fix(tests,mcp): fix CI failures and add logFile parameter to devbox_run

Fix test-server.sh and test-tools.sh to use test_passed/test_failed
(renamed in the framework refactor). Fix integration tests to use
example project fixtures instead of deleted tests/fixtures/ directory,
switch from /tmp/ to make_temp_dir(), and update device name assertions.
Add optional logFile parameter to MCP devbox_run tool for redirecting
large output to a file instead of returning inline.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: abueide

plugins/devbox-mcp/src/index.js

@@ -6,6 +6,7 @@ import {
   ListToolsRequestSchema,
 } from "@modelcontextprotocol/sdk/types.js";
 import { execFile } from "child_process";
+import { writeFile } from "fs/promises";
 import { promisify } from "util";
 
 const execFileAsync = promisify(execFile);
@@ -217,7 +218,11 @@ server.setRequestHandler(ListToolsRequestSchema, async () => {
           "- devbox.d/: Per-project configuration directory\n" +
           "- .devbox/virtenv/: Temporary runtime directory (auto-regenerated, never edit directly)\n\n" +
           "The .devbox/virtenv/ directory is automatically regenerated on 'devbox shell' or 'devbox run'. " +
-          "Any manual changes to files in .devbox/virtenv/ will be lost.",
+          "Any manual changes to files in .devbox/virtenv/ will be lost.\n\n" +
+          "OUTPUT MANAGEMENT: For commands that produce large output (builds, test suites, logs), " +
+          "use the 'logFile' parameter to write output to a file instead of returning it inline. " +
+          "This keeps context tokens low. The response will include the file path, exit status, " +
+          "and a short summary. You can then read the log file selectively if needed.",
         inputSchema: {
           type: "object",
           properties: {
@@ -250,6 +255,13 @@ server.setRequestHandler(ListToolsRequestSchema, async () => {
               description: "Timeout in milliseconds (default: 120000)",
               default: 120000,
             },
+            logFile: {
+              type: "string",
+              description:
+                "Absolute path to write stdout+stderr to instead of returning inline. " +
+                "Use for commands with large output (builds, test suites) to avoid filling context. " +
+                "When set, the response returns a short summary with the log file path.",
+            },
           },
           required: ["command"],
         },
@@ -424,7 +436,7 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
 
   switch (name) {
     case "devbox_run": {
-      const { command, args: cmdArgs = [], pure = false, env = {}, cwd, timeout } = args;
+      const { command, args: cmdArgs = [], pure = false, env = {}, cwd, timeout, logFile } = args;
 
       const devboxArgs = ["run"];
       if (pure) devboxArgs.push("--pure");
@@ -441,6 +453,44 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
 
       const result = await runDevbox(devboxArgs, { cwd, timeout });
 
+      // If logFile is specified, write output to file and return summary
+      if (logFile) {
+        const fullOutput = [
+          result.stdout || "",
+          result.stderr ? `\n--- stderr ---\n${result.stderr}` : "",
+        ].join("");
+
+        try {
+          await writeFile(logFile, fullOutput, "utf-8");
+        } catch (writeErr) {
+          return {
+            content: [
+              {
+                type: "text",
+                text: `✗ Failed to write log file: ${writeErr.message}\n\nCommand ${result.success ? "succeeded" : `failed (exit ${result.exitCode})`}`,
+              },
+            ],
+            isError: true,
+          };
+        }
+
+        const lines = fullOutput.split("\n");
+        const lineCount = lines.length;
+        const tail = lines.slice(-5).join("\n");
+
+        return {
+          content: [
+            {
+              type: "text",
+              text: result.success
+                ? `✓ Command succeeded (${lineCount} lines written to ${logFile})\n\nLast 5 lines:\n${tail}`
+                : `✗ Command failed (exit ${result.exitCode}, ${lineCount} lines written to ${logFile})\n\nLast 5 lines:\n${tail}`,
+            },
+          ],
+          isError: !result.success,
+        };
+      }
+
       return {
         content: [
           {

plugins/devbox-mcp/tests/test-server.sh

@@ -32,10 +32,10 @@ echo "Validating Node.js syntax..."
 # Change to repo root to use devbox run
 cd "${SCRIPT_DIR}/../.." || exit 1
 if devbox run node --check plugins/devbox-mcp/src/index.js >/dev/null 2>&1; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ Server JavaScript syntax is valid"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ Server JavaScript syntax has errors"
 fi
 cd "${SCRIPT_DIR}" || exit 1
@@ -45,10 +45,10 @@ echo "Checking required tools are defined..."
 required_tools="devbox_run devbox_list devbox_add devbox_info devbox_search devbox_docs_search devbox_docs_list devbox_docs_read devbox_init devbox_shell_env devbox_sync"
 for tool in $required_tools; do
   if grep -q "name: \"$tool\"" "${MCP_DIR}/src/index.js"; then
-    TEST_PASS=$((TEST_PASS + 1))
+    test_passed=$((test_passed + 1))
     echo "✓ Tool defined: $tool"
   else
-    TEST_FAIL=$((TEST_FAIL + 1))
+    test_failed=$((test_failed + 1))
     echo "✗ Tool missing: $tool"
   fi
 done
@@ -60,4 +60,4 @@ assert_file_contains "${MCP_DIR}/src/index.js" "StdioServerTransport" "Server im
 # Test 8: Check server has proper error handling
 assert_file_contains "${MCP_DIR}/src/index.js" "catch.*error" "Server has error handling"
 
-test_summary
+test_summary "devbox-mcp-server"

plugins/devbox-mcp/tests/test-tools.sh

@@ -21,10 +21,10 @@ fi
 # Test 1: Server starts without errors
 echo "Testing server startup (5 second timeout)..."
 if timeout 5 node "${MCP_DIR}/src/index.js" </dev/null 2>&1 | grep -q "Devbox MCP server running"; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ Server starts successfully"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ Server failed to start"
 fi
 
@@ -34,117 +34,117 @@ server_content="$(cat "${MCP_DIR}/src/index.js")"
 
 # devbox_run tool checks
 if echo "$server_content" | grep -q 'name: "devbox_run"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_run tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_run tool not found"
 fi
 
 if echo "$server_content" | grep -q 'command:'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_run has command parameter"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_run missing command parameter"
 fi
 
 # devbox_list tool checks
 if echo "$server_content" | grep -q 'name: "devbox_list"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_list tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_list tool not found"
 fi
 
 # devbox_add tool checks
 if echo "$server_content" | grep -q 'name: "devbox_add"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_add tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_add tool not found"
 fi
 
 if echo "$server_content" | grep -q 'packages:'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_add has packages parameter"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_add missing packages parameter"
 fi
 
 # devbox_info tool checks
 if echo "$server_content" | grep -q 'name: "devbox_info"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_info tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_info tool not found"
 fi
 
 # devbox_search tool checks
 if echo "$server_content" | grep -q 'name: "devbox_search"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_search tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_search tool not found"
 fi
 
 # devbox_docs_search tool checks
 if echo "$server_content" | grep -q 'name: "devbox_docs_search"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_docs_search tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_docs_search tool not found"
 fi
 
 if echo "$server_content" | grep -q 'maxResults'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_docs_search has maxResults parameter"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_docs_search missing maxResults parameter"
 fi
 
 # devbox_docs_list tool checks
 if echo "$server_content" | grep -q 'name: "devbox_docs_list"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_docs_list tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_docs_list tool not found"
 fi
 
 # devbox_docs_read tool checks
 if echo "$server_content" | grep -q 'name: "devbox_docs_read"'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_docs_read tool defined"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_docs_read tool not found"
 fi
 
 if echo "$server_content" | grep -q 'filePath'; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ devbox_docs_read has filePath parameter"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ devbox_docs_read missing filePath parameter"
 fi
 
 # Test 3: Check helper functions exist
 echo "Checking helper functions..."
-helper_functions="runDevbox ensureDocsRepo searchDocs listDocs readDoc"
+helper_functions="runDevbox fetchDocsList fetchRawContent searchDocs listDocs readDoc"
 for func in $helper_functions; do
   if echo "$server_content" | grep -q "function $func"; then
-    TEST_PASS=$((TEST_PASS + 1))
+    test_passed=$((test_passed + 1))
     echo "✓ $func helper function defined"
   else
-    TEST_FAIL=$((TEST_FAIL + 1))
+    test_failed=$((test_failed + 1))
     echo "✗ $func helper function not found"
   fi
 done
@@ -154,12 +154,12 @@ echo "Checking tool handlers..."
 tools="devbox_run devbox_list devbox_add devbox_info devbox_search devbox_docs_search devbox_docs_list devbox_docs_read devbox_init devbox_shell_env devbox_sync"
 for tool in $tools; do
   if echo "$server_content" | grep -q "case \"$tool\":"; then
-    TEST_PASS=$((TEST_PASS + 1))
+    test_passed=$((test_passed + 1))
     echo "✓ Handler exists for $tool"
   else
-    TEST_FAIL=$((TEST_FAIL + 1))
+    test_failed=$((test_failed + 1))
     echo "✗ Handler missing for $tool"
   fi
 done
 
-test_summary
+test_summary "devbox-mcp-tools"

tests/integration/android/test-device-mgmt.sh

@@ -20,12 +20,12 @@ REPO_ROOT="$SCRIPT_DIR/../../.."
 . "$REPO_ROOT/plugins/tests/test-framework.sh"
 
 # Setup test environment
-TEST_ROOT="/tmp/android-integration-test-$$"
+TEST_ROOT="$(make_temp_dir "android-integration")"
 mkdir -p "$TEST_ROOT/devbox.d/android/devices"
 mkdir -p "$TEST_ROOT/devbox.d/android/scripts"
 
-# Copy fixtures
-cp "$SCRIPT_DIR/../../fixtures/android/devices/"*.json "$TEST_ROOT/devbox.d/android/devices/"
+# Copy device fixtures from example project
+cp "$REPO_ROOT/examples/android/devbox.d/android/devices/"*.json "$TEST_ROOT/devbox.d/android/devices/"
 
 # Copy plugin scripts
 cp -r "$REPO_ROOT/plugins/android/virtenv/scripts/"* "$TEST_ROOT/devbox.d/android/scripts/"
@@ -43,10 +43,10 @@ cd "$TEST_ROOT"
 # Test 1: Device list command
 echo "Test: Device listing..."
 if sh "$ANDROID_SCRIPTS_DIR/user/devices.sh" list >/dev/null 2>&1; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ Device list command succeeds"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ Device list command failed"
 fi
 
@@ -56,22 +56,22 @@ if sh "$ANDROID_SCRIPTS_DIR/user/devices.sh" eval >/dev/null 2>&1; then
   assert_file_exists "$ANDROID_DEVICES_DIR/devices.lock" "Lock file created after eval"
 else
   echo "✗ Device eval command failed"
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
 fi
 
 # Test 3: Lock file structure
 echo "Test: Lock file structure..."
 if [ -f "$ANDROID_DEVICES_DIR/devices.lock" ]; then
   # Lock file should be valid JSON with devices array
   if jq -e '.devices' "$ANDROID_DEVICES_DIR/devices.lock" >/dev/null 2>&1; then
-    TEST_PASS=$((TEST_PASS + 1))
+    test_passed=$((test_passed + 1))
     echo "✓ Lock file has valid structure"
   else
-    TEST_FAIL=$((TEST_FAIL + 1))
+    test_failed=$((test_failed + 1))
     echo "✗ Lock file has invalid format"
   fi
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ Lock file not found"
 fi
 
@@ -80,10 +80,10 @@ echo "Test: Device count validation..."
 device_count=$(jq '.devices | length' "$ANDROID_DEVICES_DIR/devices.lock")
 expected_count=$(ls -1 "$ANDROID_DEVICES_DIR"/*.json | wc -l | tr -d ' ')
 if [ "$device_count" = "$expected_count" ]; then
-  TEST_PASS=$((TEST_PASS + 1))
+  test_passed=$((test_passed + 1))
   echo "✓ All devices included in lock file ($device_count devices)"
 else
-  TEST_FAIL=$((TEST_FAIL + 1))
+  test_failed=$((test_failed + 1))
   echo "✗ Device count mismatch (expected $expected_count, got $device_count)"
 fi