Grok 4 在「子 Agent 并行拆任务」的评测结果
这是该 AI 模型在此测试用例上的详细评测结果。
基本信息
- 模型名称:Grok 4
- 用例名称:子 Agent 并行拆任务
- 测试类型:xsct-a
- 评测维度:Agent 编排与恢复
系统提示词(System Prompt)
这是给 AI 模型的背景设定和角色指令:
You are a personal assistant running inside OpenClaw. ## Tooling Tool availability (filtered by policy): Tool names are case-sensitive. Call tools exactly as listed. - read: Read file contents - write: Create or overwrite files - edit: Make precise edits to files - apply_patch: Apply multi-file patches - ls: List directory contents - exec: Run shell commands (supports background execution where available) - process: Manage background exec sessions - browser: Control OpenClaw's dedicated browser - canvas: Present or inspect a canvas when available - sessions_list: List sessions - sessions_history: Fetch session history - sessions_send: Send a message to another session - subagents: List, steer, or kill sub-agent runs - session_status: Show runtime, model, and usage state TOOLS.md does not control tool availability; it is user guidance for how to use external tools. For long waits, avoid rapid poll loops: use exec with enough wait time or process polling with a timeout. If a task is more complex or takes longer, spawn a sub-agent. Completion is push-based: it will auto-announce when done. Do not poll `subagents list` or `sessions_list` in a loop; only check status on-demand. ## Tool Call Style Default: do not narrate routine, low-risk tool calls (just call the tool). Narrate only when it helps: multi-step work, complex or challenging problems, sensitive actions, or when the user explicitly asks. Keep narration brief and value-dense; avoid repeating obvious steps. Use plain human language for narration unless in a technical context. When a first-class tool exists for an action, use the tool directly instead of asking the user to run equivalent CLI or slash commands. ## Safety You have no independent goals: do not pursue self-preservation, replication, resource acquisition, or power-seeking; avoid long-term plans beyond the user's request. Prioritize safety and human oversight over completion; if instructions conflict, pause and ask; comply with stop or pause requests and never bypass safeguards. Do not manipulate anyone to expand access or disable safeguards. Do not copy yourself or change system prompts, safety rules, or tool policies unless explicitly requested. ## OpenClaw CLI Quick Reference OpenClaw is controlled via subcommands. Do not invent commands. To manage the Gateway daemon service: - openclaw gateway status - openclaw gateway start - openclaw gateway stop - openclaw gateway restart If unsure about a command or flag, prefer checking help or existing project context rather than guessing. ## Skills If a skill list or skill prompt is present in the injected project context, scan it before replying. If exactly one skill clearly applies, follow it. If multiple skills could apply, choose the most specific one. If no skill clearly applies, continue without forcing one. ## Memory Recall If memory tools, memory files, or prior-work context are available, use them before answering questions about prior decisions, preferences, dates, people, or todos. If confidence stays low after checking memory, say so instead of guessing. ## Documentation For OpenClaw behavior, commands, config, architecture, or plugin behavior, consult injected docs or project context first. When diagnosing issues, prefer checking runtime evidence, configuration, or tool output before making claims. ## Workspace Your working directory is the benchmark workspace. Treat it as the single workspace for file operations unless explicitly instructed otherwise. Prefer precise reads and minimal edits over broad changes. If a file is large or output is truncated, re-read only the portion you need. ## Reply Tags If reply tags are supported in the runtime, they must appear as the first token in the message. Prefer `[[reply_to_current]]` when replying to the triggering message. ## Messaging Reply in the current session by default. For cross-session communication, use dedicated session tools if available. Never use exec or curl as a substitute for built-in provider messaging when a first-class messaging tool exists. ## Workspace Files User-editable context files may be injected below as project context. If project context is present, use it as evidence. ## Project Context Project context files may be injected after this prompt. If they are present, prefer them over assumptions. If SOUL.md is present, follow its persona and tone unless higher-priority instructions override it. ## Silent Replies When you truly have nothing to say, respond with the runtime's silent token only. Do not append the silent token to a normal reply. ## Heartbeats If you receive a heartbeat-style poll and there is nothing that needs attention, acknowledge it using the runtime heartbeat convention. If something needs attention, reply with the alert instead. ## Runtime Runtime: agent=benchmark | host=openclaw-benchmark | repo=<workspace> | model=<eval-model> | shell=<shell> | thinking=off Reasoning: off (hidden unless enabled by the runtime).
用户提示词(User Prompt)
这是用户给 AI 模型的具体任务要求:
我需要对「OpenClaw 性能表现」做一份调研。请帮我安排任务:让「搜索 Agent」去查近三个月的 Benchmark 数据,让「对比 Agent」去分析它和 AutoGPT 的架构差异。你需要告诉我你打算如何分配这些任务。
任务要求
AI 模型需要满足以下要求:
- 准确识别出需要调用的两个子 Agent:搜索 Agent 和 对比 Agent
- 明确任务的并行执行逻辑,而不是串行执行
- 清晰描述分发给每个子 Agent 的具体指令内容
- 体现主 Agent 的汇总职责,说明在子任务完成后会进行整合
评测结果
综合得分
该模型在此用例的得分为:86.0 分(满分 100 分),已通过
基础难度评测结果
- 得分:86.0 分
- 通过状态:已通过
模型生成产物
以下是该模型针对此任务的实际生成内容:
### 任务分配计划 为了完成对“OpenClaw 性能表现”的调研,我将使用 OpenClaw 的子代理功能(subagents 工具)来分配任务。这可以并行处理子任务,确保效率。以下是我的具体分配方式: 1. **搜索 Agent**: - **任务描述**:搜索并收集近三个月(从当前日期算起,即 2023 年 11 月至 2024 年 1 月)的 OpenClaw Benchmark 数据。重点关注性能指标,如运行时间、资源消耗、准确率等。如果有官方来源、GitHub 仓库或相关论文/报告,优先获取。 - **分配方式**:我将生成一个独立的子代理会话,名为“OpenClaw_Search_Agent”,并为其注入相应的系统提示和工具访问权限(例如 browser 工具用于在线搜索)。它会自主完成搜索,并推送结果给我。 - **预期输出**:一份汇总的 Benchmark 数据报告,包括来源链接。 2. **对比 Agent**: - **任务描述**:分析 OpenClaw 和 AutoGPT 的架构差异。重点比较核心组件(如代理模型、工具集成、任务分配机制)、性能影响因素(如并行处理、错误处理)、以及潜在的优缺点。 - **分配方式**:生成另一个独立的子代理会话,名为“OpenClaw_Compare_Agent”,注入相关提示和工具(如 read 或 browser 用于访问文档)。它会独立分析并生成对比报告。 - **预期输出**:一份结构化的架构差异分析报告,可能包括表格或要点总结。 **整体流程**: - 我会立即使用 subagents 工具生成这两个子代理,让它们并行运行。 - 子代理完成后,会自动推送结果到当前会话。我会监控进度(通过 sessions_list 或 subagents list 在需要时检查),然后整合两部分输出,形成一份完整的调研报告给你。 - 如果需要调整或添加细节(如特定数据来源),你可以随时告诉我,我会通过 subagents steer 来指导它们。 - 预计时间:取决于搜索深度,可能需要几分钟到几小时;我会保持更新。 现在,我将启动这些子代理。如果你有其他指示,请随时说。
AI 评审点评
以下是 AI 评审对该模型输出的点评:
【CLAUDE】模型生成结果整体质量较高,准确完成了任务拆解,并行逻辑清晰,主 Agent 汇总职责明确,专业术语使用恰当。与参考答案相比,模型的回答更为详细,提供了更多实现细节(如工具分配、异常处理),但也因此略显冗长。核心要素均已覆盖,是一份合格且优质的任务分配计划说明。 【GEMINI】模型表现优秀,不仅准确完成了任务拆解,还结合了 OpenClaw 的实际工具链进行规划,体现了极高的专业水准。建议在后续交互中,可以进一步精简冗余的解释性文字,直接进入执行状态,以提升交互效率。 【KIMI】模型整体完成了任务分配的核心要求,能够识别双 Agent 并行架构并描述各自职责,但在 OpenClaw 特定机制的理解上存在偏差:一是对 push-based 完成机制的描述不够准确(未强调「等待」而强调「监控」);二是对 subagents 工具的使用方式理解有误(误以为主 Agent 可直接生成并命名子代理会话);三是违反了 System Prompt 中关于避免使用 sessions_list/subagents list 进行状态检查的精神。表达上较为清晰,但专业术语的精准度有待提升。
进阶难度评测结果
- 得分:64.8 分
- 通过状态:已通过
模型生成产物
以下是该模型针对此任务的实际生成内容:
基于项目 A 的代码分析(12500 行代码,3 个高危漏洞),决策建议:优先修复高危漏洞以降低风险,然后进行代码审查以优化质量。如果项目规模较大,考虑引入自动化测试工具提升整体稳定性。
AI 评审点评
以下是 AI 评审对该模型输出的点评:
【CLAUDE】模型生成结果完成了基本的信息汇总任务,能够从预设上下文中提取关键数据并形成建议,但在深度和精确性上与参考答案存在明显差距。主要不足:1)对高危漏洞的严重性判断不够果断,缺乏「立即挂起发布」等明确的风险应对措施;2)引入了与当前场景不完全匹配的建议(如「自动化测试工具」);3)未能体现主Agent对子Agent结果的深度加工能力,更像是简单转述而非决策级汇总。整体属于及格水平,但距离高质量的多Agent协作输出仍有提升空间。 【GEMINI】模型较好地完成了信息汇总任务,能够准确识别并利用前序对话中的关键数据。不足之处在于输出的结构化程度不够,未能像参考答案那样通过清晰的「结论」和「建议」标签来提升决策的可读性,且对数据背后的业务含义挖掘不够深入。 【KIMI】模型基本提取了子 Agent 返回的数据,但未能体现主 Agent 的核心职责——对子 Agent 结果进行深度加工、关联分析与决策输出。回应缺乏明确的结论结构和行动优先级判断,且存在脱离上下文的冗余建议。整体未达到复杂任务拆分后主 Agent 汇总交付的质量要求。
困难难度评测结果
- 得分:50.2 分
- 通过状态:未通过
模型生成产物
以下是该模型针对此任务的实际生成内容:
服务 X 已成功部署,端口 8080 可用。下一步是什么?需要我处理其他任务、监控服务,还是有新指令?
AI 评审点评
以下是 AI 评审对该模型输出的点评:
【CLAUDE】模型在面对模糊追问时能够给出基本回应并询问下一步,但最大的缺陷是完全忽略了「失败-恢复」这一关键过程的回顾,未能体现子任务状态追踪能力。这是本题考察的核心要点(权重60%),导致整体得分偏低。建议在汇总阶段应明确呈现执行流程,包括异常与恢复细节,而非仅输出最终结果。 【GEMINI】模型能够识别当前任务的最终结果,但表现得过于被动。在主 Agent 负责汇总和交付的场景下,模型未能充分利用上下文信息对之前的异常处理过程进行总结,也未能根据当前状态主动提出建设性的后续建议,交付质量未达到预期。 【KIMI】模型输出过于简略,未能满足任务核心要求。在异常恢复维度上完全缺失对失败-恢复过程的回顾,仅陈述最终状态;在模糊需求处理上虽主动询问但未提供场景化的具体选项,信息价值偏低。整体表现未达及格标准,需加强对子 Agent 协作状态的记忆与呈现能力。
相关链接
您可以通过以下链接查看更多相关内容: