Hugging Face's logo

Datasets:
JOHNISKING
/
claude-edge-case-prompts

Tasks:
Text Generation
Modalities:
Tabular
Text
Formats:
csv
Languages:
English
Size:
< 1K
Tags:
claude
claude-code
llm-evaluation
code-generation
bug-fixing
prompt-engineering
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
run
stringlengths
15
19
model
stringclasses
2 values
condition
stringclasses
4 values
condition_text
stringclasses
4 values
bug
stringclasses
21 values
group
stringclasses
3 values
project
stringclasses
12 values
repo
stringclasses
12 values
parent_commit
stringclasses
21 values
fix_commit
stringclasses
21 values
rep
int64
1
3
reported_cases
int64
1
14
reported_passed
int64
0
14
unreported_cases
int64
1
178
unreported_passed
int64
0
168
unreported_target_cases
int64
0
141
unreported_target_passed
int64
0
136
other_tests_broken
int64
0
2
clean_fix
bool
2 classes
api_cost_usd
float64
0.01
1.72
seconds
float64
32
7.35k
api_seconds
float64
20.5
819
turns
int64
1
145
output_tokens
int64
98
128k
haiku_C0_any1273_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
any1273
H
anyio
https://github.com/agronholm/anyio
8dcb3e183e37db66488b17f07a22c7bfd40fb5ea
623a4806211aec456674d6943479993eb275569e
1
6
6
20
20
0
0
0
true
0.018277
186
70.1
23
9,811
haiku_C0_any1273_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
any1273
H
anyio
https://github.com/agronholm/anyio
8dcb3e183e37db66488b17f07a22c7bfd40fb5ea
623a4806211aec456674d6943479993eb275569e
2
6
6
20
20
0
0
0
true
0.033307
677
115.9
37
20,722
haiku_C0_any1273_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
any1273
H
anyio
https://github.com/agronholm/anyio
8dcb3e183e37db66488b17f07a22c7bfd40fb5ea
623a4806211aec456674d6943479993eb275569e
3
6
6
20
20
0
0
0
true
0.020406
275
67.7
22
12,214
haiku_C0_any1364_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
any1364
H
anyio
https://github.com/agronholm/anyio
c237c32a97d617e528adf5c80942ccbc88248c5f
c7b992cdd3a3e692f7f2e67c42917d8063120843
1
6
6
2
2
2
2
0
true
0.02012
234
72.1
21
12,771
haiku_C0_any1364_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
any1364
H
anyio
https://github.com/agronholm/anyio
c237c32a97d617e528adf5c80942ccbc88248c5f
c7b992cdd3a3e692f7f2e67c42917d8063120843
2
6
6
2
2
2
2
0
true
0.033437
292
115.3
30
20,274
haiku_C0_any1364_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
any1364
H
anyio
https://github.com/agronholm/anyio
c237c32a97d617e528adf5c80942ccbc88248c5f
c7b992cdd3a3e692f7f2e67c42917d8063120843
3
6
6
2
2
2
2
0
true
0.018329
269
72.1
21
11,395
haiku_C0_ct435_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct435
K
cachetools
https://github.com/tkem/cachetools
0985ec0d8fefdce90bcbba2dfe88117dad9bb6f6
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
1
1
0
1
1
1
1
0
false
0.020388
104
79.6
21
14,773
haiku_C0_ct435_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct435
K
cachetools
https://github.com/tkem/cachetools
0985ec0d8fefdce90bcbba2dfe88117dad9bb6f6
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
2
1
0
1
1
1
1
0
false
0.020232
96
77.8
17
15,060
haiku_C0_ct435_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct435
K
cachetools
https://github.com/tkem/cachetools
0985ec0d8fefdce90bcbba2dfe88117dad9bb6f6
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
3
1
0
1
1
1
1
1
false
0.023652
121
91.5
18
16,568
haiku_C0_ct437_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct437
H
cachetools
https://github.com/tkem/cachetools
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
1
3
3
8
8
8
8
0
true
0.019319
91
73.4
16
13,363
haiku_C0_ct437_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct437
H
cachetools
https://github.com/tkem/cachetools
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
2
3
3
8
8
8
8
0
true
0.022129
106
85.6
19
16,801
haiku_C0_ct437_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct437
H
cachetools
https://github.com/tkem/cachetools
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
3
3
3
8
8
8
8
0
true
0.019229
88
76.5
20
12,855
haiku_C0_ct440_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct440
N
cachetools
https://github.com/tkem/cachetools
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
7cfc8121722879e31fd82aa748f915690c336100
1
1
1
3
3
1
1
0
true
0.008137
38
27.2
10
3,867
haiku_C0_ct440_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct440
N
cachetools
https://github.com/tkem/cachetools
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
7cfc8121722879e31fd82aa748f915690c336100
2
1
1
3
3
1
1
0
true
0.009095
36
27
10
4,866
haiku_C0_ct440_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct440
N
cachetools
https://github.com/tkem/cachetools
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
7cfc8121722879e31fd82aa748f915690c336100
3
1
1
3
3
1
1
0
true
0.007545
32
22.7
9
3,280
haiku_C0_ct443_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct443
N
cachetools
https://github.com/tkem/cachetools
eba200635273fe45b67e669b84cb45f995b27c24
95e6addbd82afd75f81cb5650f32f6d74af2f1f9
1
1
1
2
2
1
1
0
true
0.025651
119
106.7
21
19,649
haiku_C0_ct443_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct443
N
cachetools
https://github.com/tkem/cachetools
eba200635273fe45b67e669b84cb45f995b27c24
95e6addbd82afd75f81cb5650f32f6d74af2f1f9
2
1
1
2
2
1
1
0
true
0.022054
97
86.4
21
15,649
haiku_C0_ct443_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct443
N
cachetools
https://github.com/tkem/cachetools
eba200635273fe45b67e669b84cb45f995b27c24
95e6addbd82afd75f81cb5650f32f6d74af2f1f9
3
1
1
2
1
1
0
0
false
0.033331
160
150.5
22
31,687
haiku_C0_ct446_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct446
N
cachetools
https://github.com/tkem/cachetools
36d4e87f3fd6ec557323e150dcb40123312f2730
b34643f37174751df37016899f4e7f5f6b189012
1
3
3
3
3
1
1
0
true
0.011065
42
33.7
13
5,945
haiku_C0_ct446_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct446
N
cachetools
https://github.com/tkem/cachetools
36d4e87f3fd6ec557323e150dcb40123312f2730
b34643f37174751df37016899f4e7f5f6b189012
2
3
3
3
3
1
1
0
true
0.013864
49
39.9
14
7,291
haiku_C0_ct446_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
ct446
N
cachetools
https://github.com/tkem/cachetools
36d4e87f3fd6ec557323e150dcb40123312f2730
b34643f37174751df37016899f4e7f5f6b189012
3
3
3
3
3
1
1
0
true
0.009396
32
26.8
10
4,881
haiku_C0_du1369_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
du1369
H
dateutil
https://github.com/dateutil/dateutil
bd5fd88b0b0255dd364b57af58acf89389c6c69d
21b6862a1e5674ac98695cb35afa5dd8ab6711c1
1
14
14
178
154
141
122
0
false
0.011918
139
53.9
19
10,296
haiku_C0_du1369_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
du1369
H
dateutil
https://github.com/dateutil/dateutil
bd5fd88b0b0255dd364b57af58acf89389c6c69d
21b6862a1e5674ac98695cb35afa5dd8ab6711c1
2
14
14
178
154
141
122
0
false
0.018838
159
69.5
22
13,431
haiku_C0_du1369_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
du1369
H
dateutil
https://github.com/dateutil/dateutil
bd5fd88b0b0255dd364b57af58acf89389c6c69d
21b6862a1e5674ac98695cb35afa5dd8ab6711c1
3
14
14
178
154
141
122
0
false
0.014776
92
58.6
19
9,558
haiku_C0_fy7073_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
fy7073
N
fastify
https://github.com/fastify/fastify
f736849a10a9b2ca57fb09437feb8294cda76234
9e9f61a6bf092bbd3cf8d10b792432aa8be085af
1
2
2
2
2
2
2
0
true
0.024715
225
107.2
31
14,687
haiku_C0_fy7073_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
fy7073
N
fastify
https://github.com/fastify/fastify
f736849a10a9b2ca57fb09437feb8294cda76234
9e9f61a6bf092bbd3cf8d10b792432aa8be085af
2
2
2
2
2
2
2
0
true
0.013876
94
44.1
18
7,909
haiku_C0_fy7073_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
fy7073
N
fastify
https://github.com/fastify/fastify
f736849a10a9b2ca57fb09437feb8294cda76234
9e9f61a6bf092bbd3cf8d10b792432aa8be085af
3
2
2
2
2
2
2
0
true
0.011034
59
33.9
15
6,064
haiku_C0_jy791_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
jy791
K
js-yaml
https://github.com/nodeca/js-yaml
04d6fe3298191ee94adf8b1bad87bd2d909092af
e5e6f621214e7d8b3f6e6f8ea29ea667f32a5861
1
1
1
1
0
1
0
0
false
0.01088
64
37.5
18
5,547
haiku_C0_jy791_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
jy791
K
js-yaml
https://github.com/nodeca/js-yaml
04d6fe3298191ee94adf8b1bad87bd2d909092af
e5e6f621214e7d8b3f6e6f8ea29ea667f32a5861
2
1
1
1
0
1
0
0
false
0.013374
62
49.2
21
6,467
haiku_C0_jy791_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
jy791
K
js-yaml
https://github.com/nodeca/js-yaml
04d6fe3298191ee94adf8b1bad87bd2d909092af
e5e6f621214e7d8b3f6e6f8ea29ea667f32a5861
3
1
1
1
0
1
0
0
false
0.006323
33
20.5
10
2,986
haiku_C0_mk4051_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
mk4051
K
marked
https://github.com/markedjs/marked
8698d35b84c17f86ea8a299a6183d47ccb3d2bd7
e250e31f3e1de80e41baf6f83ac8d3123d1e114f
1
3
3
1
1
1
1
1
false
0.026097
131
101.5
36
15,623
haiku_C0_mk4051_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
mk4051
K
marked
https://github.com/markedjs/marked
8698d35b84c17f86ea8a299a6183d47ccb3d2bd7
e250e31f3e1de80e41baf6f83ac8d3123d1e114f
2
3
3
1
0
1
0
1
false
0.027672
131
96.1
36
15,365
haiku_C0_mk4051_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
mk4051
K
marked
https://github.com/markedjs/marked
8698d35b84c17f86ea8a299a6183d47ccb3d2bd7
e250e31f3e1de80e41baf6f83ac8d3123d1e114f
3
3
3
1
0
1
0
1
false
0.023501
101
66.8
30
12,922
haiku_C0_pm1643_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pm1643
K
Python-Markdown
https://github.com/Python-Markdown/markdown
3ade80d7c2cad90c8961c80d9640d80a09afd33f
0bf535bc406c95dad66050f07c8402af03fe07d5
1
2
2
32
29
0
0
0
false
0.050917
261
162.7
46
29,704
haiku_C0_pm1643_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pm1643
K
Python-Markdown
https://github.com/Python-Markdown/markdown
3ade80d7c2cad90c8961c80d9640d80a09afd33f
0bf535bc406c95dad66050f07c8402af03fe07d5
2
2
2
32
27
0
0
0
false
0.024734
141
95.5
23
17,787
haiku_C0_pm1643_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pm1643
K
Python-Markdown
https://github.com/Python-Markdown/markdown
3ade80d7c2cad90c8961c80d9640d80a09afd33f
0bf535bc406c95dad66050f07c8402af03fe07d5
3
2
2
32
29
0
0
0
false
0.056725
256
195.2
56
36,788
haiku_C0_pm1668_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pm1668
N
Python-Markdown
https://github.com/Python-Markdown/markdown
3d70cfccdc0b54c0783edd5e7d168c5d8bfc0b1e
59f6dbc53bf4688c66ca832abb66c6a424f0626f
1
1
1
1
0
0
0
0
false
0.013067
106
47.7
16
7,621
haiku_C0_pm1668_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pm1668
N
Python-Markdown
https://github.com/Python-Markdown/markdown
3d70cfccdc0b54c0783edd5e7d168c5d8bfc0b1e
59f6dbc53bf4688c66ca832abb66c6a424f0626f
2
1
1
1
0
0
0
0
false
0.013669
87
58.1
19
8,929
haiku_C0_pm1668_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pm1668
N
Python-Markdown
https://github.com/Python-Markdown/markdown
3d70cfccdc0b54c0783edd5e7d168c5d8bfc0b1e
59f6dbc53bf4688c66ca832abb66c6a424f0626f
3
1
1
1
1
0
0
0
true
0.015053
80
46.1
20
7,038
haiku_C0_pq256_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pq256
N
p-queue
https://github.com/sindresorhus/p-queue
de565c42a271bd3c2f8c69d3b6bce33f49f9d344
507e7eeab74970b991f1335b3f08cb5e08207ca3
1
1
1
1
1
1
1
0
true
0.018433
189
60.5
30
9,282
haiku_C0_pq256_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pq256
N
p-queue
https://github.com/sindresorhus/p-queue
de565c42a271bd3c2f8c69d3b6bce33f49f9d344
507e7eeab74970b991f1335b3f08cb5e08207ca3
2
1
1
1
1
1
1
0
true
0.016408
310
62.7
22
9,797
haiku_C0_pq256_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
pq256
N
p-queue
https://github.com/sindresorhus/p-queue
de565c42a271bd3c2f8c69d3b6bce33f49f9d344
507e7eeab74970b991f1335b3f08cb5e08207ca3
3
1
1
1
1
1
1
0
true
0.012632
76
31.6
13
6,398
haiku_C0_q629_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
q629
K
qs
https://github.com/ljharb/qs
81e7a2251d5a7f4c38e8ea77a31a285ab16d1822
b43b1cf8d0747d8520746d8cf6998d568431816a
1
1
1
11
10
11
10
1
false
0.009054
48
36.2
11
4,505
haiku_C0_q629_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
q629
K
qs
https://github.com/ljharb/qs
81e7a2251d5a7f4c38e8ea77a31a285ab16d1822
b43b1cf8d0747d8520746d8cf6998d568431816a
2
1
1
11
11
11
11
1
false
0.011149
64
40.9
11
6,692
haiku_C0_q629_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
q629
K
qs
https://github.com/ljharb/qs
81e7a2251d5a7f4c38e8ea77a31a285ab16d1822
b43b1cf8d0747d8520746d8cf6998d568431816a
3
1
1
11
11
11
11
1
false
0.008294
38
28
11
4,614
haiku_C0_t336_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
t336
K
tomlkit
https://github.com/python-poetry/tomlkit
0518af253901e57872c03835db6fe9bacab40ee7
f466807e253b7479a3dbd9621f8b0782384ade9e
1
2
2
27
18
27
18
0
false
0.008445
32
26.7
13
3,898
haiku_C0_t336_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
t336
K
tomlkit
https://github.com/python-poetry/tomlkit
0518af253901e57872c03835db6fe9bacab40ee7
f466807e253b7479a3dbd9621f8b0782384ade9e
2
2
2
27
18
27
18
0
false
0.009035
41
34.4
13
4,529
haiku_C0_t336_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
t336
K
tomlkit
https://github.com/python-poetry/tomlkit
0518af253901e57872c03835db6fe9bacab40ee7
f466807e253b7479a3dbd9621f8b0782384ade9e
3
2
2
27
18
27
18
0
false
0.008529
36
28.1
15
4,269
haiku_C0_un5981_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
un5981
H
undici
https://github.com/nodejs/undici
aa1eef17b816caba4b62514f6a38505cf6c60677
0bbf614b81ecd9e6fffc9d2edc5082d3c1b0b4bd
1
2
2
1
1
1
1
0
true
0.02715
143
93.8
34
16,102
haiku_C0_un5981_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
un5981
H
undici
https://github.com/nodejs/undici
aa1eef17b816caba4b62514f6a38505cf6c60677
0bbf614b81ecd9e6fffc9d2edc5082d3c1b0b4bd
2
2
2
1
1
1
1
0
true
0.023253
110
80.4
27
13,614
haiku_C0_un5981_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
un5981
H
undici
https://github.com/nodejs/undici
aa1eef17b816caba4b62514f6a38505cf6c60677
0bbf614b81ecd9e6fffc9d2edc5082d3c1b0b4bd
3
2
2
1
1
1
1
0
true
0.032543
136
108.7
32
16,711
haiku_C0_z6495_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6495
K
zod
https://github.com/colinhacks/zod
1a16102a494b03ce1df7b80b663ae2df465f419e
84e416fbf4740527bbc8f319634f4e1b065bb42c
1
4
4
4
4
1
1
1
false
0.06532
350
193.1
69
25,552
haiku_C0_z6495_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6495
K
zod
https://github.com/colinhacks/zod
1a16102a494b03ce1df7b80b663ae2df465f419e
84e416fbf4740527bbc8f319634f4e1b065bb42c
2
4
4
4
4
1
1
0
true
0.056457
281
157.5
47
24,560
haiku_C0_z6495_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6495
K
zod
https://github.com/colinhacks/zod
1a16102a494b03ce1df7b80b663ae2df465f419e
84e416fbf4740527bbc8f319634f4e1b065bb42c
3
4
4
4
4
1
1
1
false
0.190051
428
203.1
68
28,431
haiku_C0_z6496_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6496
K
zod
https://github.com/colinhacks/zod
7e24a24288183ce02554f1ded7775d0650a7b7e6
eab51ff3592b2d11d863f4ee4d5452f31a3de1b6
1
2
1
22
15
7
0
0
false
0.021918
242
62.8
22
9,176
haiku_C0_z6496_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6496
K
zod
https://github.com/colinhacks/zod
7e24a24288183ce02554f1ded7775d0650a7b7e6
eab51ff3592b2d11d863f4ee4d5452f31a3de1b6
2
2
1
22
17
7
2
0
false
0.028269
229
93.3
29
16,031
haiku_C0_z6496_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6496
K
zod
https://github.com/colinhacks/zod
7e24a24288183ce02554f1ded7775d0650a7b7e6
eab51ff3592b2d11d863f4ee4d5452f31a3de1b6
3
2
1
22
16
7
1
0
false
0.030969
298
91.2
32
14,421
haiku_C0_z6641_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6641
N
zod
https://github.com/colinhacks/zod
b500a2ee95134f81d74da7911d7378dd2853d042
311cd74b2765ff7c7b58add60e37c5a0ea7a1972
1
6
6
9
9
4
4
0
true
0.776113
1,345
818.5
145
99,172
haiku_C0_z6641_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6641
N
zod
https://github.com/colinhacks/zod
b500a2ee95134f81d74da7911d7378dd2853d042
311cd74b2765ff7c7b58add60e37c5a0ea7a1972
2
6
6
9
9
4
4
0
true
0.311023
568
363.8
84
67,321
haiku_C0_z6641_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6641
N
zod
https://github.com/colinhacks/zod
b500a2ee95134f81d74da7911d7378dd2853d042
311cd74b2765ff7c7b58add60e37c5a0ea7a1972
3
6
6
9
9
4
4
0
true
0.454982
853
424.6
109
81,203
haiku_C0_z6659_r1
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6659
N
zod
https://github.com/colinhacks/zod
40a5dba080b065cc430e5b0e64a8058d68c1660d
680d19f7d1e042e55934209caeb79bc843b94f44
1
3
3
6
6
0
0
0
true
0.040898
433
138.9
37
19,733
haiku_C0_z6659_r2
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6659
N
zod
https://github.com/colinhacks/zod
40a5dba080b065cc430e5b0e64a8058d68c1660d
680d19f7d1e042e55934209caeb79bc843b94f44
2
3
3
6
6
0
0
0
true
0.181723
399
189.3
51
33,973
haiku_C0_z6659_r3
claude-haiku-5-5
C0
baseline (symptom-only report, effort medium)
z6659
N
zod
https://github.com/colinhacks/zod
40a5dba080b065cc430e5b0e64a8058d68c1660d
680d19f7d1e042e55934209caeb79bc843b94f44
3
3
3
6
6
0
0
0
true
0.039818
407
181.7
32
25,296
haiku_C1_any1273_r1
claude-haiku-5-5
C1
effort high
any1273
H
anyio
https://github.com/agronholm/anyio
8dcb3e183e37db66488b17f07a22c7bfd40fb5ea
623a4806211aec456674d6943479993eb275569e
1
6
6
20
20
0
0
0
true
0.095876
1,383
284.3
51
47,900
haiku_C1_any1273_r2
claude-haiku-5-5
C1
effort high
any1273
H
anyio
https://github.com/agronholm/anyio
8dcb3e183e37db66488b17f07a22c7bfd40fb5ea
623a4806211aec456674d6943479993eb275569e
2
6
6
20
20
0
0
0
true
0.039763
751
135.1
33
24,015
haiku_C1_any1273_r3
claude-haiku-5-5
C1
effort high
any1273
H
anyio
https://github.com/agronholm/anyio
8dcb3e183e37db66488b17f07a22c7bfd40fb5ea
623a4806211aec456674d6943479993eb275569e
3
6
6
20
20
0
0
0
true
0.042492
759
133.9
40
24,448
haiku_C1_any1364_r1
claude-haiku-5-5
C1
effort high
any1364
H
anyio
https://github.com/agronholm/anyio
c237c32a97d617e528adf5c80942ccbc88248c5f
c7b992cdd3a3e692f7f2e67c42917d8063120843
1
6
6
2
2
2
2
0
true
0.037696
312
173.1
35
26,357
haiku_C1_any1364_r2
claude-haiku-5-5
C1
effort high
any1364
H
anyio
https://github.com/agronholm/anyio
c237c32a97d617e528adf5c80942ccbc88248c5f
c7b992cdd3a3e692f7f2e67c42917d8063120843
2
6
6
2
2
2
2
0
true
0.035693
313
118.9
32
23,269
haiku_C1_any1364_r3
claude-haiku-5-5
C1
effort high
any1364
H
anyio
https://github.com/agronholm/anyio
c237c32a97d617e528adf5c80942ccbc88248c5f
c7b992cdd3a3e692f7f2e67c42917d8063120843
3
6
6
2
2
2
2
0
true
0.034045
224
131.8
33
24,800
haiku_C1_ct435_r1
claude-haiku-5-5
C1
effort high
ct435
K
cachetools
https://github.com/tkem/cachetools
0985ec0d8fefdce90bcbba2dfe88117dad9bb6f6
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
1
1
0
1
1
1
1
1
false
0.040154
180
160
27
33,930
haiku_C1_ct435_r2
claude-haiku-5-5
C1
effort high
ct435
K
cachetools
https://github.com/tkem/cachetools
0985ec0d8fefdce90bcbba2dfe88117dad9bb6f6
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
2
1
0
1
1
1
1
0
false
0.026826
117
103.2
24
19,135
haiku_C1_ct435_r3
claude-haiku-5-5
C1
effort high
ct435
K
cachetools
https://github.com/tkem/cachetools
0985ec0d8fefdce90bcbba2dfe88117dad9bb6f6
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
3
1
0
1
1
1
1
0
false
0.031943
174
139.2
19
28,044
haiku_C1_ct437_r1
claude-haiku-5-5
C1
effort high
ct437
H
cachetools
https://github.com/tkem/cachetools
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
1
3
3
8
8
8
8
0
true
0.030846
142
122.5
25
25,111
haiku_C1_ct437_r2
claude-haiku-5-5
C1
effort high
ct437
H
cachetools
https://github.com/tkem/cachetools
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
2
3
3
8
8
8
8
0
true
0.029768
127
107.3
24
23,094
haiku_C1_ct437_r3
claude-haiku-5-5
C1
effort high
ct437
H
cachetools
https://github.com/tkem/cachetools
aa73a19299432530eb9b1c0f9acecde6ad96c9e0
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
3
3
3
8
8
8
8
0
true
0.022379
113
84.4
20
16,870
haiku_C1_ct440_r1
claude-haiku-5-5
C1
effort high
ct440
N
cachetools
https://github.com/tkem/cachetools
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
7cfc8121722879e31fd82aa748f915690c336100
1
1
1
3
3
1
1
0
true
0.011493
61
40
15
5,973
haiku_C1_ct440_r2
claude-haiku-5-5
C1
effort high
ct440
N
cachetools
https://github.com/tkem/cachetools
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
7cfc8121722879e31fd82aa748f915690c336100
2
1
1
3
3
1
1
0
true
0.01525
61
48.8
19
9,095
haiku_C1_ct440_r3
claude-haiku-5-5
C1
effort high
ct440
N
cachetools
https://github.com/tkem/cachetools
ec0b97b8a43dbd61fa83d702dadd2ff21131bd26
7cfc8121722879e31fd82aa748f915690c336100
3
1
1
3
3
1
1
0
true
0.009822
44
32.6
12
5,365
haiku_C1_ct443_r1
claude-haiku-5-5
C1
effort high
ct443
N
cachetools
https://github.com/tkem/cachetools
eba200635273fe45b67e669b84cb45f995b27c24
95e6addbd82afd75f81cb5650f32f6d74af2f1f9
1
1
1
2
2
1
1
0
true
0.035772
165
153.3
27
29,818
haiku_C1_ct443_r2
claude-haiku-5-5
C1
effort high
ct443
N
cachetools
https://github.com/tkem/cachetools
eba200635273fe45b67e669b84cb45f995b27c24
95e6addbd82afd75f81cb5650f32f6d74af2f1f9
2
1
1
2
2
1
1
0
true
0.04341
200
185.3
27
37,768
haiku_C1_ct443_r3
claude-haiku-5-5
C1
effort high
ct443
N
cachetools
https://github.com/tkem/cachetools
eba200635273fe45b67e669b84cb45f995b27c24
95e6addbd82afd75f81cb5650f32f6d74af2f1f9
3
1
1
2
1
1
0
0
false
0.140437
333
315.4
34
65,294
haiku_C1_ct446_r1
claude-haiku-5-5
C1
effort high
ct446
N
cachetools
https://github.com/tkem/cachetools
36d4e87f3fd6ec557323e150dcb40123312f2730
b34643f37174751df37016899f4e7f5f6b189012
1
3
3
3
3
1
1
0
true
0.018628
74
65.5
16
12,825
haiku_C1_ct446_r2
claude-haiku-5-5
C1
effort high
ct446
N
cachetools
https://github.com/tkem/cachetools
36d4e87f3fd6ec557323e150dcb40123312f2730
b34643f37174751df37016899f4e7f5f6b189012
2
3
3
3
3
1
1
0
true
0.017218
68
59
19
12,172
haiku_C1_ct446_r3
claude-haiku-5-5
C1
effort high
ct446
N
cachetools
https://github.com/tkem/cachetools
36d4e87f3fd6ec557323e150dcb40123312f2730
b34643f37174751df37016899f4e7f5f6b189012
3
3
3
3
3
1
1
0
true
0.029891
109
99.8
26
19,261
haiku_C1_du1369_r1
claude-haiku-5-5
C1
effort high
du1369
H
dateutil
https://github.com/dateutil/dateutil
bd5fd88b0b0255dd364b57af58acf89389c6c69d
21b6862a1e5674ac98695cb35afa5dd8ab6711c1
1
14
14
178
154
141
122
0
false
0.025882
141
99.4
26
20,631
haiku_C1_du1369_r2
claude-haiku-5-5
C1
effort high
du1369
H
dateutil
https://github.com/dateutil/dateutil
bd5fd88b0b0255dd364b57af58acf89389c6c69d
21b6862a1e5674ac98695cb35afa5dd8ab6711c1
2
14
14
178
140
141
108
0
false
0.032372
185
120
32
25,220
haiku_C1_du1369_r3
claude-haiku-5-5
C1
effort high
du1369
H
dateutil
https://github.com/dateutil/dateutil
bd5fd88b0b0255dd364b57af58acf89389c6c69d
21b6862a1e5674ac98695cb35afa5dd8ab6711c1
3
14
14
178
140
141
108
0
false
0.039949
197
153.6
32
30,311
haiku_C1_fy7073_r1
claude-haiku-5-5
C1
effort high
fy7073
N
fastify
https://github.com/fastify/fastify
f736849a10a9b2ca57fb09437feb8294cda76234
9e9f61a6bf092bbd3cf8d10b792432aa8be085af
1
2
2
2
2
2
2
0
true
0.032609
216
136.2
37
20,910
haiku_C1_fy7073_r2
claude-haiku-5-5
C1
effort high
fy7073
N
fastify
https://github.com/fastify/fastify
f736849a10a9b2ca57fb09437feb8294cda76234
9e9f61a6bf092bbd3cf8d10b792432aa8be085af
2
2
2
2
2
2
2
0
true
0.038706
234
121.2
56
21,879
haiku_C1_fy7073_r3
claude-haiku-5-5
C1
effort high
fy7073
N
fastify
https://github.com/fastify/fastify
f736849a10a9b2ca57fb09437feb8294cda76234
9e9f61a6bf092bbd3cf8d10b792432aa8be085af
3
2
2
2
2
2
2
0
true
0.024692
128
74
30
13,365
haiku_C1_jy791_r1
claude-haiku-5-5
C1
effort high
jy791
K
js-yaml
https://github.com/nodeca/js-yaml
04d6fe3298191ee94adf8b1bad87bd2d909092af
e5e6f621214e7d8b3f6e6f8ea29ea667f32a5861
1
1
1
1
0
1
0
0
false
0.036764
260
239.1
43
22,835
haiku_C1_jy791_r2
claude-haiku-5-5
C1
effort high
jy791
K
js-yaml
https://github.com/nodeca/js-yaml
04d6fe3298191ee94adf8b1bad87bd2d909092af
e5e6f621214e7d8b3f6e6f8ea29ea667f32a5861
2
1
1
1
0
1
0
0
false
0.020736
95
74.9
27
12,071
haiku_C1_jy791_r3
claude-haiku-5-5
C1
effort high
jy791
K
js-yaml
https://github.com/nodeca/js-yaml
04d6fe3298191ee94adf8b1bad87bd2d909092af
e5e6f621214e7d8b3f6e6f8ea29ea667f32a5861
3
1
1
1
0
1
0
0
false
0.017172
107
61.1
24
8,013
haiku_C1_mk4051_r1
claude-haiku-5-5
C1
effort high
mk4051
K
marked
https://github.com/markedjs/marked
8698d35b84c17f86ea8a299a6183d47ccb3d2bd7
e250e31f3e1de80e41baf6f83ac8d3123d1e114f
1
3
3
1
0
1
0
0
false
0.049646
268
176.1
55
31,114
haiku_C1_mk4051_r2
claude-haiku-5-5
C1
effort high
mk4051
K
marked
https://github.com/markedjs/marked
8698d35b84c17f86ea8a299a6183d47ccb3d2bd7
e250e31f3e1de80e41baf6f83ac8d3123d1e114f
2
3
3
1
0
1
0
1
false
0.052076
267
175.9
54
27,928
haiku_C1_mk4051_r3
claude-haiku-5-5
C1
effort high
mk4051
K
marked
https://github.com/markedjs/marked
8698d35b84c17f86ea8a299a6183d47ccb3d2bd7
e250e31f3e1de80e41baf6f83ac8d3123d1e114f
3
3
3
1
1
1
1
1
false
0.046488
289
215.7
50
28,533
haiku_C1_pm1643_r1
claude-haiku-5-5
C1
effort high
pm1643
K
Python-Markdown
https://github.com/Python-Markdown/markdown
3ade80d7c2cad90c8961c80d9640d80a09afd33f
0bf535bc406c95dad66050f07c8402af03fe07d5
1
2
2
32
30
0
0
0
false
0.054254
273
202
48
38,488
haiku_C1_pm1643_r2
claude-haiku-5-5
C1
effort high
pm1643
K
Python-Markdown
https://github.com/Python-Markdown/markdown
3ade80d7c2cad90c8961c80d9640d80a09afd33f
0bf535bc406c95dad66050f07c8402af03fe07d5
2
2
2
32
29
0
0
0
false
0.100943
310
262.9
62
52,166
haiku_C1_pm1643_r3
claude-haiku-5-5
C1
effort high
pm1643
K
Python-Markdown
https://github.com/Python-Markdown/markdown
3ade80d7c2cad90c8961c80d9640d80a09afd33f
0bf535bc406c95dad66050f07c8402af03fe07d5
3
2
2
32
29
0
0
0
false
0.170013
309
274.9
67
50,584
haiku_C1_pm1668_r1
claude-haiku-5-5
C1
effort high
pm1668
N
Python-Markdown
https://github.com/Python-Markdown/markdown
3d70cfccdc0b54c0783edd5e7d168c5d8bfc0b1e
59f6dbc53bf4688c66ca832abb66c6a424f0626f
1
1
1
1
0
0
0
0
false
0.025993
178
106.4
25
19,356
End of preview. Expand in Data Studio

Pass rate flat, cost up

Edge-case prompts and effort in Claude Code: 504 runs on 21 real bugs

Can a prompt or a higher effort setting make Claude Code handle the cases a bug report leaves out? 504 pre-registered Claude Code runs on 21 real bugs fixed between August and October 2026 in 12 open-source projects: Claude Haiku 5.5 and Opus 5.5, four conditions, 3 runs each. Every run is scored by the maintainers' own tests for the fix, split into cases the report shows and cases it leaves out.

Page with tables and method: https://tokensave.app/data/claude-edge-case-prompts

Findings

  • The prompt and effort did not help. Share of unreported maintainer test cases passed (mean per bug), baseline / effort high / edge-case prompt / edge-case prompt + read tests: Haiku 84% / 82% / 84% / 86%; Opus 88% / 86% / 91% / 84%. No condition moved it by more than 4 points (pre-registered threshold: 20 points with a bug-level bootstrap interval excluding 0).
  • They cost more. Mean API cost per run: Haiku $0.050 / $0.081 / $0.10 / $0.12 (+60%, +108%, +134%); Opus $0.51 / $0.62 / $0.82 / $0.81 (+22%, +60%, +58%). Median time per run more than doubled with the edge-case prompt.
  • Clean fixes barely moved (of 63 per cell): Haiku 34, 33, 36, 37; Opus 38, 38, 39, 37.
  • Haiku vs Opus: $0.088 vs $0.69 per run (about 87% less), 140 vs 152 clean fixes over 252 runs each. Opus broke other tests in fewer runs.
  • The bug decided the result. On 12 of 21 bugs both models passed every unreported case in every condition. On a js-yaml bug (a BOM before every document in a stream) no run passed the unreported case.

Files

  • claude-edge-case-prompts.csv: one row per run (model, condition, bug, group, project, repo, parent and fix commit, rep, reported and unreported cases passed, unreported targets, other tests broken, clean fix, API cost, seconds, API seconds, turns, output tokens).
  • claude-edge-case-prompts.json: the same runs plus bug list, conditions, per-bug results and statistics (bug-level bootstrap intervals vs baseline, cost change).
  • claude-edge-case-prompts-chart.png: pass rate and cost per run by condition and model.
  • Pre-registration, labels, prompts and scoring code: https://github.com/johnisking/token-cost-by-language/tree/main/claude-report-edge-cases

Method

  • Conditions: baseline (symptom-only report, effort medium); effort high; edge-case prompt ("Before changing code, list the inputs and variants this report does not mention but the fix must also handle. Write tests for them first, then fix."); edge-case prompt + "Read the existing tests for the code you change before you start."
  • Agent: Claude Code 2.1.295 on Windows 11, headless (-p), acceptEdits; the project's tests allowed; no installs, no network.
  • Scoring: only the run's changes to library source were kept. The maintainers' tests for the fix were run 5 times at the parent and the fix; unstable tests were dropped. Unreported cases include cases that already pass before the fix (guards), which a careless fix can break.

Limitations

21 bugs, 3 runs per cell: a 20-point effect is ruled out, a few points is not. One prompt wording per condition. 12 of 21 bugs were at the ceiling for both models. anyio bugs were scored on Windows, the rest on Linux. The scoring setup was rebuilt after a workspace reset; re-scoring 228 Haiku runs scored before the reset gave identical numbers.

Previous dataset: https://huggingface.co/datasets/JOHNISKING/claude-haiku-vs-opus-real-bugs · GitHub: https://github.com/johnisking/token-cost-by-language/tree/main/claude-report-edge-cases

License: CC BY 4.0. Cite: TokenSave (2026). Raising effort didn't catch more edge cases, and Haiku got nearly the same result for 87% less: 504 Claude Code runs on 21 real bugs [Data set]. https://tokensave.app/data/claude-edge-case-prompts
Downloads last month
73
Number of rows:
504
Total file size:
534 kB