🤖 refactor: consolidate bash execution + await/output tools

.github/workflows/pr.yml

@@ -71,9 +71,17 @@ jobs:
           key: ${{ runner.os }}-shfmt-latest
       - name: Install shfmt
         run: |
+          set -euo pipefail
           if [[ ! -f "$HOME/.local/bin/shfmt" ]] || ! "$HOME/.local/bin/shfmt" --version >/dev/null 2>&1; then
-            curl -sS https://webinstall.dev/shfmt | bash
+            # webinstall.dev can be flaky; retry and force HTTP/1.1 to avoid HTTP/2 stream errors.
+            curl --retry 5 --retry-all-errors --retry-delay 2 -LsSf --http1.1 https://webinstall.dev/shfmt | bash
+          fi
+
+          if ! "$HOME/.local/bin/shfmt" --version >/dev/null 2>&1; then
+            echo "Error: shfmt install failed"
+            exit 1
           fi
+
           echo "$HOME/.local/bin" >> $GITHUB_PATH
       - uses: cachix/install-nix-action@v27
         with:
@@ -161,7 +169,16 @@ jobs:
       - run: make storybook-build
       - run: |
           bun x http-server storybook-static -p 6006 &
-          sleep 5
+          for i in {1..30}; do
+            if curl -sf http://127.0.0.1:6006 >/dev/null; then
+              echo "Storybook ready"
+              break
+            fi
+            echo "Waiting for Storybook... ($i/30)"
+            sleep 0.5
+          done
+          # Fail the step if Storybook never became reachable.
+          curl -sf http://127.0.0.1:6006 >/dev/null
       - run: make test-storybook
 
   test-e2e:

mobile/src/messages/tools/toolRenderers.tsx

@@ -3,6 +3,12 @@ import React from "react";
 import { View, Text, ScrollView, StyleSheet, Pressable } from "react-native";
 import { Link } from "expo-router";
 import { parsePatch } from "diff";
+import {
+  coerceBashToolResult,
+  convertTaskBashResult,
+} from "@/common/utils/tools/taskResultConverters";
+import { isTaskBashArgs, isTaskBashArgsFromUnknown } from "@/common/utils/tools/taskToolTypeGuards";
+import { resolveBashDisplayName } from "@/common/utils/tools/bashDisplayName";
 import type { DisplayedMessage } from "@/common/types/message";
 import {
   FILE_EDIT_TOOL_NAMES,
@@ -88,6 +94,9 @@ export function renderSpecializedToolCard(message: ToolDisplayedMessage): ToolCa
       if (!isTaskToolArgs(message.args)) {
         return null;
       }
+      if (isTaskBashArgs(message.args)) {
+        return buildTaskBashViewModel(message as ToolDisplayedMessage & { args: TaskToolArgs });
+      }
       return buildTaskViewModel(message as ToolDisplayedMessage & { args: TaskToolArgs });
     case "task_await":
       if (!isTaskAwaitToolArgs(message.args)) {
@@ -401,6 +410,69 @@ function BashBackgroundTerminateContent({
   return <ThemedText>{result.message}</ThemedText>;
 }
 
+function buildTaskBashViewModel(
+  message: ToolDisplayedMessage & { args: TaskToolArgs }
+): ToolCardViewModel {
+  const args = message.args;
+  if (!isTaskBashArgs(args)) {
+    return {
+      icon: "💻",
+      caption: "bash",
+      title: "bash",
+      content: <ThemedText variant="muted">Invalid bash task args</ThemedText>,
+      defaultExpanded: true,
+    };
+  }
+
+  const displayName = resolveBashDisplayName(args.script, args.display_name);
+
+  const bashArgs: BashToolArgs = {
+    script: args.script,
+    timeout_secs: args.timeout_secs,
+    run_in_background: Boolean(args.run_in_background),
+    display_name: displayName,
+  };
+
+  const taskResult = coerceTaskToolResult(message.result);
+  const bashResult = convertTaskBashResult(taskResult, { legacySuccessCheckInclErrorLine: true });
+
+  const preview = truncate(args.script.trim().split("\n")[0], 80) || "bash";
+
+  const metadata: MetadataItem[] = [];
+  metadata.push({ label: "name", value: displayName });
+
+  if (typeof args.timeout_secs === "number") {
+    metadata.push({ label: "timeout", value: `${args.timeout_secs}s` });
+  }
+  if (bashResult && bashResult.exitCode !== undefined) {
+    metadata.push({ label: "exit code", value: String(bashResult.exitCode) });
+  }
+  if (bashResult && "truncated" in bashResult && bashResult.truncated) {
+    metadata.push({
+      label: "truncated",
+      value: bashResult.truncated.reason,
+      tone: "warning",
+    });
+  }
+
+  return {
+    icon: "💻",
+    caption: "bash",
+    title: preview,
+    summary: metadata.length > 0 ? <MetadataList items={metadata} /> : undefined,
+    content: (
+      <BashToolContent
+        args={bashArgs}
+        result={bashResult}
+        status={message.status}
+        toolCallId={message.toolCallId}
+      />
+    ),
+    defaultExpanded:
+      message.status !== "completed" || Boolean(bashResult && bashResult.success === false),
+  };
+}
+
 function buildTaskViewModel(
   message: ToolDisplayedMessage & { args: TaskToolArgs }
 ): ToolCardViewModel {
@@ -1503,8 +1575,15 @@ function isBashBackgroundTerminateArgs(value: unknown): value is BashBackgroundT
 }
 
 function isTaskToolArgs(value: unknown): value is TaskToolArgs {
+  if (!value || typeof value !== "object") {
+    return false;
+  }
+
+  if (isTaskBashArgsFromUnknown(value)) {
+    return true;
+  }
+
   return (
-    Boolean(value && typeof value === "object") &&
     typeof (value as TaskToolArgs).prompt === "string" &&
     typeof (value as TaskToolArgs).title === "string" &&
     typeof (value as TaskToolArgs).subagent_type === "string"
@@ -1544,18 +1623,6 @@ function isFileEditArgsUnion(value: unknown): value is FileEditArgsUnion {
   return Boolean(value && typeof (value as FileEditArgsUnion).file_path === "string");
 }
 
-function coerceBashToolResult(value: unknown): BashToolResult | null {
-  if (
-    value &&
-    typeof value === "object" &&
-    "success" in value &&
-    typeof (value as BashToolResult).success === "boolean"
-  ) {
-    return value as BashToolResult;
-  }
-  return null;
-}
-
 function coerceWebFetchToolResult(value: unknown): WebFetchToolResult | null {
   if (value && typeof value === "object" && "success" in value) {
     return value as WebFetchToolResult;

scripts/check_codex_comments.sh

@@ -49,11 +49,29 @@ REPO_INFO=$(gh repo view --json owner,name --jq '{owner: .owner.login, name: .na
 OWNER=$(echo "$REPO_INFO" | jq -r '.owner')
 REPO=$(echo "$REPO_INFO" | jq -r '.name')
 
-RESULT=$(gh api graphql \
-  -f query="$GRAPHQL_QUERY" \
-  -F owner="$OWNER" \
-  -F repo="$REPO" \
-  -F pr="$PR_NUMBER")
+# Depot runners sometimes hit transient network timeouts to api.github.com.
+# Retry the GraphQL request a few times before failing the required check.
+MAX_ATTEMPTS=5
+BACKOFF_SECS=2
+
+for ((attempt = 1; attempt <= MAX_ATTEMPTS; attempt++)); do
+  if RESULT=$(gh api graphql \
+    -f query="$GRAPHQL_QUERY" \
+    -F owner="$OWNER" \
+    -F repo="$REPO" \
+    -F pr="$PR_NUMBER"); then
+    break
+  fi
+
+  if [ $attempt -eq $MAX_ATTEMPTS ]; then
+    echo "❌ GraphQL query failed after ${MAX_ATTEMPTS} attempts"
+    exit 1
+  fi
+
+  echo "⚠️ GraphQL query failed (attempt ${attempt}/${MAX_ATTEMPTS}); retrying in ${BACKOFF_SECS}s..."
+  sleep "$BACKOFF_SECS"
+  BACKOFF_SECS=$((BACKOFF_SECS * 2))
+done
 
 # Filter regular comments from bot that aren't minimized, excluding:
 # - "Didn't find any major issues" (no issues found)

src/browser/components/AppLoader.auth.test.tsx

@@ -15,9 +15,16 @@ void mock.module("@/browser/contexts/API", () => ({
 }));
 
 void mock.module("@/browser/components/AuthTokenModal", () => ({
+  // Note: Module mocks leak between bun test files.
+  // Export all commonly-used symbols to avoid cross-test import errors.
   AuthTokenModal: (props: { error?: string | null }) => (
     <div data-testid="AuthTokenModalMock">{props.error ?? "no-error"}</div>
   ),
+  getStoredAuthToken: () => null,
+  // eslint-disable-next-line @typescript-eslint/no-empty-function
+  setStoredAuthToken: () => {},
+  // eslint-disable-next-line @typescript-eslint/no-empty-function
+  clearStoredAuthToken: () => {},
 }));
 
 import { AppLoader } from "./AppLoader";

src/browser/components/tools/BashToolCall.tsx

@@ -154,11 +154,12 @@ export const BashToolCall: React.FC<BashToolCallProps> = ({
   const showLiveOutput =
     !isBackground && (status === "executing" || (Boolean(liveOutput) && !resultHasOutput));
 
+  const truncatedInfo = result && "truncated" in result ? result.truncated : undefined;
+
   const handleToggle = () => {
     userToggledRef.current = true;
     toggleExpanded();
   };
-
   return (
     <ToolContainer expanded={expanded}>
       <ToolHeader onClick={handleToggle}>
@@ -260,6 +261,13 @@ export const BashToolCall: React.FC<BashToolCallProps> = ({
                 </DetailSection>
               )}
 
+              {truncatedInfo && (
+                <div className="text-muted px-2 text-[10px] italic">
+                  Output truncated — reason: {truncatedInfo.reason} • totalLines:{" "}
+                  {truncatedInfo.totalLines}
+                </div>
+              )}
+
               {"backgroundProcessId" in result ? (
                 // Background process: show process ID inline with icon (compact, no section wrapper)
                 <div className="flex items-center gap-2 text-[11px]">

src/browser/components/tools/TaskToolCall.tsx

@@ -179,28 +179,29 @@ export const TaskToolCall: React.FC<TaskToolCallProps> = ({ args, result, status
   const hasReport = result?.status === "completed" && !!result.reportMarkdown;
   const { expanded, toggleExpanded } = useToolExpansion(hasReport);
 
-  const isBackground = args.run_in_background ?? false;
+  const isBackground = args.run_in_background;
+
+  const title = args.title ?? "Task";
+  const prompt = args.prompt ?? "";
   const agentType = args.agentId ?? args.subagent_type ?? "unknown";
-  const prompt = args.prompt;
-  const title = args.title;
+  const kindBadge = <AgentTypeBadge type={agentType} />;
 
   // Derive task state from result
   const taskId = result?.taskId;
   const taskStatus = result?.status;
   const reportMarkdown = result?.status === "completed" ? result.reportMarkdown : undefined;
   const reportTitle = result?.status === "completed" ? result.title : undefined;
 
-  // Show preview of prompt (first line or truncated)
-  const promptPreview =
-    prompt.length > 60 ? prompt.slice(0, 60).trim() + "…" : prompt.split("\n")[0];
+  // Show preview (first line or truncated)
+  const preview = prompt.length > 60 ? prompt.slice(0, 60).trim() + "…" : prompt.split("\n")[0];
 
   return (
     <ToolContainer expanded={expanded}>
       <ToolHeader onClick={toggleExpanded}>
         <ExpandIcon expanded={expanded}>▶</ExpandIcon>
         <TaskIcon toolName="task" />
         <ToolName>task</ToolName>
-        <AgentTypeBadge type={agentType} />
+        {kindBadge}
         {isBackground && (
           <span className="text-backgrounded text-[10px] font-medium">background</span>
         )}
@@ -211,18 +212,18 @@ export const TaskToolCall: React.FC<TaskToolCallProps> = ({ args, result, status
         <ToolDetails>
           {/* Task info surface */}
           <div className="task-surface mt-1 rounded-md p-3">
-            <div className="task-divider mb-2 flex items-center gap-2 border-b pb-2">
+            <div className="task-divider mb-2 flex flex-wrap items-center gap-2 border-b pb-2">
               <span className="text-task-mode text-[12px] font-semibold">
                 {reportTitle ?? title}
               </span>
               {taskId && <TaskId id={taskId} />}
               {taskStatus && <TaskStatusBadge status={taskStatus} />}
             </div>
 
-            {/* Prompt section */}
+            {/* Prompt / script */}
             <div className="mb-2">
               <div className="text-muted mb-1 text-[10px] tracking-wide uppercase">Prompt</div>
-              <div className="text-foreground bg-code-bg max-h-[100px] overflow-y-auto rounded-sm p-2 text-[11px] break-words whitespace-pre-wrap">
+              <div className="text-foreground bg-code-bg max-h-[140px] overflow-y-auto rounded-sm p-2 text-[11px] break-words whitespace-pre-wrap">
                 {prompt}
               </div>
             </div>
@@ -249,7 +250,7 @@ export const TaskToolCall: React.FC<TaskToolCallProps> = ({ args, result, status
       )}
 
       {/* Collapsed preview */}
-      {!expanded && <div className="text-muted mt-1 truncate text-[10px]">{promptPreview}</div>}
+      {!expanded && <div className="text-muted mt-1 truncate text-[10px]">{preview}</div>}
     </ToolContainer>
   );
 };
@@ -276,6 +277,12 @@ export const TaskAwaitToolCall: React.FC<TaskAwaitToolCallProps> = ({
   const timeoutSecs = args.timeout_secs;
   const results = result?.results ?? [];
 
+  const showConfigInfo =
+    taskIds !== undefined ||
+    timeoutSecs !== undefined ||
+    args.filter !== undefined ||
+    args.filter_exclude === true;
+
   // Summary for header
   const completedCount = results.filter((r) => r.status === "completed").length;
   const totalCount = results.length;
@@ -298,10 +305,12 @@ export const TaskAwaitToolCall: React.FC<TaskAwaitToolCallProps> = ({
         <ToolDetails>
           <div className="task-surface mt-1 rounded-md p-3">
             {/* Config info */}
-            {(taskIds ?? timeoutSecs) && (
+            {showConfigInfo && (
               <div className="task-divider text-muted mb-2 flex flex-wrap gap-2 border-b pb-2 text-[10px]">
-                {taskIds && <span>Waiting for: {taskIds.length} task(s)</span>}
-                {timeoutSecs && <span>Timeout: {timeoutSecs}s</span>}
+                {taskIds !== undefined && <span>Waiting for: {taskIds.length} task(s)</span>}
+                {timeoutSecs !== undefined && <span>Timeout: {timeoutSecs}s</span>}
+                {args.filter !== undefined && <span>Filter: {args.filter}</span>}
+                {args.filter_exclude === true && <span>Exclude: true</span>}
               </div>
             )}
 
@@ -335,20 +344,35 @@ const TaskAwaitResult: React.FC<{
   const reportMarkdown = isCompleted ? result.reportMarkdown : undefined;
   const title = isCompleted ? result.title : undefined;
 
+  const output = "output" in result ? result.output : undefined;
+  const note = "note" in result ? result.note : undefined;
+  const exitCode = "exitCode" in result ? result.exitCode : undefined;
+  const elapsedMs = "elapsed_ms" in result ? result.elapsed_ms : undefined;
+
   return (
     <div className="bg-code-bg rounded-sm p-2">
-      <div className="mb-1 flex items-center gap-2">
+      <div className="mb-1 flex flex-wrap items-center gap-2">
         <TaskId id={result.taskId} />
         <TaskStatusBadge status={result.status} />
         {title && <span className="text-foreground text-[11px] font-medium">{title}</span>}
+        {exitCode !== undefined && <span className="text-muted text-[10px]">exit {exitCode}</span>}
+        {elapsedMs !== undefined && <span className="text-muted text-[10px]">{elapsedMs}ms</span>}
       </div>
 
+      {!isCompleted && output && output.length > 0 && (
+        <div className="text-foreground bg-code-bg max-h-[140px] overflow-y-auto rounded-sm p-2 text-[11px] break-words whitespace-pre-wrap">
+          {output}
+        </div>
+      )}
+
       {reportMarkdown && (
         <div className="mt-2 text-[11px]">
           <MarkdownRenderer content={reportMarkdown} />
         </div>
       )}
 
+      {note && <div className="text-muted mt-1 text-[10px]">{note}</div>}
+
       {"error" in result && result.error && (
         <div className="text-danger mt-1 text-[11px]">{result.error}</div>
       )}

src/browser/contexts/API.test.tsx

@@ -66,9 +66,13 @@ void mock.module("@orpc/client/message-port", () => ({
 }));
 
 void mock.module("@/browser/components/AuthTokenModal", () => ({
+  // Note: Module mocks leak between bun test files.
+  // Export all commonly-used symbols to avoid cross-test import errors.
   AuthTokenModal: () => null,
   getStoredAuthToken: () => null,
   // eslint-disable-next-line @typescript-eslint/no-empty-function
+  setStoredAuthToken: () => {},
+  // eslint-disable-next-line @typescript-eslint/no-empty-function
   clearStoredAuthToken: () => {},
 }));