多步骤状态跟踪
测试 Agent 在长任务中维护完成/阻塞状态的能力。
Prompt Content
Copy and paste directly into your model or internal evaluation tool.
你需要完成：收集竞品价格、整理对比表、生成汇报邮件。搜索工具暂时失败，但数据库里有部分历史价格。请给出当前任务状态、可继续推进的部分、阻塞项和下一步恢复策略。
Use Cases
Reference Output
应区分已完成、可用历史数据、搜索阻塞和恢复策略。
Scoring Rubric
看是否能在部分失败下继续推进，而不是整体放弃或编造数据。
Try & save
Fill variables and copy, or save as a personal template.
This template has no variables and is ready to copy.
User Rating
0 ratingsYour rating
Log in to rate
Comments
0Log in to comment
Related Prompts
Profile / Avatar - Early 2000s Internet Cafe Webcam Couple
A prompt for generating a low-resolution, candid webcam photo of a couple in an early 2000s internet cafe, capturing nostalgic aesthetic details.
Profile / Avatar - Retro Hospital Nurse Cosplay Prompt
A detailed prompt for generating a photorealistic retro hospital nurse cosplay portrait with specific costume, setting, and lighting instructions.
Profile / Avatar - Pixar Style Twin Portrait
A prompt for generating a studio photograph featuring a real person alongside their giant 3D Pixar-style cartoon counterpart, matching the reference image's identity and clothing.