Dataset Viewer
Auto-converted to Parquet Duplicate
adapter_backend_path
stringclasses
3 values
adapter_implementation_path
stringclasses
3 values
adapter_kind
stringclasses
1 value
adapter_module_hashes_path
stringclasses
3 values
adapter_plan_schema
stringclasses
3 values
adapter_restore_path
stringclasses
2 values
adapter_seed_sha256
stringlengths
64
64
⌀
cloud_launch_ready
bool
2 classes
dependency_lock_sha256
stringclasses
88 values
difficulty_status
stringclasses
1 value
evaluation_only
bool
1 class
execution_path
stringlengths
60
63
gate_passed
bool
1 class
gate_report_path
stringlengths
71
74
generated_alternatives_run
bool
1 class
generated_reference
bool
1 class
image_pull_qualified
bool
1 class
image_receipt_path
stringlengths
57
60
in_stratified_sample
bool
2 classes
instruction
stringlengths
451
26.6k
language
stringclasses
2 values
leak_audit_run
bool
1 class
provenance_path
stringlengths
64
67
provided_reference
bool
1 class
provider_image_source_attestation
bool
1 class
qwen_difficulty
float64
0
1
⌀
qwen_protocol
stringclasses
3 values
qwen_solve_rate
float64
0
1
⌀
qwen_solved
int64
0
8
⌀
qwen_status
stringclasses
4 values
qwen_valid_attempts
int64
4
8
⌀
qwen_wilson95_high
float64
0.32
1
⌀
qwen_wilson95_low
float64
0
0.68
⌀
rfc008_complete
bool
1 class
semantic_nonoverlap_established
bool
1 class
source_commit
stringlengths
40
40
source_dataset
stringclasses
3 values
source_license_spdx
stringclasses
4 values
source_pr_url
stringclasses
130 values
source_repo
stringclasses
27 values
source_revision
stringclasses
3 values
source_split
stringclasses
1 value
target_model
stringclasses
1 value
target_model_calibrated
bool
1 class
task_bundle_license_spdx
stringclasses
2 values
task_file_inventory_path
stringlengths
62
65
task_id
stringlengths
11
64
task_inventory_sha256
stringlengths
64
64
task_path
stringlengths
51
54
task_source_sha256
stringlengths
64
64
terminal_proof_path
stringlengths
68
68
⌀
trainability_established
bool
1 class
training_admissible
bool
1 class
upstream_repo
stringclasses
27 values
usage
stringclasses
1 value
validation_profile
stringclasses
1 value
verified
bool
1 class
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/0379675216b16a1676ff1bef.json
true
scale800_v1/20261112/gate-evidence/0379675216b16a1676ff1bef/report.json
false
false
true
scale800_v1/20261112/images/0379675216b16a1676ff1bef.json
false
UPDATE query generates incorrect SQL with subqueries and expressions When using UPDATE queries with subqueries or complex expressions, the generated SQL wraps all values in parentheses incorrectly. This causes subqueries to lose their required parentheses while simple expressions get unnecessary ones. #### Steps to R...
go
false
scale800_v1/20261112/tasks/0379675216b16a1676ff1bef/provenance.json
true
null
0.75
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.25
1
scored
4
0.699358
0.045587
false
false
2df379c9c99de79a1483ac96f1f196e3ba239b50
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/0379675216b16a1676ff1bef.json
Masterminds__squirrel.1ded5784.func_pm_ctrl_invert_if__coeoegdw
e8cb014e78abda927762d1b8ab96e5b8a6a1a0f4f1656973bcfc5256f0a4a914
scale800_v1/20261112/tasks/0379675216b16a1676ff1bef
e8fae4b316bc754ef85c76f2ad5d448c84a7dbb394d62bf2a5931ec08947bcd5
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/204fec1c84ae0f2e06a9e22a.json
true
scale800_v1/20261112/gate-evidence/204fec1c84ae0f2e06a9e22a/report.json
false
false
true
scale800_v1/20261112/images/204fec1c84ae0f2e06a9e22a.json
false
LIKE operators incorrectly reject non-null values #### Description When using LIKE operators with non-null values, the query builder incorrectly throws an error "cannot use null with like operators" even when the value is not null. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.com/...
go
false
scale800_v1/20261112/tasks/204fec1c84ae0f2e06a9e22a/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
05d652fd412e240cef4d9374d963f770cf6f86a4
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/204fec1c84ae0f2e06a9e22a.json
Masterminds__squirrel.1ded5784.func_pm_ctrl_invert_if__qz42ljwj
67c1d4be571b31d80e68fe097eef0987166357daec961c1089096f6087c588ad
scale800_v1/20261112/tasks/204fec1c84ae0f2e06a9e22a
58424fd2fa34ace929074c4cd99b070c637c393ed1258ed79532f8c20eeaaae4
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/e13424c206f80ad064d19122.json
true
scale800_v1/20261112/gate-evidence/e13424c206f80ad064d19122/report.json
false
false
true
scale800_v1/20261112/images/e13424c206f80ad064d19122.json
false
Update statements fail with "must specify a table" error even when table is provided #### Description When trying to build UPDATE SQL statements using the UpdateBuilder, all update operations are failing with the error "update statements must specify a table" even when a table name is clearly provided via the `Table(...
go
false
scale800_v1/20261112/tasks/e13424c206f80ad064d19122/provenance.json
true
null
null
null
null
null
not_started
null
null
null
false
false
7b9f2cb24fa54fcfcd394e2cfe64a176ae9cf462
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/e13424c206f80ad064d19122.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__55rcqsj4
bafb2446cb156e9309f6275898277375c8f30590beecf9b7f4c576701ae7348a
scale800_v1/20261112/tasks/e13424c206f80ad064d19122
84012ca014f665de0e5428727f65d4f3ab336800fdfc6ba9906c339f7c6caee8
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/4b59905844c4c20d628a8fba.json
true
scale800_v1/20261112/gate-evidence/4b59905844c4c20d628a8fba/report.json
false
false
true
scale800_v1/20261112/images/4b59905844c4c20d628a8fba.json
false
SelectBuilder ExecContext returns wrong error when RunWith is set #### Description When using SelectBuilder with ExecContext and a RunWith runner is properly set, the method incorrectly returns a "cannot run; no Runner set (RunWith)" error instead of executing the query or returning the appropriate context-related er...
go
false
scale800_v1/20261112/tasks/4b59905844c4c20d628a8fba/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
818f6f5954faf29c197023d34a43c01f9914deaf
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/4b59905844c4c20d628a8fba.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__617tm58v
bff3f1d3df8dd1807f31cdacb9bdad4d34f2fcc7c0be1373f5ffcc3a2e4ee7a2
scale800_v1/20261112/tasks/4b59905844c4c20d628a8fba
afd2f6c944f83ac7fcc2626ab278c650e56c547bada40f7e21463925992f759f
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/63a1ee21d2764851ad2f8a2f.json
true
scale800_v1/20261112/gate-evidence/63a1ee21d2764851ad2f8a2f/report.json
false
false
true
scale800_v1/20261112/images/63a1ee21d2764851ad2f8a2f.json
false
UpdateBuilder.MustSql() doesn't panic on valid SQL #### Description The `UpdateBuilder.MustSql()` method is supposed to panic when there's an error generating SQL, but it's currently panicking when the SQL generation is successful instead. #### Steps/Code to Reproduce ```go package main import ( "github.com/Ma...
go
false
scale800_v1/20261112/tasks/63a1ee21d2764851ad2f8a2f/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
8cf0f2d6e8bf28d318aead941a19eb8b34202a3f
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/63a1ee21d2764851ad2f8a2f.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__6rsggjlj
0e8e1c175a5c0d78959fe1c5b48d87158b18d6268d7821f47c91fbf0cde59058
scale800_v1/20261112/tasks/63a1ee21d2764851ad2f8a2f
87ec3567ae51e95b41f773de6e392f90c2c7d189c54cf465f4c6af542ff73177
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/228d31f82de518ad7ce9e2a3.json
true
scale800_v1/20261112/gate-evidence/228d31f82de518ad7ce9e2a3/report.json
false
false
true
scale800_v1/20261112/images/228d31f82de518ad7ce9e2a3.json
false
SQL CASE statements generate empty clauses #### Description When building SQL queries with CASE statements, the generated SQL contains empty WHEN/THEN clauses instead of the expected values and conditions. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.com/Masterminds/squirrel" ) f...
go
false
scale800_v1/20261112/tasks/228d31f82de518ad7ce9e2a3/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
b6b30d085883cb57a5062a0579bde23f642f1f85
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/228d31f82de518ad7ce9e2a3.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__6z3s0vh2
828a3ef916b109ec4c5c7a5b00eeae42c2847cf66d80c1ff64f9c96c5c9f2024
scale800_v1/20261112/tasks/228d31f82de518ad7ce9e2a3
53838b3301ecb669b22b66e5cf83b210d377b4ade70d675ba7abc3f230bd98d3
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/aab73369ea0473ac7a23c937.json
true
scale800_v1/20261112/gate-evidence/aab73369ea0473ac7a23c937/report.json
false
false
true
scale800_v1/20261112/images/aab73369ea0473ac7a23c937.json
false
QueryRowContext returns wrong error when Runner is set #### Description When using `QueryRowContext` with a Runner that is set but doesn't implement `QueryRowerContext`, the method returns the wrong error message. Instead of returning "cannot run; no Runner set (RunWith)" it returns "cannot QueryRow; Runner is not a ...
go
false
scale800_v1/20261112/tasks/aab73369ea0473ac7a23c937/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
fb5137d39f1e9b6fbab92c44814483a300349b61
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/aab73369ea0473ac7a23c937.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__7w8cqbkc
47f7672dd2518efe67fb0d07def373fc809bf0fe72898ecbcb2a91d0dc18287c
scale800_v1/20261112/tasks/aab73369ea0473ac7a23c937
667ca23029e7e70d58bb81fdb93ca99505bca8b0feb576b6186e95d515682654
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/291e43155072beb9d3f5fc55.json
true
scale800_v1/20261112/gate-evidence/291e43155072beb9d3f5fc55/report.json
false
false
true
scale800_v1/20261112/images/291e43155072beb9d3f5fc55.json
false
Empty Eq map generates incorrect SQL When using an empty `Eq` map in WHERE clauses, the generated SQL is incorrect. Instead of evaluating to true (which should be the case for an empty condition map), it's generating `(1=1)` in all cases. #### Steps to Reproduce ```go package main import ( "fmt" "github.com...
go
false
scale800_v1/20261112/tasks/291e43155072beb9d3f5fc55/provenance.json
true
null
null
qwen38-auto48-64k-2m-nothink-r3-a4-v1
null
null
scoring
null
null
null
false
false
6ab105238a430c7c63c0c965863d41d6f671a32a
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/291e43155072beb9d3f5fc55.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__c7xrz7e0
d087701bfdbb4c1cc4f695ee343e771a9c8585b67193438812eb257663d04bfb
scale800_v1/20261112/tasks/291e43155072beb9d3f5fc55
cb7061522e84ae2ebec5ee69cf1766e5c8cf8f3d27418cf758553ad358743f2a
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/4ad0407ebf71072a9c73462d.json
true
scale800_v1/20261112/gate-evidence/4ad0407ebf71072a9c73462d/report.json
false
false
true
scale800_v1/20261112/images/4ad0407ebf71072a9c73462d.json
false
Delete operations fail when no runner is set #### Description When attempting to execute a DELETE query without setting a runner, the operation fails with a nil pointer dereference instead of returning the expected `RunnerNotSet` error. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github...
go
false
scale800_v1/20261112/tasks/4ad0407ebf71072a9c73462d/provenance.json
true
null
null
qwen38-auto48-64k-2m-nothink-r3-a4-v1
null
null
scoring
null
null
null
false
false
5b9af1361c4da745505126677e881bbddffbb72b
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/4ad0407ebf71072a9c73462d.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__csy4cg8m
a835247836607e9763a27000aab9518f3041c6933f3341a884bc289d99ed38e9
scale800_v1/20261112/tasks/4ad0407ebf71072a9c73462d
4fe4b123c40c4ed1da38284f9f9d84a04d40e09ea0487bb7c484e13044d0a2a3
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/5a0f91c6659cf8c75ae685f2.json
true
scale800_v1/20261112/gate-evidence/5a0f91c6659cf8c75ae685f2/report.json
false
false
true
scale800_v1/20261112/images/5a0f91c6659cf8c75ae685f2.json
false
Row.Scan() returns nil error when it should return the actual error #### Description When calling `Scan()` on a `Row` instance that has an error, the method incorrectly returns `nil` instead of the actual error. This causes nil pointer dereferences when the calling code expects an error to be returned but gets `nil` ...
go
false
scale800_v1/20261112/tasks/5a0f91c6659cf8c75ae685f2/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
265974711a78f0cb35eec179977ffb6883a460f0
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/5a0f91c6659cf8c75ae685f2.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__ox0ispzd
07d90220c2edb5954c93b8c3ea4630e86a9e8d6785b3e6f1dfdea6f16052de77
scale800_v1/20261112/tasks/5a0f91c6659cf8c75ae685f2
d7d4da0816c8f5e3aa312695203fc77db27910007dc5be1a66f2182274f19b1a
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/2f0dab40d3f2524b799cd56a.json
true
scale800_v1/20261112/gate-evidence/2f0dab40d3f2524b799cd56a/report.json
false
false
true
scale800_v1/20261112/images/2f0dab40d3f2524b799cd56a.json
false
INSERT statements broken - validation logic inverted #### Description INSERT statement generation is completely broken. All INSERT queries fail with validation errors even when they should be valid. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.com/Masterminds/squirrel" ) func mai...
go
false
scale800_v1/20261112/tasks/2f0dab40d3f2524b799cd56a/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
1046c2496381ce80bec1ba4722b51258e2858be7
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/2f0dab40d3f2524b799cd56a.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__pg3kixez
1985940f0d5e0895088a110a7fca38b89363437979394a214437aab54731dc22
scale800_v1/20261112/tasks/2f0dab40d3f2524b799cd56a
df58d2b9ec1c2bb7cc945303d6e184263fee526b5aa4e7cdde10e146bf16e315
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/7438abb7d48a4dfbae8173a1.json
true
scale800_v1/20261112/gate-evidence/7438abb7d48a4dfbae8173a1/report.json
false
false
true
scale800_v1/20261112/images/7438abb7d48a4dfbae8173a1.json
false
SQL query arguments not being collected properly #### Description When building SQL queries, the arguments array is not being populated correctly, resulting in `nil` arguments being returned instead of the expected parameter values. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.com...
go
false
scale800_v1/20261112/tasks/7438abb7d48a4dfbae8173a1/provenance.json
true
null
0.5
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.5
2
scored
4
0.849961
0.150039
false
false
506808060c00d119c96c7c50788628bdace8744e
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/7438abb7d48a4dfbae8173a1.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__vmu4mbav
ca109db2436e1e84b671f3725366ab253f9be92ebb1efbafccca2caca68e3d23
scale800_v1/20261112/tasks/7438abb7d48a4dfbae8173a1
a706328980f3a3b5f72c68ccb83c6d0faf3d50d6166f20d7fdb37311993d3518
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/f3845f49027c4c0172c7108b.json
true
scale800_v1/20261112/gate-evidence/f3845f49027c4c0172c7108b/report.json
false
false
true
scale800_v1/20261112/images/f3845f49027c4c0172c7108b.json
false
Where clauses not being added when predicate is non-nil and non-empty #### Description The Where method in SelectBuilder is not properly adding WHERE clauses when the predicate is a valid non-nil, non-empty value. This affects all SQL generation that relies on WHERE conditions. #### Steps/Code to Reproduce ```go pa...
go
false
scale800_v1/20261112/tasks/f3845f49027c4c0172c7108b/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
1536e9193db34d5ef53bcd21acae460c703f5f60
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/f3845f49027c4c0172c7108b.json
Masterminds__squirrel.1ded5784.func_pm_flip_operators__ya57pp6o
3407e7d8b8a21a258e3797867404aaab363c250b4fae0cd74c0b9b0f0275828b
scale800_v1/20261112/tasks/f3845f49027c4c0172c7108b
e1b436a9db98a7c3889f39b1d6d8b0819dad0fd4fa3ab093e5942059a6390c01
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/dbbc5c592b066b5c088db026.json
true
scale800_v1/20261112/gate-evidence/dbbc5c592b066b5c088db026/report.json
false
false
true
scale800_v1/20261112/images/dbbc5c592b066b5c088db026.json
false
Placeholder replacement causes slice bounds out of range panic #### Description When processing SQL strings with escaped question marks (`??`) at the end of the string, the placeholder replacement function panics with a "slice bounds out of range" error. #### Steps/Code to Reproduce ```go package main import ( ...
go
false
scale800_v1/20261112/tasks/dbbc5c592b066b5c088db026/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
54070447c816d06f4ac4d50c48997eb8e5cd1f85
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/dbbc5c592b066b5c088db026.json
Masterminds__squirrel.1ded5784.func_pm_op_change__sdfcg12h
906d26cd3f0781554b2820fd4960460e5a482a984d72868ebae6eb3e84267945
scale800_v1/20261112/tasks/dbbc5c592b066b5c088db026
aaf68c2c53f263c84d0cca258062a25643f9ce68dc042aeafdfe57d89b8f3ca6
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/a68c99a609480287442d62ac.json
true
scale800_v1/20261112/gate-evidence/a68c99a609480287442d62ac/report.json
false
false
true
scale800_v1/20261112/images/a68c99a609480287442d62ac.json
false
UPDATE statements generate extra WHERE, LIMIT, OFFSET clauses when they shouldn't #### Description When building UPDATE statements with squirrel, the generated SQL includes empty WHERE, LIMIT, and OFFSET clauses even when none were specified. This results in malformed SQL like `UPDATE table SET x = '1', y = 'val' WHE...
go
false
scale800_v1/20261112/tasks/a68c99a609480287442d62ac/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
a6dfd33efdda2f90eb5dd43bba26d4c2ea5ccc08
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/a68c99a609480287442d62ac.json
Masterminds__squirrel.1ded5784.func_pm_op_change_const__1lejzmql
373ef534ca1e5c00e8f179d288580eb0d7511010c731ca22fb752cb2d192894f
scale800_v1/20261112/tasks/a68c99a609480287442d62ac
d002dc5f5110a013aa5cac664414df2a63e214fb33aaf923eb904904c2e63907
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/5e15e16e1f0fbe96e2283896.json
true
scale800_v1/20261112/gate-evidence/5e15e16e1f0fbe96e2283896/report.json
false
false
true
scale800_v1/20261112/images/5e15e16e1f0fbe96e2283896.json
true
INSERT statements with suffixes not properly handled #### Description When building INSERT statements with suffixes (like RETURNING clauses), the suffix is not being appended to the generated SQL when there's only one suffix present. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.co...
go
false
scale800_v1/20261112/tasks/5e15e16e1f0fbe96e2283896/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-v1
1
8
scored
8
1
0.675592
false
false
b8011f33870c224ca154fe59db235071027b29f8
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/5e15e16e1f0fbe96e2283896.json
Masterminds__squirrel.1ded5784.func_pm_op_change_const__4mi9lm2k
6dd7832cb56508606b747d63bb67706deb74faeffc78859ccaadde2b626df394
scale800_v1/20261112/tasks/5e15e16e1f0fbe96e2283896
267b2fc2fb50a6b434627db2e9b026693fb19897aabf63934dc0b5d87c9b80da
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/70706cc1cfa0288ad094ab2a.json
true
scale800_v1/20261112/gate-evidence/70706cc1cfa0288ad094ab2a/report.json
false
false
true
scale800_v1/20261112/images/70706cc1cfa0288ad094ab2a.json
true
Escaped placeholder handling broken in expr.ToSql() #### Description When using escaped placeholders (`??`) in SQL expressions, the placeholder escaping logic is not working correctly. The escaped placeholders are not being processed properly, leading to incorrect SQL output and potential runtime panics. #### Steps/...
go
false
scale800_v1/20261112/tasks/70706cc1cfa0288ad094ab2a/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-v1
1
8
scored
8
1
0.675592
false
false
9b2588afb6a9569252397d1e9076ddde4e61fc25
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/70706cc1cfa0288ad094ab2a.json
Masterminds__squirrel.1ded5784.func_pm_op_change_const__al8x4eaq
e9912aebeaae7b1c00fc426d5cdf408bf6f58a1a73541e044bbedbb173e3502e
scale800_v1/20261112/tasks/70706cc1cfa0288ad094ab2a
a24548c24dd5e2ad6349c47822f3f569bfcc01099e40359555c24307fce47e11
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/97f7062db477be020f45124b.json
true
scale800_v1/20261112/gate-evidence/97f7062db477be020f45124b/report.json
false
false
true
scale800_v1/20261112/images/97f7062db477be020f45124b.json
false
DELETE query builder generates invalid SQL with empty OFFSET clause #### Description When building DELETE queries, the query builder incorrectly appends an empty "OFFSET " clause to the generated SQL, resulting in malformed queries. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.com...
go
false
scale800_v1/20261112/tasks/97f7062db477be020f45124b/provenance.json
true
null
1
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0
0
scored
4
0.489891
0
false
false
fa1ad2a03153e8afb1557865f5ca3467ad7785a5
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/97f7062db477be020f45124b.json
Masterminds__squirrel.1ded5784.func_pm_op_change_const__jzhulypy
e63a0431ef1c634b841f5dd5161349d04468fd65ab041e351ba5a216ef3088c1
scale800_v1/20261112/tasks/97f7062db477be020f45124b
d9affb4609dcef8366e86c7cd5afb5719733dea5e3e8737a19c41e0956bfa4f4
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/3b7af9cb30f6f4569db5b2dc.json
true
scale800_v1/20261112/gate-evidence/3b7af9cb30f6f4569db5b2dc/report.json
false
false
true
scale800_v1/20261112/images/3b7af9cb30f6f4569db5b2dc.json
false
SelectBuilder.Where() doesn't handle empty string predicates correctly When passing an empty string to the `Where()` method, it should be treated the same as `nil` and ignored, but currently it adds an empty `WHERE` clause to the SQL output. #### Steps to Reproduce ```go package main import ( "fmt" "github....
go
false
scale800_v1/20261112/tasks/3b7af9cb30f6f4569db5b2dc/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
64730277ae09e2c630c55d513a4cb7c0822d018c
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/3b7af9cb30f6f4569db5b2dc.json
Masterminds__squirrel.1ded5784.func_pm_op_swap__oqebv5zg
52823f8feda3ba90356da2052bc9bb9bde76bf837dc2429c2d1583dea4079a93
scale800_v1/20261112/tasks/3b7af9cb30f6f4569db5b2dc
cc9808bb9f1a45888eede3951b00b3e00feb22fb5d4abd74b5b71264335ccc3a
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/035ded4cf9df8f7710194e95.json
true
scale800_v1/20261112/gate-evidence/035ded4cf9df8f7710194e95/report.json
false
false
true
scale800_v1/20261112/images/035ded4cf9df8f7710194e95.json
true
ExecWith doesn't handle ToSql errors properly When using ExecWith with a Sqlizer that returns an error from ToSql(), the error is ignored and passed through to db.Exec() with potentially invalid query/args. #### Steps/Code to Reproduce ```go package main import ( "database/sql" "fmt" "github.com/Masterm...
go
false
scale800_v1/20261112/tasks/035ded4cf9df8f7710194e95/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-v1
1
8
scored
8
1
0.675592
false
false
bac06a2b99748f24f84c84959d22cfb9ca285ce9
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/035ded4cf9df8f7710194e95.json
Masterminds__squirrel.1ded5784.func_pm_remove_cond__tkhou5al
ff2937c591edc6e67d8f349a0adea7c932deffca4dc17dcc056aa057419ef2d2
scale800_v1/20261112/tasks/035ded4cf9df8f7710194e95
4ab0d30fec047fa1a43a6ced09dfa97752fa0fc75c115f8ea5d852cb672c6f56
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/bbf9c3f598c6781a641d82b4.json
true
scale800_v1/20261112/gate-evidence/bbf9c3f598c6781a641d82b4/report.json
false
false
true
scale800_v1/20261112/images/bbf9c3f598c6781a641d82b4.json
false
Where clause arguments getting dropped when multiple arguments provided #### Description When creating where clauses with multiple arguments, the first argument is being dropped unexpectedly. This affects SQL generation where placeholder values don't match up with the generated SQL. #### Steps/Code to Reproduce ```...
go
false
scale800_v1/20261112/tasks/bbf9c3f598c6781a641d82b4/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
37d9ddc7ab57ab5d56868ada5ad1a30ad2f0246b
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/bbf9c3f598c6781a641d82b4.json
Masterminds__squirrel.1ded5784.lm_modify__0vqzytrm
672a9fc5539087a1a35555529bca0d5962fff4c3e1c5e820bdf9b58f3e53919c
scale800_v1/20261112/tasks/bbf9c3f598c6781a641d82b4
f909aafbcf885ab56521c9fbe99116b8731a310423a9cce598eafb7a826cfb27
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/051c64ba88ae502156a11b2f.json
true
scale800_v1/20261112/gate-evidence/051c64ba88ae502156a11b2f/report.json
false
false
true
scale800_v1/20261112/images/051c64ba88ae502156a11b2f.json
false
DeleteBuilder QueryRowContext returns empty result when using context methods #### Description When using `QueryRowContext` on a `DeleteBuilder`, the method returns an empty result instead of the expected SQL query and parameters. This appears to be related to how the builder data is retrieved in the context method. ...
go
false
scale800_v1/20261112/tasks/051c64ba88ae502156a11b2f/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
d4bf9b8443ceaefeb700697352be54497b5f513a
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/051c64ba88ae502156a11b2f.json
Masterminds__squirrel.1ded5784.lm_modify__1qf54jqz
4fd47b384604e54c4b9f466e5218a4001e8ada3c9e1f176f31a5caed7c8070cc
scale800_v1/20261112/tasks/051c64ba88ae502156a11b2f
a0fa290444b175903157b75828e10889d94a35333013dbda61ad65db85170238
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/92c7bf4f1dc5b3195f3988c4.json
true
scale800_v1/20261112/gate-evidence/92c7bf4f1dc5b3195f3988c4/report.json
false
false
true
scale800_v1/20261112/images/92c7bf4f1dc5b3195f3988c4.json
false
Expression arguments not properly appended in ToSql method When using expressions with arguments, the arguments are not being properly included in the final SQL output. This causes issues when the expression contains placeholders that need to be filled with the provided arguments. #### Steps to Reproduce ```go packa...
go
false
scale800_v1/20261112/tasks/92c7bf4f1dc5b3195f3988c4/provenance.json
true
null
0.5
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.5
2
scored
4
0.849961
0.150039
false
false
02b8d7246ed0e3baae55806c73c091a00a6bf17d
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/92c7bf4f1dc5b3195f3988c4.json
Masterminds__squirrel.1ded5784.lm_modify__22kgn3n5
0b047a8962d9785f8d00e816b01c5b699482da2720d69112d9fd7870d9b474c1
scale800_v1/20261112/tasks/92c7bf4f1dc5b3195f3988c4
97de9cce43f51e064f26fd68e8ed7096ccd77e3adc3073e1932e6086a1131842
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/507df911abc96c1e03764826.json
true
scale800_v1/20261112/gate-evidence/507df911abc96c1e03764826/report.json
false
false
true
scale800_v1/20261112/images/507df911abc96c1e03764826.json
false
Update builder loses placeholder format when created from StatementBuilderType #### Description When creating an UpdateBuilder using `StatementBuilderType.Update()`, the placeholder format configuration is lost, causing a nil pointer dereference when trying to generate SQL. #### Steps/Code to Reproduce ```go packag...
go
false
scale800_v1/20261112/tasks/507df911abc96c1e03764826/provenance.json
true
null
null
qwen38-auto48-64k-2m-nothink-r3-a4-v1
null
null
scoring
null
null
null
false
false
dd3d89009f1ba50c674565e979ee10612bb5c27e
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/507df911abc96c1e03764826.json
Masterminds__squirrel.1ded5784.lm_modify__2byob5te
f79303a6c4e96baa64d377f8a9065efe188562dffd29c6923df5be67f3a1be89
scale800_v1/20261112/tasks/507df911abc96c1e03764826
ef648c85b325f9d322a21a5f0611bf7aae2b46af35e943fcc40b74c46b280aba
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/dc833de1d068d820dd8d8270.json
true
scale800_v1/20261112/gate-evidence/dc833de1d068d820dd8d8270/report.json
false
false
true
scale800_v1/20261112/images/dc833de1d068d820dd8d8270.json
true
SQL query generation adds unwanted separator at beginning #### Description When building SQL queries, an extra separator (comma or "AND") is being added at the beginning of clauses, resulting in malformed SQL. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.com/Masterminds/squirrel" ...
go
false
scale800_v1/20261112/tasks/dc833de1d068d820dd8d8270/provenance.json
true
null
1
qwen38-auto48-64k-2m-nothink-r3-v1
0
0
scored
8
0.324408
0
false
false
94b054e3ddf090d08944b0a0b1d4baf8d5aee0fd
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/dc833de1d068d820dd8d8270.json
Masterminds__squirrel.1ded5784.lm_modify__2c1twnc6
d45ca047ba56dfc1ff24cacfc52d4535bf81deb3cd71df0d98e5ac2ec11ecc34
scale800_v1/20261112/tasks/dc833de1d068d820dd8d8270
c57c870e633e1ee8ef969e54b8e8245760fd246517dc82f4f657d06da15fd3de
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/7032a495b6dfdd4f852483ee.json
true
scale800_v1/20261112/gate-evidence/7032a495b6dfdd4f852483ee/report.json
false
false
true
scale800_v1/20261112/images/7032a495b6dfdd4f852483ee.json
true
Statement cache not working correctly with PrepareContext #### Description The statement cache in `StmtCache.PrepareContext` is not functioning as expected. When preparing the same query multiple times, it should reuse the cached prepared statement, but instead it's creating new prepared statements each time. #### S...
go
false
scale800_v1/20261112/tasks/7032a495b6dfdd4f852483ee/provenance.json
true
null
null
qwen38-auto48-64k-2m-nothink-r3-v1
null
null
scoring
null
null
null
false
false
068d1fdaaae529dee496047f1b420942883e2d4e
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/7032a495b6dfdd4f852483ee.json
Masterminds__squirrel.1ded5784.lm_modify__2qc5gboi
4641de58ced37c2623bb0b1c0e4184daf6e23a94e40e0de2cbfe31470912dcd4
scale800_v1/20261112/tasks/7032a495b6dfdd4f852483ee
f089367db2a3c4a18bc74ba948de3504c6d224bc3572eabcb58069bf6784ac85
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/59c483ae9a4fcb3127e2afbf.json
true
scale800_v1/20261112/gate-evidence/59c483ae9a4fcb3127e2afbf/report.json
false
false
true
scale800_v1/20261112/images/59c483ae9a4fcb3127e2afbf.json
true
Expr function reverses argument order unexpectedly #### Description The `Expr` function is reversing the order of arguments passed to it, causing SQL placeholders to be bound to the wrong values. #### Steps/Code to Reproduce ```go package main import ( "fmt" "github.com/Masterminds/squirrel" ) func main()...
go
false
scale800_v1/20261112/tasks/59c483ae9a4fcb3127e2afbf/provenance.json
true
null
null
qwen38-auto48-64k-2m-nothink-r3-v1
null
null
scoring
null
null
null
false
false
299dd016c8e917638c10699df2771f7fc36ae5ae
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/59c483ae9a4fcb3127e2afbf.json
Masterminds__squirrel.1ded5784.lm_modify__39gxbrjn
c27761fad92e1b4ddaeb590db0801096899aec7f14516a483ce8acc48b99f4a2
scale800_v1/20261112/tasks/59c483ae9a4fcb3127e2afbf
e6afe9db6bf0b4205b9e770c290ddd5f217cf9358bb28b6299d31da6333cb3d3
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/f23005cea993e49b33bc22d8.json
true
scale800_v1/20261112/gate-evidence/f23005cea993e49b33bc22d8/report.json
false
false
true
scale800_v1/20261112/images/f23005cea993e49b33bc22d8.json
false
Gt comparison operator generates incorrect SQL #### Description The `Gt` (greater than) comparison operator is generating incorrect SQL output. When using `Gt` to create a "greater than" condition, it produces "less than or equal to" (`<=`) SQL instead of the expected "greater than" (`>`) SQL. #### Steps/Code to Rep...
go
false
scale800_v1/20261112/tasks/f23005cea993e49b33bc22d8/provenance.json
true
null
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
64108f503eb397d43ddaf0c80053d5871c63cd5c
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/f23005cea993e49b33bc22d8.json
Masterminds__squirrel.1ded5784.lm_modify__5gl22gfp
6f945241760131c46ffac3adc4d9a5d57aaedb281288218317ea1a1d0de48076
scale800_v1/20261112/tasks/f23005cea993e49b33bc22d8
bb6ed8120dae38a518114da15ff854907c0d01947fb7bf212298f3079e1d4cdb
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/7896a61bebda22964e0bbf78.json
true
scale800_v1/20261112/gate-evidence/7896a61bebda22964e0bbf78/report.json
false
false
true
scale800_v1/20261112/images/7896a61bebda22964e0bbf78.json
false
SQL conjunction expressions missing parentheses When building SQL queries with multiple conditions using AND/OR operators, the generated SQL is missing parentheses around the conjunction expressions. This can lead to incorrect query logic due to operator precedence issues. #### Steps/Code to Reproduce ```go package ...
go
false
scale800_v1/20261112/tasks/7896a61bebda22964e0bbf78/provenance.json
true
null
0.5
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.5
2
scored
4
0.849961
0.150039
false
false
7502390dc0c0afac8cdde9e8e8324e2ed7c0fc9b
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/7896a61bebda22964e0bbf78.json
Masterminds__squirrel.1ded5784.lm_modify__60oyywsi
1f692d8613709864eef0fda5f954dc47edf1bc0e9c93d58cfd9f8a60fa085659
scale800_v1/20261112/tasks/7896a61bebda22964e0bbf78
808a9059c2d62b756eef4b16410864996434d181d33eaf0f7c1d3c384ca6d62a
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
null
null
null
null
null
null
null
true
null
unmeasured
false
scale800_v1/20261112/execution/ad7880aa1a119092b965313a.json
true
scale800_v1/20261112/gate-evidence/ad7880aa1a119092b965313a/report.json
false
false
true
scale800_v1/20261112/images/ad7880aa1a119092b965313a.json
false
Update builder panics with nil pointer dereference when no runner is set When calling `Exec()` on an update builder without setting a runner, the code now creates an empty `sql.DB{}` instance instead of returning the expected `RunnerNotSet` error. This causes a nil pointer dereference panic when the empty DB tries to ...
go
false
scale800_v1/20261112/tasks/ad7880aa1a119092b965313a/provenance.json
true
null
null
qwen38-auto48-64k-2m-nothink-r3-a4-v1
null
null
scoring
null
null
null
false
false
752988b934f185386d97043097c46f2d171b40dc
SWE-bench/SWE-smith-go
MIT
null
swesmith/Masterminds__squirrel.1ded5784
d2f6ecc2cf8ae1af00fc790801bbacfa2913e015
train
Qwen/Qwen3.8-27B
false
null
scale800_v1/20261112/inventories/ad7880aa1a119092b965313a.json
Masterminds__squirrel.1ded5784.lm_modify__6pvf4uk8
1b96010876a12788e97cdadbb8983841ffd68ef7aae0ab6f12ca6030f333c9ff
scale800_v1/20261112/tasks/ad7880aa1a119092b965313a
c7c4a719cdd65610da3208e529317012eb86414dac1d5505e608309919738916
null
false
false
masterminds/squirrel
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
3727a1e6f9314e0e35a60ba672da5117d39ac44587a8472d8fb13c9ebfaaf7d6
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/3949ddfba4ac56ce4f54635a.json
true
scale800_py_v1/20261112/gate-evidence/3949ddfba4ac56ce4f54635a/report.json
false
false
true
scale800_py_v1/20261112/images/3949ddfba4ac56ce4f54635a.json
false
Cache decorator returns timestamp instead of cached value #### Description The `@cache` decorator is returning timestamps instead of the actual cached values when retrieving items from the cache. #### Steps/Code to Reproduce ```python from funcy import cache @cache(timeout=60) def add_one(x): return x + 1 res...
python
false
scale800_py_v1/20261112/provenance/3949ddfba4ac56ce4f54635a.json
true
false
0.75
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.25
1
scored
4
0.699358
0.045587
false
false
76d77689e15abf517fc8139560ff28761a1f6b91
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/3949ddfba4ac56ce4f54635a.json
Suor__funcy.207a7810.combine_file__186umsl2
e094fbd39285be8a6b3a8c9bf3ddfe018b5d267f0d99378ad07b85e7c4e54db3
scale800_py_v1/20261112/tasks/3949ddfba4ac56ce4f54635a
22f1575c18148371d751032cebbdb8aa117677d387669dc8b82c7846140d4343
scale800_py_v1/20261112/terminal-proof/3949ddfba4ac56ce4f54635a.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
941e5280837831ed5d8ae10b0d809cd8b2a6f72772bbb0146ea9c4028c84ca38
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/26911a7ffcb4e1663c6eb417.json
true
scale800_py_v1/20261112/gate-evidence/26911a7ffcb4e1663c6eb417/report.json
false
false
true
scale800_py_v1/20261112/images/26911a7ffcb4e1663c6eb417.json
false
Decorator argument order swapped causing TypeError When using decorators with both positional and keyword arguments, the arguments are being passed in the wrong order to `make_decorator`, causing `TypeError` exceptions. #### Steps/Code to Reproduce ```python from funcy import decorator @decorator def my_decorator(f...
python
false
scale800_py_v1/20261112/provenance/26911a7ffcb4e1663c6eb417.json
true
false
0.75
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.25
1
scored
4
0.699358
0.045587
false
false
4837ea08a0a06fb6dcb5e71edce20c7a32b63c36
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/26911a7ffcb4e1663c6eb417.json
Suor__funcy.207a7810.combine_file__1vbqy9bz
baad414cd7369b2871dff1b8babebbe3c40376efc079761b45eb001bbb4e2957
scale800_py_v1/20261112/tasks/26911a7ffcb4e1663c6eb417
3c0a06bae5a8abb8132faf9401d8bf9d181c1f6186356f3c91b4e1313782000a
scale800_py_v1/20261112/terminal-proof/26911a7ffcb4e1663c6eb417.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
d4c6537d6aa8ede0ba9cb2dd1a755140b58ad532aacc3da6b273dd7f8186dd20
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/9922e787c0d95273b408dd1f.json
true
scale800_py_v1/20261112/gate-evidence/9922e787c0d95273b408dd1f/report.json
false
false
true
scale800_py_v1/20261112/images/9922e787c0d95273b408dd1f.json
false
Debug decorators broken after recent changes #### Description The debug decorators in `funcy.debug` are not working correctly. When using `log_calls`, `log_errors`, `print_errors`, and `log_durations`, the output format is completely wrong and some functionality is broken. #### Steps/Code to Reproduce ```python fro...
python
false
scale800_py_v1/20261112/provenance/9922e787c0d95273b408dd1f.json
true
false
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
e108dfb4e1b0f9e9fee9cf49fc28f024a3031d3c
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/9922e787c0d95273b408dd1f.json
Suor__funcy.207a7810.combine_file__3cucpzij
affe04341e95d369a507de1b86db4cecf7b0a785173e442d60c632c5707b44c0
scale800_py_v1/20261112/tasks/9922e787c0d95273b408dd1f
560f94727f5233ae3f3a446bea1f565c8427f1dc778e1df910e5b07a6480d8d3
scale800_py_v1/20261112/terminal-proof/9922e787c0d95273b408dd1f.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
40b95248a443173abc15e0f43729af443fcb850a56afaadc797192c87de9170b
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/939ef5a89514c4fad0370885.json
true
scale800_py_v1/20261112/gate-evidence/939ef5a89514c4fad0370885/report.json
false
false
true
scale800_py_v1/20261112/images/939ef5a89514c4fad0370885.json
false
# Broken curry and compose functions in funcy ## Description I've discovered that the `curry` and `compose` functions in funcy are not working correctly after a recent update. ## How to reproduce Here's a simple script that demonstrates the issue: ```python from funcy import curry, compose # Curry example def add...
python
false
scale800_py_v1/20261112/provenance/939ef5a89514c4fad0370885.json
true
false
null
qwen38-auto48-64k-2m-nothink-r3-a4-v1
null
null
incomplete
null
null
null
false
false
9554a1c03292d0ef23c6df01d22999f6b85a946e
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/939ef5a89514c4fad0370885.json
Suor__funcy.207a7810.combine_file__3u9hti2d
401b78f7ced2f72fc20079fa6734b27bd3c2ece913ce3b9abb50eaeeb418d2d6
scale800_py_v1/20261112/tasks/939ef5a89514c4fad0370885
b9bd3687d14135e3e45d61e4cadabd5e8a95e8a9f8c69e2391139836bd8bf6e7
scale800_py_v1/20261112/terminal-proof/939ef5a89514c4fad0370885.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
6f046b3d09556f1ad270e0da6e5bc82c94d782d62b5740a0c7b7d8efbc67d038
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/c21a55263e586618fa102531.json
true
scale800_py_v1/20261112/gate-evidence/c21a55263e586618fa102531/report.json
false
false
true
scale800_py_v1/20261112/images/c21a55263e586618fa102531.json
true
Lookuper functions broken after recent changes Description The `make_lookuper` function is not working correctly. When trying to use a lookuper function with arguments, it's passing the arguments as a tuple instead of unpacking them, and the lookup behavior has changed unexpectedly. Steps/Code to Reproduce ```pytho...
python
false
scale800_py_v1/20261112/provenance/c21a55263e586618fa102531.json
true
false
0
qwen38-auto48-64k-2m-nothink-r3-v1
1
8
scored
8
1
0.675592
false
false
c129dbb1ced9d23712b8458475c5bc1a80f0c303
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/c21a55263e586618fa102531.json
Suor__funcy.207a7810.combine_file__3y0j7te5
f2f942770d0cae5c8ecb82eaaac9d649bc212ede66a906056d1d1e83a7b5a3a8
scale800_py_v1/20261112/tasks/c21a55263e586618fa102531
f7094cc8d98f1f3a3b8d2f66e62e83d41a772f31976771abcc44a17fe7753d8c
scale800_py_v1/20261112/terminal-proof/c21a55263e586618fa102531.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
63b49a614659e6bb2421fd146a9822d285286dff12d255f298e1cf6c4cb1c1ff
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/30d48b405b3bd14fc468767b.json
true
scale800_py_v1/20261112/gate-evidence/30d48b405b3bd14fc468767b/report.json
false
false
true
scale800_py_v1/20261112/images/30d48b405b3bd14fc468767b.json
false
# Bug: Decorator argument access is broken in funcy **Describe the bug** I've discovered an issue with the decorator functionality in funcy. When trying to access arguments in decorated functions, I'm getting unexpected behavior. The decorator seems to be passing arguments in the wrong order or returning None when it ...
python
false
scale800_py_v1/20261112/provenance/30d48b405b3bd14fc468767b.json
true
false
0.25
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.75
3
scored
4
0.954413
0.300642
false
false
ae014f971c04672216999c7140483bef6c2c5332
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/30d48b405b3bd14fc468767b.json
Suor__funcy.207a7810.combine_file__4ho5rovv
143039ac07d8c95c9aec55102313f53d3c2245de4fcc469c969fc6e5560c73fe
scale800_py_v1/20261112/tasks/30d48b405b3bd14fc468767b
f7ea70f395280d7e131d520c1e4d677da424ba3452a200f6cdad0fa56245817f
scale800_py_v1/20261112/terminal-proof/30d48b405b3bd14fc468767b.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
e2193d98943bc11470bc0d5d9e81d22596b7307333a4558204edcffe5a605b6f
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/668b03ef0019cea5bdb3c757.json
true
scale800_py_v1/20261112/gate-evidence/668b03ef0019cea5bdb3c757/report.json
false
false
true
scale800_py_v1/20261112/images/668b03ef0019cea5bdb3c757.json
false
# Multiple issues with function composition and currying Hi, I've found several issues with the latest changes to the `funcs.py` module. The problems affect multiple functions including `identity`, `curry`, `compose`, and `rcompose`. ## Current Behaviour When using the `identity` function, it now returns the negati...
python
false
scale800_py_v1/20261112/provenance/668b03ef0019cea5bdb3c757.json
true
false
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
6821b20e1d4710035278b5de12f7b16d99bbe727
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/668b03ef0019cea5bdb3c757.json
Suor__funcy.207a7810.combine_file__5c2gq3ju
e0f77d1dbc753c4255fe8ea49a6444020b0ccface06fe5f4e7428e0f112bd90f
scale800_py_v1/20261112/tasks/668b03ef0019cea5bdb3c757
d9c21793b960c96c3c0093e14917d406aafb0513799c1df78820920b25888442
scale800_py_v1/20261112/terminal-proof/668b03ef0019cea5bdb3c757.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
0a6865c0642f3f7a31306a6875a850dd3a38d760d060da5a1268962aadbc748e
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/35395ababd3193ef05ed8b52.json
true
scale800_py_v1/20261112/gate-evidence/35395ababd3193ef05ed8b52/report.json
false
false
true
scale800_py_v1/20261112/images/35395ababd3193ef05ed8b52.json
false
re_all returns matches in reversed order #### Description The `re_all` function is returning regex matches in reversed order instead of the expected order they appear in the string. #### Steps/Code to Reproduce ```python from funcy import re_all # Simple digit matching result = re_all(r'\d+', 'x34y12') print("Expe...
python
false
scale800_py_v1/20261112/provenance/35395ababd3193ef05ed8b52.json
true
false
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
cecb12f3b616a3997fd8b355ea34dadbac298670
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/35395ababd3193ef05ed8b52.json
Suor__funcy.207a7810.combine_file__5kzv0kej
2d985a206f80f203bd3eb309f7ede8c37db97d79e205424ee82ae00658a62e69
scale800_py_v1/20261112/tasks/35395ababd3193ef05ed8b52
c1e398f6088ed8c55cf9a8c044a60ec79a902d213a10cc4138720be775b7677c
scale800_py_v1/20261112/terminal-proof/35395ababd3193ef05ed8b52.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
b57875bc4d93796314ec7a11509ac1af2835e59574623c0f5dde7c7ebeb900a0
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/b856fc7a2ddae40c89919abc.json
true
scale800_py_v1/20261112/gate-evidence/b856fc7a2ddae40c89919abc/report.json
false
false
true
scale800_py_v1/20261112/images/b856fc7a2ddae40c89919abc.json
false
# Multiple bugs in flow control decorators I've found several issues with the flow control decorators in the latest version. The problems affect `limit_error_rate` and `once_per` functionality. ## Issue 1: `limit_error_rate` decorator has inverted timeout logic When using the `limit_error_rate` decorator, it's not b...
python
false
scale800_py_v1/20261112/provenance/b856fc7a2ddae40c89919abc.json
true
false
null
null
null
null
not_started
null
null
null
false
false
85fab930615fb0bc8aeaca9b0ace10890d843339
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/b856fc7a2ddae40c89919abc.json
Suor__funcy.207a7810.combine_file__5pn9f8oq
d4ee9797f731d81ecc75ae561ce0c7269041fe3354430d04c80b1ddd6751b4f5
scale800_py_v1/20261112/tasks/b856fc7a2ddae40c89919abc
ec971fd7fd65026c2d282d11066570555b83316d901462405271b560d5cad10d
scale800_py_v1/20261112/terminal-proof/b856fc7a2ddae40c89919abc.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
9988c02ce55452586450c465c68bc31a91c2033117b04b4c8caddf1e0140c404
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/9a8b784bc52ceaddef028e5c.json
true
scale800_py_v1/20261112/gate-evidence/9a8b784bc52ceaddef028e5c/report.json
false
false
true
scale800_py_v1/20261112/images/9a8b784bc52ceaddef028e5c.json
true
cached_property returns None instead of the computed value Description I found an issue with the `cached_property` decorator in funcy. When using it, the property always returns `None` instead of the actual computed value. Here's a simple reproduction: ```python from funcy import cached_property class A(object): ...
python
false
scale800_py_v1/20261112/provenance/9a8b784bc52ceaddef028e5c.json
true
false
0.125
qwen38-auto48-64k-2m-nothink-r3-v1
0.875
7
scored
8
0.977583
0.529112
false
false
b8a0d70cec592067728bf9197da97fb80ba265f8
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/9a8b784bc52ceaddef028e5c.json
Suor__funcy.207a7810.combine_file__71pmnmxj
788eaa994fc0c65ed99e584d5f09d3bcafa27a420bc637037e7e9a053776df16
scale800_py_v1/20261112/tasks/9a8b784bc52ceaddef028e5c
8824c380ffdc58eb89e024a3dc4e262765a5878f9cf5e6c6da6a91d03a1ed206
scale800_py_v1/20261112/terminal-proof/9a8b784bc52ceaddef028e5c.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
1dfd4e1c3b09980be212f2b4947d6a4573089c6b0646793e6947062b9102a897
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/b3d966b28f645967f9865b79.json
true
scale800_py_v1/20261112/gate-evidence/b3d966b28f645967f9865b79/report.json
false
false
true
scale800_py_v1/20261112/images/b3d966b28f645967f9865b79.json
false
# Multiple issues with collection manipulation functions I've discovered several bugs in the collection manipulation functions that are causing unexpected behavior. Here are the issues I'm encountering: ## Issue 1: `walk_keys` function applies transformation to values instead of keys When using the `walk_keys` funct...
python
false
scale800_py_v1/20261112/provenance/b3d966b28f645967f9865b79.json
true
false
1
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0
0
scored
4
0.489891
0
false
false
7cd62410999b8f695a0a60a3087402786bde2e72
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/b3d966b28f645967f9865b79.json
Suor__funcy.207a7810.combine_file__724bnxkn
c953c8131d53de3f5000e9cd95b948c1b0bd80266ec6568a2912476e97bc112a
scale800_py_v1/20261112/tasks/b3d966b28f645967f9865b79
203a96aa82e4a2397cbf1aa158232407d425cfdc1d7de48465a6ee8626c50104
scale800_py_v1/20261112/terminal-proof/b3d966b28f645967f9865b79.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
7899c699f152ef3108dafb2a149e39e5cb2da0111d0f03926ce90fa2d7266256
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/c6452146ba578d43089abdb0.json
true
scale800_py_v1/20261112/gate-evidence/c6452146ba578d43089abdb0/report.json
false
false
true
scale800_py_v1/20261112/images/c6452146ba578d43089abdb0.json
false
# tap() function in debug.py is broken and returns None instead of the input value ## Description The `tap()` function in debug.py is not working as expected. It's supposed to print a value and return it, but it's returning `None` instead and the printing functionality is also broken. ## Steps to Reproduce ```pytho...
python
false
scale800_py_v1/20261112/provenance/c6452146ba578d43089abdb0.json
true
false
0.25
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.75
3
scored
4
0.954413
0.300642
false
false
b2efbbf4814950cb6b8bc29ec5bd22fd2ff31ccc
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/c6452146ba578d43089abdb0.json
Suor__funcy.207a7810.combine_file__81deq3s8
aabaeed8965b1270f721e4ea8e2e14aa316cf3b70440ff94fddd66ac6133b16f
scale800_py_v1/20261112/tasks/c6452146ba578d43089abdb0
c5c6cef28a61688e641f555467e6f9c936ca8b0f0b432d0d2853750abaf1b8cb
scale800_py_v1/20261112/terminal-proof/c6452146ba578d43089abdb0.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
5969d218921232f4bced236dd1b06f604f588335f58557915a0f2bcc279f946b
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/d67c204a619693dae2a4a5d6.json
true
scale800_py_v1/20261112/gate-evidence/d67c204a619693dae2a4a5d6/report.json
false
false
true
scale800_py_v1/20261112/images/d67c204a619693dae2a4a5d6.json
false
# Multiple issues with collection manipulation functions I've discovered several bugs in the collection manipulation functions that are causing unexpected behavior. ## Issue 1: `walk_keys` function is broken When trying to use the `walk_keys` function to transform dictionary keys, I'm getting a TypeError about unhas...
python
false
scale800_py_v1/20261112/provenance/d67c204a619693dae2a4a5d6.json
true
false
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
91e30b91a842da3115990068b3309a66117a654e
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/d67c204a619693dae2a4a5d6.json
Suor__funcy.207a7810.combine_file__8y50osx1
f30c4af01c0571e0c471c50f6b58efb441132e7669e0442151ed57b659e52b77
scale800_py_v1/20261112/tasks/d67c204a619693dae2a4a5d6
af292b882381012e0e0887c45bfe3c1ea7505adb354aa4cc99f6eae77f41352a
scale800_py_v1/20261112/terminal-proof/d67c204a619693dae2a4a5d6.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
8bcf6cf5cee424f46a7c1095bac43d9f3f2241b05a7f3bbc2d583b28b31641bc
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/633a5c12c342e7d74de80591.json
true
scale800_py_v1/20261112/gate-evidence/633a5c12c342e7d74de80591/report.json
false
false
true
scale800_py_v1/20261112/images/633a5c12c342e7d74de80591.json
false
String regex functions broken after recent changes #### Description The string regex functions `re_find`, `re_test`, and `cut_suffix` are not working correctly. When using regex patterns with strings, they're returning unexpected results or None instead of the expected matches. #### Steps/Code to Reproduce ```pytho...
python
false
scale800_py_v1/20261112/provenance/633a5c12c342e7d74de80591.json
true
false
0
qwen38-auto48-64k-2m-nothink-r3-a4-v1
1
4
scored
4
1
0.510109
false
false
b88f0de344beda08fbe03402343ca7a6ce62ba31
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/633a5c12c342e7d74de80591.json
Suor__funcy.207a7810.combine_file__9xbr3kzj
7680b438904d5119ea5419c4f19adb4744dc9c7debaf7354a97114bda5849dba
scale800_py_v1/20261112/tasks/633a5c12c342e7d74de80591
be474e2d50627c95df085b0cfbf1c82a2742c84eec6a442f94027503c90c7915
scale800_py_v1/20261112/terminal-proof/633a5c12c342e7d74de80591.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
9b4b01bc03ecd941aeff968a2a0e51b58f85780e64e7f56ce26f1444b19b0fb8
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/42063e73927b1ac951a6daef.json
true
scale800_py_v1/20261112/gate-evidence/42063e73927b1ac951a6daef/report.json
false
false
true
scale800_py_v1/20261112/images/42063e73927b1ac951a6daef.json
false
walk_keys function applies transformation to values instead of keys #### Description The `walk_keys` function is incorrectly applying the transformation function to values instead of keys, and also truncating the collection by excluding the last element. #### Steps/Code to Reproduce ```python from funcy import walk...
python
false
scale800_py_v1/20261112/provenance/42063e73927b1ac951a6daef.json
true
false
1
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0
0
scored
4
0.489891
0
false
false
6f4f1d48a990626457937b2243acdf1b35c3d8b3
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/42063e73927b1ac951a6daef.json
Suor__funcy.207a7810.combine_file__b3usvi20
a6d096efb4b65cb8a213c86028582caffb72ae1b36dd4fe453295e4244d7f3e7
scale800_py_v1/20261112/tasks/42063e73927b1ac951a6daef
4c26a838520e6e02e5695e22667849d3dad9a9abb929d03bb5b1fe480fbb71f7
scale800_py_v1/20261112/terminal-proof/42063e73927b1ac951a6daef.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
3c7d624665770899ea334641dc79ba6e248e460006b07e3689c6d9860f467d22
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/e67416af0c3338bfe49e5c5c.json
true
scale800_py_v1/20261112/gate-evidence/e67416af0c3338bfe49e5c5c/report.json
false
false
true
scale800_py_v1/20261112/images/e67416af0c3338bfe49e5c5c.json
false
# Bug in all_fn and none_fn functions ## Description I've discovered an issue with the `all_fn` and `none_fn` functions in the funcolls module. These functions are not behaving as expected according to their documentation. According to the docstring, `all_fn` should construct "a predicate, which holds when all fs hol...
python
false
scale800_py_v1/20261112/provenance/e67416af0c3338bfe49e5c5c.json
true
false
0.25
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0.75
3
scored
4
0.954413
0.300642
false
false
c810c7340cd18257f55830b14912686c71072937
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/e67416af0c3338bfe49e5c5c.json
Suor__funcy.207a7810.combine_file__be6tat1i
967ec0b0d785017f07f4c5d6474dd2d76979bc0407bcdd92771bcd3ee3dabecb
scale800_py_v1/20261112/tasks/e67416af0c3338bfe49e5c5c
ba695192dbd8b4c8248d7cce348e16c9b1750eb1a77a4e6729ab8859a690fb7a
scale800_py_v1/20261112/terminal-proof/e67416af0c3338bfe49e5c5c.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
7df498742be2a493d5b270452c04ff46e98cd2499c77c6aa988379c7a0eea779
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/77ace3cd203fb06746f13a4e.json
true
scale800_py_v1/20261112/gate-evidence/77ace3cd203fb06746f13a4e/report.json
false
false
true
scale800_py_v1/20261112/images/77ace3cd203fb06746f13a4e.json
true
lmapcat function not working correctly with multiple sequences #### Description The `lmapcat` function appears to be broken when used with multiple sequences. Instead of mapping the function over each sequence separately and concatenating the results, it's passing the sequences as a single tuple argument to the funct...
python
false
scale800_py_v1/20261112/provenance/77ace3cd203fb06746f13a4e.json
true
false
0.625
qwen38-auto48-64k-2m-nothink-r3-v1
0.375
3
scored
8
0.694258
0.136844
false
false
c841dbb48cad068b2de65468c7c4dd7d53f9cb4d
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/77ace3cd203fb06746f13a4e.json
Suor__funcy.207a7810.combine_file__bxevtrf7
9964ffc69225ef87d04946f325b23a069b7ee05f6753ae3e3fe9248f1bdd73d0
scale800_py_v1/20261112/tasks/77ace3cd203fb06746f13a4e
eaf02ae33543ad573e216bb8f7401b5c8ea80ec49f3c0ff16f54eaf5d5d47ad6
scale800_py_v1/20261112/terminal-proof/77ace3cd203fb06746f13a4e.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/backend.py
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b
python-git-patch-v1
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/module-sha256.json
openenv-python-hf-jobs-semantic-plan-v2
scale800_py_v1/20261112/implementations/a2f8064fdc1af59c548eb41b/repo/outputs/full-hub-run-20261110/scale800-pilot-v1/production/hf-check-gate-python-v2/restore.py
92640b89215bd474eaeed92fde8b9d00dda2d2a96180528b0f7380d62eb90bda
false
71791663068c58a7f7937988b03a78992a4a47904002ed23cfeb0d85de7c4197
unmeasured
false
scale800_py_v1/20261112/execution/652bc992ecc5da4fbb880e98.json
true
scale800_py_v1/20261112/gate-evidence/652bc992ecc5da4fbb880e98/report.json
false
false
true
scale800_py_v1/20261112/images/652bc992ecc5da4fbb880e98.json
false
LabeledContextDecorator breaks when used with non-callable labels #### Description When using debug decorators like `log_errors` or `print_errors` with a non-callable label, the decorator fails to work properly. The logic for detecting callable vs non-callable labels appears to be inverted. #### Steps/Code to Reprod...
python
false
scale800_py_v1/20261112/provenance/652bc992ecc5da4fbb880e98.json
true
false
1
qwen38-auto48-64k-2m-nothink-r3-a4-v1
0
0
scored
4
0.489891
0
false
false
7492e339aff24d31b5d8272c223b81355bf3cd64
SWE-bench/SWE-smith-py
BSD-3-Clause
null
swesmith/Suor__funcy.207a7810
77cab9055d42ab4a5c25c89a8f937096db13558e
train
Qwen/Qwen3.8-27B
false
MIT
scale800_py_v1/20261112/inventories/652bc992ecc5da4fbb880e98.json
Suor__funcy.207a7810.combine_file__dsudit38
29121a4958140d6f3172823cb75453d395ca41f32cefb1b95b0a3a46645a6deb
scale800_py_v1/20261112/tasks/652bc992ecc5da4fbb880e98
c2a382d0948c784ee8410298ccfb52510f4e9608b72818c05e2a378aa7bf2311
scale800_py_v1/20261112/terminal-proof/652bc992ecc5da4fbb880e98.json
false
false
suor/funcy
quarantine_only
harbor-semantic-jobs-v1
false
End of preview. Expand in Data Studio

Agentic coding environments

This private repository preserves the original write-42 pipeline pilot and adds six original coding environments in the separate synthetic_gap_v1 split. See the split section below for controls, benchmark coverage and calibration status.

Original pipeline pilot

This snapshot contains one synthetic smoke-test task, not a corpus of real coding tasks. It asks an agent to write 42 to a file. It tests the OpenEnv/Harbor pipeline and has no demonstrated training value.

The real Docker pilot accepted one control and excluded two bad controls. The accepted task scored NOP=0 and oracle=1. Exact original validation evidence, including excluded reports, is retained under evidence/ with checksums. Only the accepted task is under tasks/.

Profile: harbor-oracle-floor-v1. Reports remain WARN because other checks were skipped. Full semantic, security, reward-hacking and trainability qualification has not been established. The original snapshot preceded model evaluation; subsequent live audit results are recorded below. No RL training has been performed. See publication.json for source hashes and the exact implementation commit. generate_controls.py preserves the generator; reproduce through the original example module in an OpenEnv checkout.

Container images are not included. Historical local sha256: image IDs in the evidence cannot be pulled from a registry. Task files retain the Dockerfile; a fresh build and validation are required after downloading this repository. Historical absolute paths remain only in the original evidence.

This snapshot is a synthetic pilot; future real Hub-sourced corpora will be identified separately.

Live HF Jobs coding audit — 2026-11-10

The supplemental harbor-coding-audit-v1 pilot completed successfully on HF Jobs, using Qwen/Qwen3.8-27B through the DeepInfra inference provider.

  • Two generated reference candidates and the provided reference each passed three fresh replays (9/9).
  • The no-op scored 0; all three trusted wrong-answer controls scored 0.
  • Reference, verifier rejection, reliability, bounded leakage probes, and target difficulty checks all passed.
  • Target difficulty: 8/8 independent attempts solved, zero infrastructure errors; solve-rate Wilson 95% interval 0.6756–1.0.
  • All 42 distinct solver/verifier Sandbox Jobs were independently confirmed terminated.
  • All 67 checksummed audit artifacts were downloaded and verified.

Detailed report · Raw evidence · HF Job

Run: coding-83614ebe192502f1249acef887ae0c44c4f87b725914ea7585fe31b59c1d6d3c. Implementation commit: 762c190448e7106b3c4f38de8f9192582c51642d. The exact source, task, configuration, and dependency lock are preserved in inputs/83614ebe192502f1249acef887ae0c44c4f87b725914ea7585fe31b59c1d6d3c/bundle.zip. The cloud audit uses prebuilt registry images pinned by OCI digest, separately from the historical local Docker image IDs in the original pilot evidence.

The first attempt remains preserved: seven successful target attempts and one startup infrastructure error caused its audit to withhold approval. The rerun used a tested, bounded control-plane readiness reconciliation and preserved all original admission gates.

This demonstrates end-to-end pipeline execution and the named supplemental checks on a synthetic control. Full RFC 008 qualification and trainability remain unestablished; the report retains rfc008_complete: false and trainability_established: false.

Original synthetic capability split — 2026-11-10

The named synthetic_gap_v1 split adds six original mini-repository tasks: atomic artifact publication, transactional SQLite migration, framed stream recovery, incremental build caching, layered CLI configuration and async retry budgets. These families are absent from the original admitted write-42 corpus. This is not a claim that they are absent from every unvalidated Hub task.

Split index · Task bundles · Reproduction and integration · Benchmark coverage and exact-version caveats.

Each task has public starter source, a hidden behavioral verifier, a reference, a valid alternative and three task-specific wrong implementations. Sixty controls passed on local Python 3.12 and sixty passed again in offline Linux/amd64 Docker containers: baseline and wrong implementations fail, both valid solutions pass three fresh replays each. Every baseline retains adjacent passing tests. Evidence binds each replay and packaged task to its source hash. Five focused packager regressions and an independent review cover evidence admission and valid-alternative fairness. Local Docker images use the pinned Python 3.12.16 registry base; hidden tests and solutions are excluded from each public build context.

The tasks are assistant-authored original material using Tasksmith-inspired controls, not a Tasksmith PR-conversion run or its verified label. Benchmark instances, patches, tests and rollouts were not imported; complete instance-level deduplication was not run. Treat these rows as training candidates and keep benchmark evaluation payloads separate.

Qwen3.8-27B difficulty remains unmeasured on this new split. No model calls, HF Jobs or RL training ran for this contribution. qwen38-gap48-v1 is a separate proposed 48-call/32,768-token/600-second protocol, not the pilot's protocol. Audit templates deliberately omit execution.agent_image: public-only starter images must be published under real registry digests before cloud preparation. The preparation script retains per-task source/config/plan hashes and image pins. Full RFC 008 qualification, hostile-code containment and training benefit remain unestablished. Existing pilot inputs, results and evidence are preserved.

Tasksmith-generated environments — unverified

Load config="tasksmith_gap_v1", split="unverified" for two constructed native environments retained for quarantine and further evaluation. Their generation provenance, per-source licenses, native bundles and original failure receipts are under tasksmith_gap_v1/20261110/. These records have not passed the separate quality or Qwen audits and have no qualified published image pins. They are not training-admissible; validation, calibration, nonoverlap and trainability flags remain false. This config is separate from the default synthetic split.

Cloud audit screening update - 2026-11-10

The six original synthetic tasks completed this cloud screening attempt. The provided references passed 18 replay trials and all 18 trusted negatives were rejected; 120 recorded child Jobs were independently terminal. Generated candidate scripts had 36 execution errors, so target evaluation was skipped and model difficulty remains unmeasured. This does not change the original task bytes, split rows, or qualification flags. Raw immutable reports remain retained; a bounded follow-up identified Bash response fences and added explicit public-source context. No RFC-completeness or trainability claim follows.

The screening summary is results/full-hub-20261110/synthetic-v3-screening-summary.json.

synthetic_gap_v2: original polyglot candidates

The synthetic_gap_v2 config exposes six originally authored candidate tasks in Go, Rust, C, SQL, TypeScript and JavaScript through its candidates split. Native task bytes, provenance and mode inventories are under synthetic_gap_v2/20261110/; the index is splits/synthetic_gap_v2.jsonl.

The publisher reports 132 passing local/Linux control executions: 66 host controls and 66 Linux amd64 controls. This card registration did not rerun those controls. Qwen target difficulty remains unmeasured. The candidates remain unverified and do not establish RFC 008 completeness, trainability, training admission, semantic novelty/nonoverlap, hostile-code isolation, image qualification or cloud readiness.

HTTP downloads preserve published file bytes, not verified POSIX execution modes. Consumers must restore modes from the recorded inventory before local execution; this registration makes no remote POSIX-mode verification claim.

tasksmith_gap_v2: quarantined native Tasksmith candidates

The tasksmith_gap_v2 config exposes 13 unique native Tasksmith PR-conversion tasks in its candidates split, indexed by splits/tasksmith_gap_v2.jsonl. Native tasks, provenance, failure lineage and recorded mode inventories are under tasksmith_gap_v2/20261110/. The immutable publication is ad8f9fd640a29204bbc8fa10efb27677bc8203aa.

The campaign used a 24-source pool and 28 attempts: 13 tasks were constructed, seven received native Tasksmith quality verification, six remain unverified, and 11 sources were blocked. Generation stopped at panel_exhausted before the target of 12 native-quality-verified tasks. These native labels do not establish broader qualification. All 13 rows retain verified: false and usage: quarantine_only. Qwen target calls were zero; calibration, RFC 008 completeness, trainability, training admission, semantic nonoverlap, complete benchmark coverage, published image qualification and cloud readiness remain unestablished/false. No training was run for this cohort.

The publication reports all 6,758 files byte-verified. Hub readback verifies file bytes, not POSIX execution modes; consumers must restore the recorded modes before execution. This card registration checks the immutable index, schema and manifest and does not repeat the full bundle download or native quality runs.

Recorded native accounting totals USD 53.922916 across both phases, comprising model usage plus estimated compute. This is not a provider invoice or a ledger settlement.

synthetic_gap_v3: original terminal capability candidates

The synthetic_gap_v3 config exposes four originally authored tasks through its candidates split: service-route-repair, git-release-bundle, audio-timeline-render and grouped-calibration-report. The index is splits/synthetic_gap_v3.jsonl; native tasks and provenance are under synthetic_gap_v3/20261110/. The immutable publication is 29db70c4a00d6f31eb6bf857cd8b8f456a722df0.

The publisher reports 45 passing Linux controls and 22 passing host controls. Host replays cover the audio and calibration tasks; service and Git host status is not_run. Control statuses and test-method counts are source-bound local claims. This card registration checks metadata and the dataset loader and does not rerun or independently verify those runtime controls.

All four rows retain usage: quarantine_only. All nine broader qualification flags remain false: verified, target_model_calibrated, rfc008_complete, trainability_established, training_admissible, semantic_nonoverlap_established, published_image_pins_qualified, cloud_launch_ready and hostile_code_isolation_established. The cohort manifest also retains complete_benchmark_coverage_established: false. Target-model calls were zero and Qwen difficulty remains unmeasured; no model, image, novelty, cloud, RFC or training qualification is established.

Hub byte readback does not verify POSIX execution modes. Consumers must restore modes from synthetic_gap_v3/20261110/native-file-modes.json before execution.

qwen38_coding_audit_v1: measured results and incomplete audit

The qwen38_coding_audit_v1 config exposes measured (6 complete measurements) and incomplete (1 record). Exact metadata and artifact hashes are under results/full-hub-20261110/calibration-snapshot-v1/. These are supplemental harbor-coding-audit-v1 results, not full RFC 008 or training qualification. Existing task snapshots, their historical labels, and all prior splits remain unchanged; immutable run/report references identify the selected measurement.

The measured rows have 8 valid, independently graded Qwen3.8-27B attempts each: atomic-artifact-publish 0/8, sqlite-event-migration 0/8, framed-stream-recovery 1/8, incremental-build-graph 0/8, layered-config-cli 0/8 and native-cancel-async-tasks 2/8. Wilson intervals and exact audit difficulty labels are copied from the reports. The incomplete async-retry-budget audit has 7 valid records, 0 solves among those records, and one HTTP 504; its failed request has no reward or grader ID. It is excluded from the measured split. No earlier attempts or retries are pooled.

Synthetic limits were 48 calls / 4096 response tokens / 32768 cumulative tokens, with 600-second execution timeout. Native limits were 16 / 2048 / 16384 with 900-second timeout. All valid attempts stopped at conservative input reservation or response limits after 2-5 completed turns. Results are conditional on these recorded protocols, not intrinsic task hardness. Compare the native and synthetic protocols separately. The native Terminal Bench source remains evaluation-only.

model_calibrated denotes complete measurements under that protocol only. verified, rfc008_complete, trainability_established, training_admissible, served_weights_attested and strong_hostile_code_isolation remain false on all rows. Strong verifier integrity and learning benefit are not established. Known selected child Jobs were independently terminal; the incomplete audit's missing grader ID leaves its broader cleanup qualification incomplete. Provider invoices and failed-request billed usage remain unknown; no actual-cost claim is made. No new compute or inference was performed to publish this snapshot.

The metadata builder and checksum inventory are retained alongside the rows. Report and evidence paths use ordinary Hugging Face repo/revision/path fields; source_summary_path is an operator-local relative locator bound by its hash. The execution pipeline is maintained in OpenEnv's examples/coding_audit and examples/hub_validation_pipeline; the immutable input bundles retain the actual producer sources and configuration. Provider-served weights are not commit-pinned.

tasksmith_gap_v3: quarantined native recovery candidates

The tasksmith_gap_v3 config exposes 14 unique native recovery rows through its candidates split, indexed by splits/tasksmith_gap_v3.jsonl. Native artifacts, source identities, mode inventories, failure lineage and accounting disclosures are retained under tasksmith_gap_v3/20261110/recovery/, published immutably at 733f7ee15c7b88dbe0d2cb0d7945b6c877b7c693.

These are 14 newly native-accepted recoveries: eight prepared-profile constructions and six reused V2 native task inputs. Together with seven disjoint prior native-verified source URLs they give 21 cumulative unique native acceptances within the original 24-source campaign, not 21 new tasks. Three NetworkX sources remain blocked. All 14 rows remain quarantine_only with every broader qualification and coverage flag false. Recovery target-model calls were zero; Qwen calibration, complete benchmark coverage, semantic nonoverlap, image attestation, cloud readiness, RFC completeness, trainability and training admission are unestablished. The existing calibration snapshot is separate.

Native Tasksmith acceptance is not a claim that every blind solver completed without exception. Cattrs PR #765 retained verifier reward 1 and native acceptance alongside AgentTimeoutError. final_blind_solver_completed_without_exception is false for that row and true for the other 13; the exception field preserves the timeout explicitly. All 14 owned Modal workers were independently queried terminal according to the retained terminal proof. This registration checks metadata and dataset loading; it does not rerun native quality checks or perform model, cloud or image work.

Recorded V3 accounting is USD 42.994439: model usage 27.086014 plus conservative compute estimate 15.908425. Cumulative V2+V3 accounting is USD 96.917355: model usage 50.633517 plus conservative compute estimate 46.283838. These are not provider invoices or complete actual-cost settlements. The separately reported partial model usage USD 0.7532611500000002 is incomplete and excluded from those native accounted totals. Its full USD 4 operation remains uncertain/reserved. The shared USD 600 expansion and original USD 100 holds remain reserved, with no settlement or new reservation from this registration.

The publication reports exact immutable byte readback for all 7,786 files and local mode restoration followed by 14 archived native-proof replays. Hub byte readback does not establish remote POSIX execution modes. Consumers must restore the recorded native file modes before execution. Prior sources, indices, calibration results, configurations and card content are preserved.

synthetic_gap_v4: original capability candidates

The synthetic_gap_v4 config exposes six manually authored original synthetic tasks through its candidates split: capacity-plan-artifact, go-expression-pipeline, native-cache-lifetime, pki-chain-artifacts, sqlite-query-budget and typescript-reactive-workbench. These are not Tasksmith-generated tasks. The index is splits/synthetic_gap_v4.jsonl, with native material and provenance under synthetic_gap_v4/20261111/. The immutable data publication is 822a2ab16d2d8b1e78305060968d202586cf8a7a.

The publisher reports expected outcomes across 71 Linux control trials and 48 host control trials, covering 125 test methods. Those trials include deliberately failing baselines and wrong-answer controls as well as legitimate implementations; they are not 119 positive solutions. Host full panels cover four tasks; native-cache-lifetime and pki-chain-artifacts have host status not_run_full_panel. linux_controls_passed: true records the expected Linux control outcomes. Registration verifies immutable metadata and the dataset loader; it does not rerun these controls. Native C evidence uses ASan/UBSan with detect_leaks=0 and bounded linked-allocation tracking; general leak freedom is unestablished. PKI keys are deliberately public synthetic fixtures, not production credentials.

All six rows retain usage: quarantine_only. All ten broader qualification flags remain false: verified, target_model_calibrated, rfc008_complete, trainability_established, training_admissible, semantic_nonoverlap_established, complete_benchmark_coverage_established, published_image_pins_qualified, cloud_launch_ready and hostile_code_isolation_established. Qwen difficulty remains unmeasured; model, image, cloud, RFC, coverage, novelty and training qualification are unestablished.

Hub byte readback does not verify POSIX execution modes. Consumers must restore modes from synthetic_gap_v4/20261111/native-file-modes.json before execution.

synthetic_gap_v5: original raster and finite-chain candidates

The synthetic_gap_v5 config exposes two manually authored original tasks through its candidates split: raster-track-association and chain-posterior-inference. They are not Tasksmith-generated. The index is splits/synthetic_gap_v5.jsonl; native tasks and provenance are under synthetic_gap_v5/20261111/. The immutable data publication is 4be6e6dec8a258637ce23f6b92271f58f7794bb7.

The publisher reports expected outcomes across 28 host and 28 Linux control trials covering 45 test methods. These include expected negative-control rejections as well as legitimate implementations. linux_controls_passed: true records those expected outcomes. Registration verifies pinned metadata and the actual dataset loader; it does not rerun the controls.

Both rows retain usage: quarantine_only. All ten broader qualification flags remain false: verified, target_model_calibrated, rfc008_complete, trainability_established, training_admissible, semantic_nonoverlap_established, complete_benchmark_coverage_established, published_image_pins_qualified, cloud_launch_ready and hostile_code_isolation_established. Target-model calls were zero and Qwen difficulty remains unmeasured. Model calibration, cloud launch, published-image, full RFC008, training, benchmark coverage and novelty qualification remain unestablished.

The raster task does not establish encoded-video, OCR or vision-model equivalence. The finite-chain posterior task does not establish general Bayesian network or sampling equivalence.

HTTP byte verification does not attest POSIX execution modes. Consumers must restore recorded modes from synthetic_gap_v5/20261111/native-file-modes.json before execution.

tasksmith_gap_v4: quarantined native repository candidates

The tasksmith_gap_v4 config exposes three unique native Tasksmith rows through its candidates split: Werkzeug PR #3335, pytest PR #14372 and SQLGlot PR #8722. The index is splits/tasksmith_gap_v4.jsonl; native artifacts, source identities, failure and repair lineage, mode inventories and accounting disclosures are under tasksmith_gap_v4/20261111/cohort/. The immutable data publication is 4ec2381e17a1d4e8103d3ac1c1d6746d06c8d6a8.

All three rows have tasksmith_verified: true and a final blind trial that completed without exception. The final blind rewards are 0 for Werkzeug, 1 for pytest and 0 for SQLGlot. These are author/quality-model observations, not Qwen measurements or a claim that every blind solver solved its task. Two native task inputs were reused exactly from the initial failed V4 attempt; pytest was a fresh prepared-profile construction. Those initial failures and recovery evidence are retained. The three newly accepted source URLs are disjoint from the 21 prior native-accepted source URLs in the V2/V3 campaign, giving 24 cumulative unique native acceptances, not 24 new tasks. The initial pilot split is separate.

Every row remains usage: quarantine_only. All twelve broader flags stay false: verified, target_model_calibrated, rfc008_complete, trainability_established, training_admissible, semantic_nonoverlap_established, benchmark_holdout_deduplication_established, complete_benchmark_coverage_established, published_image_pins_qualified, image_source_attestation, cloud_launch_ready and hostile_code_isolation_established. Qwen difficulty remains unmeasured; semantic holdout nonoverlap, complete coverage, image/cloud qualification, full RFC compliance and training benefit or admission are unestablished.

The publisher reports all six V4 workers independently terminal. This registration verifies immutable metadata and dataset loading; it does not rerun native quality checks, query worker status, launch models or qualify runtime isolation.

Recorded V4 accounting is USD 23.371185; cumulative native campaign accounting is USD 120.288540. These combine recorded model usage with conservative compute estimates and are not provider invoices or complete actual-cost settlements. USD 5 of uncertain native operations remains held. The shared USD 600 expansion and original USD 100 reservations remain unchanged; registration makes no settlement or new reservation. Historical accounting disclosures remain preserved.

The publisher reports immutable byte readback of all 3,945 added files. Registration does not repeat that full payload download. Hub byte readback does not establish remote POSIX modes; consumers must restore the recorded modes before execution. Previous sources, indices, configurations, calibration results and card content are preserved.

Current accepted task views

This private dataset's default scale800_v1/train contains every accepted Go and Python task. scale800_best_v1/train keeps complete target-model measurements with 0 < qwen_solved < qwen_valid_attempts.

The former config named default is available as legacy_default_v1, with its original files, features and splits preserved.

import os
from datasets import load_dataset
repo = 'harness-hub/agentic-coding-envs'
token = os.environ['HF_TOKEN']
tasks = load_dataset(repo, split='train', token=token)
best = load_dataset(repo, name='scale800_best_v1', split='train', token=token, streaming=True)

Qwen3.8-27B scores use current qwen38-auto48-64k-2m-nothink* protocols: thinking off, a 2,000,000-token attempt budget, and either four-of-four or eight-of-eight valid attempts. Partial attempts have null numerical scores; protocols and attempts are never pooled. Generated alternative solutions and leakage audits were NOT RUN by this reused-gate scoring profile. Source flags such as target_model_calibrated=false, difficulty_status=unmeasured and usage=quarantine_only are retained; new qwen_status and qwen_* fields describe the joined current measurement. A train split name or mixed score does not establish trainability or semantic holdout nonoverlap.

in_stratified_sample is bound only to the owner's published roster and seed receipts: {"drawn_at":"2026-11-12T19:39:14.117186+00:00","membership_scope":"Selected IDs in the owner's frozen population; later nonselected IDs are false. No outcome or attempt-count inference.","population_count":825,"population_revision":"3afa3ffb350eae408cdc4962952452726a680559","sample_count":158,"sample_locator":{"path":"results/qwen-difficulty-auto-v1/stratified-samples/srs6-3afa-v1/d00ee4f9f721077b46e3657074b30ecc60fe9af743e30f46f908beae31b3124a/sample.json","repo_id":"harness-hub/agentic-coding-envs","revision":"edaa524aa77900bd6048896dff3f49d42dde4104","sha256":"dfc0b212570e239277a43189c2ee3e809787996c9d1c529d7f8b818ab84cee5d"},"seed_hex":"3340a71c3d236dd72732f5d05ff6144ead6ff6064d77004f29384dea010a1112","seed_locator":{"path":"results/qwen-difficulty-auto-v1/stratified-samples/srs6-3afa-v1/d00ee4f9f721077b46e3657074b30ecc60fe9af743e30f46f908beae31b3124a/seed.json","repo_id":"harness-hub/agentic-coding-envs","revision":"edaa524aa77900bd6048896dff3f49d42dde4104","sha256":"4d02842e0279a8446854639af36ff90720893c54da9ef8a66bccca61042c8f09"},"seed_recorded_at":"2026-11-12T19:34:40.949053+00:00","selected_task_ids_locator":{"path":"results/qwen-difficulty-auto-v1/stratified-samples/srs6-3afa-v1/d00ee4f9f721077b46e3657074b30ecc60fe9af743e30f46f908beae31b3124a/selected-task-ids.json","repo_id":"harness-hub/agentic-coding-envs","revision":"edaa524aa77900bd6048896dff3f49d42dde4104","sha256":"272ecee123c0ecda53b90044bc7736b73f13c3690290b623b53c0961f9321d26"},"status":"verified"}.

Downloads last month
1,029

Collection including harness-hub/harness-adapters