I am researching how major international news organizationsNews & Media Publishers
Replay
Every recorded step of a run, as the agent saw the screen: the frame, the action it took, what it said, and which rubrics the judge's evidence had reached by then. 4 runs recorded with screenshots.
Recorded browser path
Claude Opus 5.5
- 100%
- Rubric score
- 100
- Steps
- 10/10
- Rubrics passed
Task promptI am researching how major international news organizations cover climate-related extreme weather and want to build a reusable monitoring methodology rather than simply collecting articles. Compare coverage from at least 12 major publishers across three months, classify stories by event type, geography, framing, use of scientific evidence, attribution, and solutions coverage, and identify patterns while avoiding claims that cannot be supported by the sample. Develop a reproducible research workflow and a structured dataset.
I am researching how major international news organizations cover climate-related extreme weather and want to build a reusable monitoring methodology rather than simply collecting articles. Compare coverage from at least 12 major publishers across three months, classify stories by event type, geography, framing, use of scientific evidence, attribution, and solutions coverage, and identify patterns while avoiding claims that cannot be supported by the sample. Develop a reproducible research workflow and a structured dataset.