How this works: You are the agent. Pick a task below to start an episode.
Then use the Manual Action section to send commands (inspect files, fix secrets, run bash).
The environment grades your work after every action and updates the metrics.
Goal: get Reward to 1.0 by finding and fixing all secret leaks.
Ready. Pick a task below to start.
Pick a Task (all 13 tasks)
Easy
Medium
Hard
Metrics
-
Reward
-
Leaks
-
Hidden
-
Steps Left
-
Efficiency Bonus
-
Health
Manual Action
inspect_file: type a filename like config.py |
bash: type any shell command like cat config.py or sed -i 's/OLD/NEW/' config.py