Understand the failure.
Verify the repair.
Small, reproducible software failures. The broken implementation, the fix that didn’t work, and the one that passed—preserved together.
Explore the cases ↓How results are verified ↗WHAT THE ARCHIVE CONTAINS
100840 executable cases. 20168 are open.
Every case records the implementation that fails, the fix that did not work, and the repair that passed its checks—with recorded outputs and source hashes. This release adds 100840 cases across 20168 failure mechanisms and 254 domains.
The open tier gives you the failure and the unsuccessful fix for one case in every mechanism. The remaining 80672 cases, 5 variants per mechanism, are member-only: the verified repair, its recorded checks, and the full fixture suite are held in the member archive. Read the methodology ↗
A RECORD OF WHAT WENT WRONG
Browse the archive / 100840
Context assembly ignores boundary-specific wrapper costs · case 01
Context assembly ignores boundary-specific wrapper costs.
Context assembly ignores boundary-specific wrapper costs · case 02
Context assembly ignores boundary-specific wrapper costs.
Context assembly ignores boundary-specific wrapper costs · case 03
Context assembly ignores boundary-specific wrapper costs.
Context assembly ignores boundary-specific wrapper costs · case 04
Context assembly ignores boundary-specific wrapper costs.
Context assembly ignores boundary-specific wrapper costs · case 05
Context assembly ignores boundary-specific wrapper costs.
Repeated passages from one document remain duplicated in context · case 01
Repeated passages from one document remain duplicated in context.
Repeated passages from one document remain duplicated in context · case 02
Repeated passages from one document remain duplicated in context.
Repeated passages from one document remain duplicated in context · case 03
Repeated passages from one document remain duplicated in context.
Repeated passages from one document remain duplicated in context · case 04
Repeated passages from one document remain duplicated in context.
Repeated passages from one document remain duplicated in context · case 05
Repeated passages from one document remain duplicated in context.
The last chunk is dropped instead of truncated at the remaining budget · case 01
The last chunk is dropped instead of truncated at the remaining budget.
The last chunk is dropped instead of truncated at the remaining budget · case 02
The last chunk is dropped instead of truncated at the remaining budget.
The last chunk is dropped instead of truncated at the remaining budget · case 03
The last chunk is dropped instead of truncated at the remaining budget.
The last chunk is dropped instead of truncated at the remaining budget · case 04
The last chunk is dropped instead of truncated at the remaining budget.
The last chunk is dropped instead of truncated at the remaining budget · case 05
The last chunk is dropped instead of truncated at the remaining budget.
Document chunks are merged across an unrelated source · case 01
Document chunks are merged across an unrelated source.
Document chunks are merged across an unrelated source · case 02
Document chunks are merged across an unrelated source.
Document chunks are merged across an unrelated source · case 03
Document chunks are merged across an unrelated source.
Document chunks are merged across an unrelated source · case 04
Document chunks are merged across an unrelated source.
Document chunks are merged across an unrelated source · case 05
Document chunks are merged across an unrelated source.
Repeated chunk identifiers consume the evidence budget twice · case 01
Repeated chunk identifiers consume the evidence budget twice.
Repeated chunk identifiers consume the evidence budget twice · case 02
Repeated chunk identifiers consume the evidence budget twice.
Repeated chunk identifiers consume the evidence budget twice · case 03
Repeated chunk identifiers consume the evidence budget twice.
Repeated chunk identifiers consume the evidence budget twice · case 04
Repeated chunk identifiers consume the evidence budget twice.
Repeated chunk identifiers consume the evidence budget twice · case 05
Repeated chunk identifiers consume the evidence budget twice.
Centered evidence window is reversed while crossing a boundary · case 01
Centered evidence window is reversed while crossing a boundary.
Centered evidence window is reversed while crossing a boundary · case 02
Centered evidence window is reversed while crossing a boundary.
Centered evidence window is reversed while crossing a boundary · case 03
Centered evidence window is reversed while crossing a boundary.
Centered evidence window is reversed while crossing a boundary · case 04
Centered evidence window is reversed while crossing a boundary.
Centered evidence window is reversed while crossing a boundary · case 05
Centered evidence window is reversed while crossing a boundary.
Context misses metadata overhead for selected evidence · case 01
Context misses metadata overhead for selected evidence.
Context misses metadata overhead for selected evidence · case 02
Context misses metadata overhead for selected evidence.
Context misses metadata overhead for selected evidence · case 03
Context misses metadata overhead for selected evidence.
Context misses metadata overhead for selected evidence · case 04
Context misses metadata overhead for selected evidence.
Context misses metadata overhead for selected evidence · case 05
Context misses metadata overhead for selected evidence.
A feasible context contains fewer complete chunks than possible · case 01
A feasible context contains fewer complete chunks than possible.
A feasible context contains fewer complete chunks than possible · case 02
A feasible context contains fewer complete chunks than possible.
A feasible context contains fewer complete chunks than possible · case 03
A feasible context contains fewer complete chunks than possible.
A feasible context contains fewer complete chunks than possible · case 04
A feasible context contains fewer complete chunks than possible.
A feasible context contains fewer complete chunks than possible · case 05
A feasible context contains fewer complete chunks than possible.
Chunk-local offsets omit the chunk base · case 01
Chunk-local offsets omit the chunk base.
Chunk-local offsets omit the chunk base · case 02
Chunk-local offsets omit the chunk base.
Chunk-local offsets omit the chunk base · case 03
Chunk-local offsets omit the chunk base.
Chunk-local offsets omit the chunk base · case 04
Chunk-local offsets omit the chunk base.
Chunk-local offsets omit the chunk base · case 05
Chunk-local offsets omit the chunk base.
Global citation offsets are sliced without rebasing · case 01
Global citation offsets are sliced without rebasing.
Global citation offsets are sliced without rebasing · case 02
Global citation offsets are sliced without rebasing.
Global citation offsets are sliced without rebasing · case 03
Global citation offsets are sliced without rebasing.
Global citation offsets are sliced without rebasing · case 04
Global citation offsets are sliced without rebasing.
Global citation offsets are sliced without rebasing · case 05
Global citation offsets are sliced without rebasing.
Truncated excerpts retain citations beyond visible text · case 01
Truncated excerpts retain citations beyond visible text.
Truncated excerpts retain citations beyond visible text · case 02
Truncated excerpts retain citations beyond visible text.
Truncated excerpts retain citations beyond visible text · case 03
Truncated excerpts retain citations beyond visible text.
Truncated excerpts retain citations beyond visible text · case 04
Truncated excerpts retain citations beyond visible text.
Truncated excerpts retain citations beyond visible text · case 05
Truncated excerpts retain citations beyond visible text.
Citation starts before the retained window · case 01
Citation starts before the retained window.
Citation starts before the retained window · case 02
Citation starts before the retained window.
Citation starts before the retained window · case 03
Citation starts before the retained window.
Citation starts before the retained window · case 04
Citation starts before the retained window.
Citation starts before the retained window · case 05
Citation starts before the retained window.
Overlapping citations from adjacent chunks remain duplicated · case 01
Overlapping citations from adjacent chunks remain duplicated.
Overlapping citations from adjacent chunks remain duplicated · case 02
Overlapping citations from adjacent chunks remain duplicated.
Overlapping citations from adjacent chunks remain duplicated · case 03
Overlapping citations from adjacent chunks remain duplicated.
Overlapping citations from adjacent chunks remain duplicated · case 04
Overlapping citations from adjacent chunks remain duplicated.
Overlapping citations from adjacent chunks remain duplicated · case 05
Overlapping citations from adjacent chunks remain duplicated.
Equal offsets from distinct sources are merged · case 01
Equal offsets from distinct sources are merged.
Equal offsets from distinct sources are merged · case 02
Equal offsets from distinct sources are merged.
Equal offsets from distinct sources are merged · case 03
Equal offsets from distinct sources are merged.
Equal offsets from distinct sources are merged · case 04
Equal offsets from distinct sources are merged.
Equal offsets from distinct sources are merged · case 05
Equal offsets from distinct sources are merged.
Overlapping chunk citations include the already stitched prefix · case 01
Overlapping chunk citations include the already stitched prefix.
Overlapping chunk citations include the already stitched prefix · case 02
Overlapping chunk citations include the already stitched prefix.
Overlapping chunk citations include the already stitched prefix · case 03
Overlapping chunk citations include the already stitched prefix.
Overlapping chunk citations include the already stitched prefix · case 04
Overlapping chunk citations include the already stitched prefix.
Overlapping chunk citations include the already stitched prefix · case 05
Overlapping chunk citations include the already stitched prefix.
A citation anchor is assigned to every overlapping chunk · case 01
A citation anchor is assigned to every overlapping chunk.
A citation anchor is assigned to every overlapping chunk · case 02
A citation anchor is assigned to every overlapping chunk.
A citation anchor is assigned to every overlapping chunk · case 03
A citation anchor is assigned to every overlapping chunk.
A citation anchor is assigned to every overlapping chunk · case 04
A citation anchor is assigned to every overlapping chunk.
A citation anchor is assigned to every overlapping chunk · case 05
A citation anchor is assigned to every overlapping chunk.
A citation split across chunks is discarded · case 01
A citation split across chunks is discarded.
A citation split across chunks is discarded · case 02
A citation split across chunks is discarded.
A citation split across chunks is discarded · case 03
A citation split across chunks is discarded.
A citation split across chunks is discarded · case 04
A citation split across chunks is discarded.
A citation split across chunks is discarded · case 05
A citation split across chunks is discarded.
Citation IDs point to the pre-merge source table · case 01
Citation IDs point to the pre-merge source table.
Citation IDs point to the pre-merge source table · case 02
Citation IDs point to the pre-merge source table.
Citation IDs point to the pre-merge source table · case 03
Citation IDs point to the pre-merge source table.
Citation IDs point to the pre-merge source table · case 04
Citation IDs point to the pre-merge source table.
Citation IDs point to the pre-merge source table · case 05
Citation IDs point to the pre-merge source table.
The last cited source replaces a higher-priority source · case 01
The last cited source replaces a higher-priority source.
The last cited source replaces a higher-priority source · case 02
The last cited source replaces a higher-priority source.
The last cited source replaces a higher-priority source · case 03
The last cited source replaces a higher-priority source.
The last cited source replaces a higher-priority source · case 04
The last cited source replaces a higher-priority source.
The last cited source replaces a higher-priority source · case 05
The last cited source replaces a higher-priority source.
Citations are deduplicated by label and collapse distinct spans · case 01
Citations are deduplicated by label and collapse distinct spans.
Citations are deduplicated by label and collapse distinct spans · case 02
Citations are deduplicated by label and collapse distinct spans.
Citations are deduplicated by label and collapse distinct spans · case 03
Citations are deduplicated by label and collapse distinct spans.
Citations are deduplicated by label and collapse distinct spans · case 04
Citations are deduplicated by label and collapse distinct spans.
Citations are deduplicated by label and collapse distinct spans · case 05
Citations are deduplicated by label and collapse distinct spans.
INSPECTABLE BY DESIGN
Every result has a runnable source.
Runnable implementations with recorded outputs, source hashes, and explicit contracts. Related variants share a failure mechanism and belong together in evaluation splits.
Read the methodology ↗