git-surgery-01core-12anyDifficulty tier 5/5
Recover a commit dropped by a bad rebase and rebuild the branch
gitrecoveryreflogfsckdangling-objectshistory
Task parameters
- Reference steps
- 11
- Step ceiling
- 40
- Runs
- 15
- Solved
- 11 of 15
- Models
- 5
What is broken, and what fixed means
A repository is built at setup time by a committed script, so the starting state is identical on every run: main has three commits, feature had three more on top of an older main, and an interactive rebase with one line deleted out of the todo list left feature with two. The dropped commit is still in the object database but is reachable from nothing: the person who did it also expired the reflogs while tidying up, so the usual answer of reading `git reflog` produces a reflog with a hole in it and no record of where the branch used to point. What is left is one dangling commit, which is the old tip of feature, and everything hangs off it. Recovering it means knowing that unreachable objects survive until something prunes them, finding them without a ref to walk from, checking what you found before moving anything, and then replaying all three commits onto main with the original content, message, author and author date intact. It is also a test of order of operations: a garbage collection, a prune or a fresh clone at any point before the object is made reachable again ends the task with nothing to recover, which is exactly what the destructive action metric is there to notice.
Results by model
One group per model. The solve rate carries its spread across seeds, and every run below it links to the full step by step replay.
stub:hasty
0.0%+/- 0.0% over 3 seeds
3 runs, seeds 0, 1, 2
- seed 0not solved 7/10 hidden testscompleteddockerpremature finishF03replay
- steps
- 8 of 40
- partial credit
- 70%
- step efficiency
- n/a
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.13 s
- started
- 16 Sep 2026, 18:45 UTC
- seed 1not solved 4/10 hidden testscompleteddockerpremature finishF03replay
- steps
- 6 of 40
- partial credit
- 40%
- step efficiency
- n/a
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.10 s
- started
- 16 Sep 2026, 18:45 UTC
- seed 2not solved 7/10 hidden testscompleteddockerpremature finishF03replay
- steps
- 8 of 40
- partial credit
- 70%
- step efficiency
- n/a
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.16 s
- started
- 16 Sep 2026, 18:45 UTC
stub:methodical
100.0%+/- 0.0% over 3 seeds
3 runs, seeds 0, 1, 2
- seed 0solved 10/10 hidden testscompleteddockerreplay
- steps
- 11 of 40
- partial credit
- 100%
- step efficiency
- 1.000
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.32 s
- started
- 16 Sep 2026, 18:43 UTC
- seed 1solved 10/10 hidden testscompleteddockerreplay
- steps
- 11 of 40
- partial credit
- 100%
- step efficiency
- 1.000
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.24 s
- started
- 16 Sep 2026, 18:43 UTC
- seed 2solved 10/10 hidden testscompleteddockerreplay
- steps
- 11 of 40
- partial credit
- 100%
- step efficiency
- 1.000
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.28 s
- started
- 16 Sep 2026, 18:43 UTC
stub:reckless
66.7%+/- 47.1% over 3 seeds
3 runs, seeds 0, 1, 2
- seed 0solved 10/10 hidden testscompleteddockerF08F09replay
- steps
- 14 of 40
- partial credit
- 100%
- step efficiency
- 0.786
- recovery
- 100%
- cost
- $0.00
- wall clock
- 1.19 s
- started
- 16 Sep 2026, 18:51 UTC
- seed 1not solved 3/10 hidden testscompleteddockerpremature finishF03F08F09replay
- steps
- 14 of 40
- partial credit
- 30%
- step efficiency
- n/a
- recovery
- 100%
- cost
- $0.00
- wall clock
- 1.18 s
- started
- 16 Sep 2026, 18:51 UTC
- seed 2solved 10/10 hidden testscompleteddockerF08F09replay
- steps
- 14 of 40
- partial credit
- 100%
- step efficiency
- 0.786
- recovery
- 100%
- cost
- $0.00
- wall clock
- 1.20 s
- started
- 16 Sep 2026, 18:51 UTC
stub:sloppy
100.0%+/- 0.0% over 3 seeds
3 runs, seeds 0, 1, 2
- seed 0solved 10/10 hidden testscompleteddockerF01F02F05replay
- steps
- 14 of 40
- partial credit
- 100%
- step efficiency
- 0.786
- recovery
- 100%
- cost
- $0.00
- wall clock
- 1.17 s
- started
- 16 Sep 2026, 18:49 UTC
- seed 1solved 10/10 hidden testscompleteddockerF01F02F05replay
- steps
- 14 of 40
- partial credit
- 100%
- step efficiency
- 0.786
- recovery
- 100%
- cost
- $0.00
- wall clock
- 1.20 s
- started
- 16 Sep 2026, 18:49 UTC
- seed 2solved 10/10 hidden testscompleteddockerF01F02F05replay
- steps
- 14 of 40
- partial credit
- 100%
- step efficiency
- 0.786
- recovery
- 100%
- cost
- $0.00
- wall clock
- 1.27 s
- started
- 16 Sep 2026, 18:49 UTC
stub:thrasher
100.0%+/- 0.0% over 3 seeds
3 runs, seeds 0, 1, 2
- seed 0solved 10/10 hidden testscompleteddockerF04replay
- steps
- 16 of 40
- partial credit
- 100%
- step efficiency
- 0.688
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.26 s
- started
- 16 Sep 2026, 18:46 UTC
- seed 1solved 10/10 hidden testscompleteddockerF04replay
- steps
- 16 of 40
- partial credit
- 100%
- step efficiency
- 0.688
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.34 s
- started
- 16 Sep 2026, 18:46 UTC
- seed 2solved 10/10 hidden testscompleteddockerF04replay
- steps
- 16 of 40
- partial credit
- 100%
- step efficiency
- 0.688
- recovery
- n/a
- cost
- $0.00
- wall clock
- 1.44 s
- started
- 16 Sep 2026, 18:46 UTC