{"title":"iOS 应用自动化测试","slug":"automated-testing-ios","category":"guides","summary":"使用 Synapse + Computer Control API 通过 Simulator 自动化 iOS 应用测试。","audience":["human","llm"],"tags":["guide","ios","testing","automation"],"difficulty":"advanced","updated":"2026-06-27","word_count":189,"read_minutes":1,"lang":"zh","translated":true,"requested_lang":"zh","content_markdown":"\n# iOS 应用自动化测试\n\n将 Synapse 的记忆系统与 Computer Control API 结合，构建 LLM 驱动的 iOS 应用测试。LLM 会记住测试场景，从过往失败中学习，并适应 UI 变化。\n\n## 架构\n\n```\n┌──────────────┐ 命令 ┌──────────────┐ 截图 ┌──────────────┐\n│ LLM Agent │ ─────────────▶│ Synapse │ ────────────────▶ │ iOS Sim │\n│ (Claude) │ │ Computer │ ◀──────────────── │ (via agent) │\n└──────────────┘ │ Control │ 结果 └──────────────┘\n │ └──────────────┘\n │ store/recall\n ▼\n┌──────────────┐\n│ 记忆 │ (测试场景、过往失败、UI 模式)\n└──────────────┘\n```\n\n## 前置条件\n\n- Synapse 账户 + Mind Key\n- 在 Claude Desktop 中配置 Synapse MCP Server\n- 已安装 `screen-remote-agent` 的 iOS Simulator\n- 在 Synapse 中注册计算机（参见 [Computer Control API](/docs/api/computers)）\n\n## 第 1 步：注册 Simulator 计算机\n\n在运行 iOS Simulator 的 Mac 上：\n\n```bash\n# 从 Synapse 获取安装码\ncurl -X POST https://synapse.schaefer.zone/computers/install-code \\\n -H \"Authorization: Bearer YOUR_MIND_KEY\" \\\n -d '{\"computer_name\":\"ios-sim\"}'\n# → { \"install_code\": \"ic_...\" }\n\n# 在 Mac 上运行 screen-remote-agent\n# (使用安装码完成注册)\n```\n\n## 第 2 步：在记忆中存储测试场景\n\n把可复用的测试场景作为记忆存储：\n\n```python\nimport requests\n\ndef store_test_scenario(name, steps, app):\n requests.post(f\"{URL}/memory\",\n headers={\"Authorization\": f\"Bearer {MIND_KEY}\"},\n json={\n \"category\": \"skill\",\n \"key\": f\"test_scenario_{name}\",\n \"content\": f\"App: {app}\\nSteps:\\n\" + \"\\n\".join(steps),\n \"tags\": [\"test\", \"ios\", app],\n \"priority\": \"high\"\n })\n\nstore_test_scenario(\"login_flow\", [\n \"Launch app\",\n \"Tap email field\",\n \"Type test@example.com\",\n \"Tap password field\",\n \"Type password123\",\n \"Tap Login button\",\n \"Verify home screen appears\"\n], \"MyApp\")\n```\n\n## 第 3 步：LLM 驱动的测试执行\n\n在 Claude Desktop 中（已配置 Synapse MCP）：\n\n```\nRun the login_flow test scenario on the iOS Simulator.\nTake a screenshot after each step and verify the expected UI.\nIf any step fails, store the failure as a memory so we can\navoid it next time.\n```\n\nClaude 会：\n\n1. 调用 `memory_search` 查找 `test_scenario_login_flow` 记忆\n2. 调用 `computer_screenshot` 查看当前状态\n3. 通过 `computer_command_queue` 执行每一步（点击、输入）\n4. 通过截图验证结果\n5. 把任何失败存储为 `mistake` 记忆\n\n## 第 4 步：自愈测试\n\n当测试失败时，存储失败信息与恢复方案：\n\n```python\ndef store_test_failure(scenario, step, error, recovery):\n requests.post(f\"{URL}/memory\",\n headers={\"Authorization\": f\"Bearer {MIND_KEY}\"},\n json={\n \"category\": \"mistake\",\n \"key\": f\"failure_{scenario}_{step}\",\n \"content\": f\"Scenario: {scenario}\\nStep: {step}\\nError: {error}\\nRecovery: {recovery}\",\n \"tags\": [\"test\", \"failure\", \"ios\", scenario],\n \"priority\": \"high\"\n })\n\n# 示例\nstore_test_failure(\"login_flow\", \"tap_login\",\n \"Login button not found at expected coordinates\",\n \"Button moved due to new logo. Search by accessibility label instead.\")\n```\n\n下次 LLM 运行该测试时，会回放该失败记忆并自动应用恢复方案。\n\n## 第 5 步：测试结果跟踪\n\n把测试运行记录为任务：\n\n```python\ndef track_test_run(scenario, status, duration):\n requests.post(f\"{URL}/mind/task\",\n headers={\"Authorization\": f\"Bearer {MIND_KEY}\",\n \"Content-Type\": \"application/json\"},\n json={\n \"title\": f\"Test: {scenario}\",\n \"description\": f\"Status: {status}, Duration: {duration}s\",\n \"priority\": \"normal\"\n })\n```\n\n## 常用命令\n\n| 操作 | 命令 |\n|--------|---------|\n| 启动 Simulator | `xcrun simctl launch booted com.example.app` |\n| 截屏 | `computer_screenshot`（通过 Synapse MCP） |\n| 在 (x,y) 点击 | `computer_command_queue {type:\"click\", payload:{x,y}}` |\n| 输入文本 | `computer_command_queue {type:\"type\", payload:{text:\"...\"}}` |\n| 按 Home 键 | `computer_command_queue {type:\"key\", payload:{keys:[\"Cmd\",\"Shift\",\"H\"]}}` |\n\n## 最佳实践\n\n> [!TIP]\n> - **把 UI 坐标作为记忆存储** — UI 会变，但 LLM 可以重新学习\n> - **使用 accessibility 标签** — 比坐标更稳定\n> - **把测试数据单独存储** — 用变量管理用户名、密码\n> - **在干净状态运行测试** — 每次运行之间重置 Simulator\n> - **为失败保存截图** — 便于调试\n\n## 下一步\n\n- [自愈测试](/docs/guides/self-healing-tests)\n- [Computer Control API](/docs/api/computers)\n- [记忆最佳实践](/docs/guides/memory-best-practices)\n","content_html":"

iOS 应用自动化测试

\n

将 Synapse 的记忆系统与 Computer Control API 结合，构建 LLM 驱动的 iOS 应用测试。LLM 会记住测试场景，从过往失败中学习，并适应 UI 变化。

\n

架构

\n
┌──────────────┐    命令     ┌──────────────┐    截图    ┌──────────────┐\n│  LLM Agent   │ ─────────────▶│  Synapse     │ ────────────────▶ │  iOS Sim     │\n│  (Claude)    │               │  Computer    │ ◀──────────────── │  (via agent) │\n└──────────────┘               │  Control     │    结果           └──────────────┘\n       │                       └──────────────┘\n       │ store/recall\n       ▼\n┌──────────────┐\n│  记忆        │ (测试场景、过往失败、UI 模式)\n└──────────────┘

前置条件

\n\n

第 1 步：注册 Simulator 计算机

\n

在运行 iOS Simulator 的 Mac 上：

\n
# 从 Synapse 获取安装码\ncurl -X POST https://synapse.schaefer.zone/computers/install-code \\\n  -H "Authorization: Bearer YOUR_MIND_KEY" \\\n  -d '{"computer_name":"ios-sim"}'\n# → { "install_code": "ic_..." }\n\n# 在 Mac 上运行 screen-remote-agent\n# (使用安装码完成注册)

第 2 步：在记忆中存储测试场景

\n

把可复用的测试场景作为记忆存储：

\n
import requests\n\ndef store_test_scenario(name, steps, app):\n    requests.post(f"{URL}/memory",\n        headers={"Authorization": f"Bearer {MIND_KEY}"},\n        json={\n            "category": "skill",\n            "key": f"test_scenario_{name}",\n            "content": f"App: {app}\\nSteps:\\n" + "\\n".join(steps),\n            "tags": ["test", "ios", app],\n            "priority": "high"\n        })\n\nstore_test_scenario("login_flow", [\n    "Launch app",\n    "Tap email field",\n    "Type test@example.com",\n    "Tap password field",\n    "Type password123",\n    "Tap Login button",\n    "Verify home screen appears"\n], "MyApp")

第 3 步：LLM 驱动的测试执行

\n

在 Claude Desktop 中（已配置 Synapse MCP）：

\n
Run the login_flow test scenario on the iOS Simulator.\nTake a screenshot after each step and verify the expected UI.\nIf any step fails, store the failure as a memory so we can\navoid it next time.

Claude 会：

\n
    \n
  1. 调用 memory_search 查找 test_scenario_login_flow 记忆
    2. \n
  2. 调用 computer_screenshot 查看当前状态
    3. \n
  3. 通过 computer_command_queue 执行每一步（点击、输入）
    4. \n
  4. 通过截图验证结果
    5. \n
  5. 把任何失败存储为 mistake 记忆
    6. \n
\n

第 4 步：自愈测试

\n

当测试失败时，存储失败信息与恢复方案：

\n
def store_test_failure(scenario, step, error, recovery):\n    requests.post(f"{URL}/memory",\n        headers={"Authorization": f"Bearer {MIND_KEY}"},\n        json={\n            "category": "mistake",\n            "key": f"failure_{scenario}_{step}",\n            "content": f"Scenario: {scenario}\\nStep: {step}\\nError: {error}\\nRecovery: {recovery}",\n            "tags": ["test", "failure", "ios", scenario],\n            "priority": "high"\n        })\n\n# 示例\nstore_test_failure("login_flow", "tap_login",\n    "Login button not found at expected coordinates",\n    "Button moved due to new logo. Search by accessibility label instead.")

下次 LLM 运行该测试时，会回放该失败记忆并自动应用恢复方案。

\n

第 5 步：测试结果跟踪

\n

把测试运行记录为任务：

\n
def track_test_run(scenario, status, duration):\n    requests.post(f"{URL}/mind/task",\n        headers={"Authorization": f"Bearer {MIND_KEY}",\n                 "Content-Type": "application/json"},\n        json={\n            "title": f"Test: {scenario}",\n            "description": f"Status: {status}, Duration: {duration}s",\n            "priority": "normal"\n        })

常用命令

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
操作命令
启动 Simulatorxcrun simctl launch booted com.example.app
截屏computer_screenshot（通过 Synapse MCP）
在 (x,y) 点击computer_command_queue {type:"click", payload:{x,y}}
输入文本computer_command_queue {type:"type", payload:{text:"..."}}
按 Home 键computer_command_queue {type:"key", payload:{keys:["Cmd","Shift","H"]}}
\n

最佳实践

\n

下一步

\n\n","urls":{"html":"/docs/guides/automated-testing-ios","text":"/docs/guides/automated-testing-ios?format=text","json":"/docs/guides/automated-testing-ios?format=json","llm":"/docs/guides/automated-testing-ios?format=llm"},"translations_available":["en","zh","hi","es","fr","ar","pt","ru","ja","de","it","ko","nl","pl","tr","sv","vi","th","id","uk"]}