Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): (None, {}), NamedSplit('test'): ('text', {})}
Error code: FileFormatMismatchBetweenSplitsError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
SkillsBench Trajectories — April 2026
Benchmark trajectories for Claude Code on SkillsBench.
Models
- Claude Opus 4.7 (
claude-opus-4-7) - Claude Sonnet 4.6 (
claude-sonnet-4-6)
Conditions
opus47-with-skills-t1— Opus 4.7 with skills, trial 1opus47-without-skills-t1— Opus 4.7 without skills, trial 1sonnet46-with-skills-t1— Sonnet 4.6 with skills, trial 1sonnet46-without-skills-t1— Sonnet 4.6 without skills, trial 1
Auth
Claude Code OAuth subscription (Max 20x tier)
Environment
Daytona cloud sandboxes
Trial 1 Results (in progress)
| Condition | Passed | Failed | Errored | Raw Score |
|---|---|---|---|---|
| Opus 4.7 +skills | 35 | 34 | 13 | 42.7% |
| Opus 4.7 -skills | 23 | 45 | 14 | 28.0% |
| Sonnet 4.6 +skills | TBD | |||
| Sonnet 4.6 -skills | TBD |
Structure
jobs/
<condition>-t<trial>/
<job_timestamp>/
<task_name>__<hash>/
result.json # rewards, errors, timing
config.json # run configuration
trajectory/
acp_trajectory.jsonl # full agent trace
agent/
claude_agent_acp.txt # agent log
verifier/
reward.txt # final score
ctrf.json # test results
- Downloads last month
- 4,960