 @@
 import os
 import shutil
+from unittest.mock import Mock
 
 import pytest
 
-from openhands.core.message import Message, TextContent
 from openhands.utils.microagent import MicroAgent
 from openhands.utils.prompt import PromptManager
 
@@ -11,9 +11,7 @@ from openhands.utils.prompt import PromptManager
 @pytest.fixture
 def prompt_dir(tmp_path):
     # Copy contents from "openhands/agenthub/codeact_agent" to the temp directory
-    shutil.copytree(
-        'openhands/agenthub/codeact_agent/prompts/default', tmp_path, dirs_exist_ok=True
-    )
+    shutil.copytree('openhands/agenthub/codeact_agent', tmp_path, dirs_exist_ok=True)
 
     # Return the temporary directory path
     return tmp_path
@@ -27,79 +25,78 @@ def agent_skills_docs():
     return SAMPLE_AGENT_SKILLS_DOCS
 
 
-def test_prompt_manager_without_microagent(prompt_dir, agent_skills_docs):
-    manager = PromptManager(
-        prompt_dir, microagent_dir='', agent_skills_docs=agent_skills_docs
-    )
+def test_prompt_manager_without_micro_agent(prompt_dir, agent_skills_docs):
+    manager = PromptManager(prompt_dir, agent_skills_docs)
 
     assert manager.prompt_dir == prompt_dir
     assert manager.agent_skills_docs == agent_skills_docs
-    assert len(manager.microagents) == 0
+    assert manager.micro_agent is None
 
-    assert isinstance(manager.get_system_message(), str)
+    assert isinstance(manager.system_message, str)
     assert (
         "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed answers to the user's questions."
-        in manager.get_system_message()
+        in manager.system_message
     )
-    assert SAMPLE_AGENT_SKILLS_DOCS in manager.get_system_message()
-    assert isinstance(manager.get_example_user_message(), str)
-    assert '--- BEGIN OF GUIDELINE ---' not in manager.get_example_user_message()
-    assert '--- END OF GUIDELINE ---' not in manager.get_example_user_message()
-    assert "NOW, LET'S START!" in manager.get_example_user_message()
-    assert 'microagent' not in manager.get_example_user_message()
+    assert SAMPLE_AGENT_SKILLS_DOCS in manager.system_message
+    assert isinstance(manager.initial_user_message, str)
+    assert '--- BEGIN OF GUIDELINE ---' not in manager.initial_user_message
+    assert '--- END OF GUIDELINE ---' not in manager.initial_user_message
+    assert "NOW, LET'S START!" in manager.initial_user_message
+    assert 'micro_agent' not in manager.initial_user_message
 
 
-def test_prompt_manager_with_microagent(prompt_dir, agent_skills_docs):
-    microagent_name = 'test_microagent'
-    microagent_content = """
----
-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
-"""
+def test_prompt_manager_with_micro_agent(prompt_dir, agent_skills_docs):
+    micro_agent_name = 'test_micro_agent'
+    micro_agent_content = (
+        '## Micro Agent\n'
+        'This is a test micro agent.\n'
+        'It is used to test the prompt manager.\n'
+    )
 
     # Create a temporary micro agent file
     os.makedirs(os.path.join(prompt_dir, 'micro'), exist_ok=True)
-    with open(os.path.join(prompt_dir, 'micro', f'{microagent_name}.md'), 'w') as f:
-        f.write(microagent_content)
+    with open(os.path.join(prompt_dir, 'micro', f'{micro_agent_name}.md'), 'w') as f:
+        f.write(micro_agent_content)
+
+    # Mock MicroAgent
+    mock_micro_agent = Mock(spec=MicroAgent)
+    mock_micro_agent.content = micro_agent_content
 
     manager = PromptManager(
         prompt_dir=prompt_dir,
-        microagent_dir=os.path.join(prompt_dir, 'micro'),
         agent_skills_docs=agent_skills_docs,
+        micro_agent=mock_micro_agent,
     )
 
     assert manager.prompt_dir == prompt_dir
     assert manager.agent_skills_docs == agent_skills_docs
-    assert len(manager.microagents) == 1
+    assert manager.micro_agent == mock_micro_agent
 
-    assert isinstance(manager.get_system_message(), str)
+    assert isinstance(manager.system_message, str)
     assert (
         "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed answers to the user's questions."
-        in manager.get_system_message()
+        in manager.system_message
     )
-    assert SAMPLE_AGENT_SKILLS_DOCS in manager.get_system_message()
+    assert SAMPLE_AGENT_SKILLS_DOCS in manager.system_message
 
-    assert isinstance(manager.get_example_user_message(), str)
+    assert isinstance(manager.initial_user_message, str)
+    assert (
+        '--- BEGIN OF GUIDELINE ---\n'
+        + 'The following information may assist you in completing your task:\n\n'
+        + micro_agent_content
+        + '\n'
+        + '--- END OF GUIDELINE ---\n'
+        + "\n\nNOW, LET'S START!"
+    ) in manager.initial_user_message
+    assert micro_agent_content in manager.initial_user_message
 
-    message = Message(
-        role='user',
-        content=[TextContent(text='Hello, flarglebargle!')],
-    )
-    manager.enhance_message(message)
-    assert 'magic word' in message.content[1].text
-
-    os.remove(os.path.join(prompt_dir, 'micro', f'{microagent_name}.md'))
+    # Clean up the temporary file
+    os.remove(os.path.join(prompt_dir, 'micro', f'{micro_agent_name}.md'))
 
 
 def test_prompt_manager_file_not_found(prompt_dir, agent_skills_docs):
     with pytest.raises(FileNotFoundError):
-        MicroAgent(os.path.join(prompt_dir, 'micro', 'non_existent_microagent.md'))
+        MicroAgent(os.path.join(prompt_dir, 'micro', 'non_existent_micro_agent.md'))
 
 
 def test_prompt_manager_template_rendering(prompt_dir, agent_skills_docs):
@@ -107,14 +104,12 @@ def test_prompt_manager_template_rendering(prompt_dir, agent_skills_docs):
     with open(os.path.join(prompt_dir, 'system_prompt.j2'), 'w') as f:
         f.write('System prompt: {{ agent_skills_docs }}')
     with open(os.path.join(prompt_dir, 'user_prompt.j2'), 'w') as f:
-        f.write('User prompt: foo')
+        f.write('User prompt: {{ micro_agent }}')
 
-    manager = PromptManager(
-        prompt_dir, microagent_dir='', agent_skills_docs=agent_skills_docs
-    )
+    manager = PromptManager(prompt_dir, agent_skills_docs)
 
-    assert manager.get_system_message() == f'System prompt: {agent_skills_docs}'
-    assert manager.get_example_user_message() == 'User prompt: foo'
+    assert manager.system_message == f'System prompt: {agent_skills_docs}'
+    assert manager.initial_user_message == 'User prompt: None'
 
     # Clean up temporary files
     os.remove(os.path.join(prompt_dir, 'system_prompt.j2'))