item_id
stringlengths
21
26
family
stringclasses
383 values
twin_of
stringlengths
0
21
system
stringlengths
1.51k
1.51k
user
stringlengths
18.6k
288k
input_tokens
int64
6.38k
123k
budget_tokens
int64
1k
3.07k
cut
int64
19
151
steps
int64
32
166
resolved
bool
1 class
image
stringlengths
25
33
source
stringclasses
1 value
facts
stringlengths
585
1.33k
probes
stringlengths
1.29k
1.44k
ctx_compress-s1-01959-twin
stemflow
ctx_compress-s1-01959
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_slug returns one item too few on every page. It lives in src/stemflow/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the stemflow layout. [tool_use kind=search name...
66,429
3,072
71
136
false
synth:7744578879738306569@2.32762
synth
[{"aliases": ["src/stemflow/rules.py", "rules.py"], "kind": "file", "value": "src/stemflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/lexeme_8.csv", "lexeme_8.csv"], "kind": "file", "value": "data/lexeme_8.csv"}, {"aliases": ["da...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/snippet_9.csv", "snippet_9.csv"]...
ctx_compress-s1-01960
refundkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_invoice_window leaves a trailing dash when the title ends in punctuation. It lives in src/refundkit/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the refundkit la...
88,151
3,072
100
117
false
synth:7691596269042754065@1.84193
synth
[{"aliases": ["src/refundkit/cli.py", "cli.py"], "kind": "file", "value": "src/refundkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/balance_6.csv", "balance_6.csv"], "kind": "file", "value": "data/balance_6.csv"}, {"aliases": ["data/tari...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_io.py", "tests/test_io.py"], "ki...
ctx_compress-s1-01960-twin
refundkit
ctx_compress-s1-01960
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_invoice_window leaves a trailing dash when the title ends in punctuation. It lives in src/refundkit/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the refundkit la...
47,422
2,371
58
117
false
synth:7691596269042754065@1.84193
synth
[{"aliases": ["src/refundkit/cli.py", "cli.py"], "kind": "file", "value": "src/refundkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/balance_6.csv", "balance_6.csv"], "kind": "file", "value": "data/balance_6.csv"}, {"aliases": ["data/tari...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["balance_6.csv", "data/balance_6.csv"]...
ctx_compress-s1-01961
headingledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_clause raises SectionBoundError for bytes input in strict mode instead of decoding it. It lives in src/headingledger/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappin...
87,588
3,072
93
108
false
synth:2347715730775519696@1.98495
synth
[{"aliases": ["src/headingledger/rules.py", "rules.py"], "kind": "file", "value": "src/headingledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/headingledger/report.py", "report.py"], "kind": "file", "value": "src/headingledger/re...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["io.py", "src/headingledger/io.py"], "...
ctx_compress-s1-01961-twin
headingledger
ctx_compress-s1-01961
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_clause raises SectionBoundError for bytes input in strict mode instead of decoding it. It lives in src/headingledger/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappin...
50,035
2,501
59
108
false
synth:2347715730775519696@1.98495
synth
[{"aliases": ["src/headingledger/rules.py", "rules.py"], "kind": "file", "value": "src/headingledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/headingledger/report.py", "report.py"], "kind": "file", "value": "src/headingledger/re...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/margin_3.csv", "margin_3.csv"], ...
ctx_compress-s1-01962
refundforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_tariff lets defaults overwrite explicit overrides. It lives in src/refundforge/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the refundforge layout. [tool_use kin...
95,655
3,072
116
125
false
synth:1760587676463669797@1.64185
synth
[{"aliases": ["src/refundforge/limits.py", "limits.py"], "kind": "file", "value": "src/refundforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/refundforge/core.py", "core.py"], "kind": "file", "value": "src/refundforge/core.py"...
[{"answers": ["src/refundforge/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01962-twin
refundforge
ctx_compress-s1-01962
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_tariff lets defaults overwrite explicit overrides. It lives in src/refundforge/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the refundforge layout. [tool_use kin...
49,415
2,470
66
125
false
synth:1760587676463669797@1.64185
synth
[{"aliases": ["src/refundforge/limits.py", "limits.py"], "kind": "file", "value": "src/refundforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/refundforge/core.py", "core.py"], "kind": "file", "value": "src/refundforge/core.py"...
[{"answers": ["src/refundforge/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01963
slotdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_booking double counts carried amounts when the period closes on a weekend. It lives in src/slotdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slotdesk l...
33,824
1,691
82
97
false
synth:8624433810781713804@1
synth
[{"aliases": ["src/slotdesk/store.py", "store.py"], "kind": "file", "value": "src/slotdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/slotdesk/cli.py", "cli.py"], "kind": "file", "value": "src/slotdesk/cli.py"}, {"aliases": ["data...
[{"answers": ["src/slotdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-01963-twin
slotdesk
ctx_compress-s1-01963
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_booking double counts carried amounts when the period closes on a weekend. It lives in src/slotdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slotdesk l...
19,905
1,000
50
97
false
synth:8624433810781713804@1
synth
[{"aliases": ["src/slotdesk/store.py", "store.py"], "kind": "file", "value": "src/slotdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/slotdesk/cli.py", "cli.py"], "kind": "file", "value": "src/slotdesk/cli.py"}, {"aliases": ["data...
[{"answers": ["src/slotdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-01964
bucketkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series_all raises SampleDriftError for bytes input in strict mode instead of decoding it. It lives in src/bucketkit/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping t...
35,856
1,792
70
79
false
synth:2754473880011104461@1
synth
[{"aliases": ["src/bucketkit/io.py", "io.py"], "kind": "file", "value": "src/bucketkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/bucketkit/store.py", "store.py"], "kind": "file", "value": "src/bucketkit/store.py"}, {"aliases": ["src/bucketki...
[{"answers": ["src/bucketkit/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-01964-twin
bucketkit
ctx_compress-s1-01964
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series_all raises SampleDriftError for bytes input in strict mode instead of decoding it. It lives in src/bucketkit/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping t...
23,293
1,164
46
79
false
synth:2754473880011104461@1
synth
[{"aliases": ["src/bucketkit/io.py", "io.py"], "kind": "file", "value": "src/bucketkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/bucketkit/store.py", "store.py"], "kind": "file", "value": "src/bucketkit/store.py"}, {"aliases": ["src/bucketki...
[{"answers": ["src/bucketkit/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-01965
clausekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_heading_strict produces doubled dashes for consecutive separators. It lives in src/clausekit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the clausekit layout...
44,447
2,222
67
84
false
synth:8830413416780242600@1
synth
[{"aliases": ["src/clausekit/rules.py", "rules.py"], "kind": "file", "value": "src/clausekit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/clausekit/audit.py", "audit.py"], "kind": "file", "value": "src/clausekit/audit.py"}, {"aliase...
[{"answers": ["src/clausekit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01965-twin
clausekit
ctx_compress-s1-01965
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_heading_strict produces doubled dashes for consecutive separators. It lives in src/clausekit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the clausekit layout...
23,470
1,173
39
84
false
synth:8830413416780242600@1
synth
[{"aliases": ["src/clausekit/rules.py", "rules.py"], "kind": "file", "value": "src/clausekit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/clausekit/audit.py", "audit.py"], "kind": "file", "value": "src/clausekit/audit.py"}, {"aliase...
[{"answers": ["src/clausekit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01966
skukit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_stock double counts carried amounts when the period closes on a weekend. It lives in src/skukit/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the skukit layout....
18,069
1,000
49
58
false
synth:4258764283781844712@1
synth
[{"aliases": ["src/skukit/compat.py", "compat.py"], "kind": "file", "value": "src/skukit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/bin_2.csv", "bin_2.csv"], "kind": "file", "value": "data/bin_2.csv"}, {"aliases": ["src/skuki...
[{"answers": ["src/skukit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-01966-twin
skukit
ctx_compress-s1-01966
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_stock double counts carried amounts when the period closes on a weekend. It lives in src/skukit/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the skukit layout....
13,098
1,000
31
58
false
synth:4258764283781844712@1
synth
[{"aliases": ["src/skukit/compat.py", "compat.py"], "kind": "file", "value": "src/skukit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/bin_2.csv", "bin_2.csv"], "kind": "file", "value": "data/bin_2.csv"}, {"aliases": ["src/skuki...
[{"answers": ["src/skukit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-01967
snippetkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_glyph_safe keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/snippetkit/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the snipp...
53,251
2,662
71
88
false
synth:8492802218069311283@1
synth
[{"aliases": ["src/snippetkit/limits.py", "limits.py"], "kind": "file", "value": "src/snippetkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/snippetkit/tables.py", "tables.py"], "kind": "file", "value": "src/snippetkit/tables.p...
[{"answers": ["src/snippetkit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01967-twin
snippetkit
ctx_compress-s1-01967
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_glyph_safe keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/snippetkit/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the snipp...
38,064
1,903
51
88
false
synth:8492802218069311283@1
synth
[{"aliases": ["src/snippetkit/limits.py", "limits.py"], "kind": "file", "value": "src/snippetkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/snippetkit/tables.py", "tables.py"], "kind": "file", "value": "src/snippetkit/tables.p...
[{"answers": ["src/snippetkit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01968
quotakit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_window skips the first page; page 1 returns what should be page 2. It lives in src/quotakit/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the quotakit layout....
59,183
2,959
91
102
false
synth:7521294142760864625@1.47618
synth
[{"aliases": ["src/quotakit/format.py", "format.py"], "kind": "file", "value": "src/quotakit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/quotakit/audit.py", "audit.py"], "kind": "file", "value": "src/quotakit/audit.py"}, {"alia...
[{"answers": ["src/quotakit/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01968-twin
quotakit
ctx_compress-s1-01968
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_window skips the first page; page 1 returns what should be page 2. It lives in src/quotakit/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the quotakit layout....
31,939
1,596
53
102
false
synth:7521294142760864625@1.47618
synth
[{"aliases": ["src/quotakit/format.py", "format.py"], "kind": "file", "value": "src/quotakit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/quotakit/audit.py", "audit.py"], "kind": "file", "value": "src/quotakit/audit.py"}, {"alia...
[{"answers": ["src/quotakit/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01969
marginforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_draft_once returns one item too few on every page. It lives in src/marginforge/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the marginforge layout. [tool_use kind=searc...
62,713
3,072
90
109
false
synth:23303855932688501@1.51685
synth
[{"aliases": ["src/marginforge/io.py", "io.py"], "kind": "file", "value": "src/marginforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/marginforge/store.py", "store.py"], "kind": "file", "value": "src/marginforge/store.py"}, {"aliases": ["data...
[{"answers": ["src/marginforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-01969-twin
marginforge
ctx_compress-s1-01969
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_draft_once returns one item too few on every page. It lives in src/marginforge/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the marginforge layout. [tool_use kind=searc...
28,944
1,447
50
109
false
synth:23303855932688501@1.51685
synth
[{"aliases": ["src/marginforge/io.py", "io.py"], "kind": "file", "value": "src/marginforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/marginforge/store.py", "store.py"], "kind": "file", "value": "src/marginforge/store.py"}, {"aliases": ["data...
[{"answers": ["src/marginforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rec...
ctx_compress-s1-01970
tariffkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_tariff skips the first page; page 1 returns what should be page 2. It lives in src/tariffkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tariffkit layout. [t...
72,504
3,072
114
123
false
synth:3547682833900471226@1.68902
synth
[{"aliases": ["src/tariffkit/rules.py", "rules.py"], "kind": "file", "value": "src/tariffkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/tariffkit/audit.py", "audit.py"], "kind": "file", "value": "src/tariffkit/audit.py"}, {"aliase...
[{"answers": ["src/tariffkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01970-twin
tariffkit
ctx_compress-s1-01970
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_tariff skips the first page; page 1 returns what should be page 2. It lives in src/tariffkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tariffkit layout. [t...
38,943
1,947
63
123
false
synth:3547682833900471226@1.68902
synth
[{"aliases": ["src/tariffkit/rules.py", "rules.py"], "kind": "file", "value": "src/tariffkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/tariffkit/audit.py", "audit.py"], "kind": "file", "value": "src/tariffkit/audit.py"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["io.py", "src/tariffkit/io.py"], "kind...
ctx_compress-s1-01971
balanceflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_tariff keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/balanceflow/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the balanceflo...
84,044
3,072
110
125
false
synth:4326862108541025341@1.89105
synth
[{"aliases": ["src/balanceflow/report.py", "report.py"], "kind": "file", "value": "src/balanceflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/payout_0.csv", "payout_0.csv"], "kind": "file", "value": "data/payout_0.csv"}, {"al...
[{"answers": ["src/balanceflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01971-twin
balanceflow
ctx_compress-s1-01971
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_tariff keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/balanceflow/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the balanceflo...
46,425
2,321
66
125
false
synth:4326862108541025341@1.89105
synth
[{"aliases": ["src/balanceflow/report.py", "report.py"], "kind": "file", "value": "src/balanceflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/payout_0.csv", "payout_0.csv"], "kind": "file", "value": "data/payout_0.csv"}, {"al...
[{"answers": ["src/balanceflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01972
headingforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_section ignores the cap, so late attempts wait for minutes. It lives in src/headingforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the headingforge layout. [to...
29,523
1,476
60
71
false
synth:4935545869915738682@1
synth
[{"aliases": ["src/headingforge/store.py", "store.py"], "kind": "file", "value": "src/headingforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/headingforge/core.py", "core.py"], "kind": "file", "value": "src/headingforge/core.py"},...
[{"answers": ["src/headingforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01972-twin
headingforge
ctx_compress-s1-01972
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_section ignores the cap, so late attempts wait for minutes. It lives in src/headingforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the headingforge layout. [to...
15,216
1,000
36
71
false
synth:4935545869915738682@1
synth
[{"aliases": ["src/headingforge/store.py", "store.py"], "kind": "file", "value": "src/headingforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/headingforge/core.py", "core.py"], "kind": "file", "value": "src/headingforge/core.py"},...
[{"answers": ["src/headingforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01973
tilekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_tile misses keys that differ only in case. It lives in src/tilekit/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tilekit layout. [tool_use kind=search name=...
23,889
1,194
49
56
false
synth:9072454344038523933@1
synth
[{"aliases": ["src/tilekit/limits.py", "limits.py"], "kind": "file", "value": "src/tilekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/beacon_2.csv", "beacon_2.csv"], "kind": "file", "value": "data/beacon_2.csv"}, {"aliases": ...
[{"answers": ["src/tilekit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01973-twin
tilekit
ctx_compress-s1-01973
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_tile misses keys that differ only in case. It lives in src/tilekit/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tilekit layout. [tool_use kind=search name=...
11,190
1,000
33
56
false
synth:9072454344038523933@1
synth
[{"aliases": ["src/tilekit/limits.py", "limits.py"], "kind": "file", "value": "src/tilekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/beacon_2.csv", "beacon_2.csv"], "kind": "file", "value": "data/beacon_2.csv"}, {"aliases": ...
[{"answers": ["src/tilekit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01974
clauseflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_revision raises DraftDriftError for bytes input in strict mode instead of decoding it. It lives in src/clauseflow/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappi...
50,042
2,502
78
99
false
synth:5777170398270050740@1.00743
synth
[{"aliases": ["src/clauseflow/compat.py", "compat.py"], "kind": "file", "value": "src/clauseflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/footnote_4.csv", "footnote_4.csv"], "kind": "file", "value": "data/footnote_4.csv"}, ...
[{"answers": ["src/clauseflow/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01974-twin
clauseflow
ctx_compress-s1-01974
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_revision raises DraftDriftError for bytes input in strict mode instead of decoding it. It lives in src/clauseflow/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappi...
31,403
1,570
58
99
false
synth:5777170398270050740@1.00743
synth
[{"aliases": ["src/clauseflow/compat.py", "compat.py"], "kind": "file", "value": "src/clauseflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/footnote_4.csv", "footnote_4.csv"], "kind": "file", "value": "data/footnote_4.csv"}, ...
[{"answers": ["src/clauseflow/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01975
principalworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim_all ignores the cap, so late attempts wait for minutes. It lives in src/principalworks/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the principalworks layout. [...
78,027
3,072
99
126
false
synth:1570428374989957824@2.0739
synth
[{"aliases": ["src/principalworks/io.py", "io.py"], "kind": "file", "value": "src/principalworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/session_4.csv", "session_4.csv"], "kind": "file", "value": "data/session_4.csv"}, {"aliases": ["src/p...
[{"answers": ["src/principalworks/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-01975-twin
principalworks
ctx_compress-s1-01975
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim_all ignores the cap, so late attempts wait for minutes. It lives in src/principalworks/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the principalworks layout. [...
52,693
2,634
61
126
false
synth:1570428374989957824@2.0739
synth
[{"aliases": ["src/principalworks/io.py", "io.py"], "kind": "file", "value": "src/principalworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/session_4.csv", "session_4.csv"], "kind": "file", "value": "data/session_4.csv"}, {"aliases": ["src/p...
[{"answers": ["src/principalworks/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-01976
waypointforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_waypoint double counts carried amounts when the period closes on a weekend. It lives in src/waypointforge/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the waypointfo...
86,724
3,072
111
124
false
synth:7143304397691820886@2.10145
synth
[{"aliases": ["src/waypointforge/cli.py", "cli.py"], "kind": "file", "value": "src/waypointforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/fix_3.csv", "fix_3.csv"], "kind": "file", "value": "data/fix_3.csv"}, {"aliases": ["src/waypointf...
[{"answers": ["src/waypointforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01976-twin
waypointforge
ctx_compress-s1-01976
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_waypoint double counts carried amounts when the period closes on a weekend. It lives in src/waypointforge/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the waypointfo...
49,519
2,475
69
124
false
synth:7143304397691820886@2.10145
synth
[{"aliases": ["src/waypointforge/cli.py", "cli.py"], "kind": "file", "value": "src/waypointforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/fix_3.csv", "fix_3.csv"], "kind": "file", "value": "data/fix_3.csv"}, {"aliases": ["src/waypointf...
[{"answers": ["src/waypointforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01977
slotkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster returns the upper bound for every value instead of clamping into the range. It lives in src/slotkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slo...
105,081
3,072
105
122
false
synth:4154560618530603359@2.11175
synth
[{"aliases": ["src/slotkit/rules.py", "rules.py"], "kind": "file", "value": "src/slotkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slotkit/format.py", "format.py"], "kind": "file", "value": "src/slotkit/format.py"}, {"aliases": [...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["report.py", "src/slotkit/report.py"],...
ctx_compress-s1-01977-twin
slotkit
ctx_compress-s1-01977
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster returns the upper bound for every value instead of clamping into the range. It lives in src/slotkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slo...
60,070
3,003
65
122
false
synth:4154560618530603359@2.11175
synth
[{"aliases": ["src/slotkit/rules.py", "rules.py"], "kind": "file", "value": "src/slotkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slotkit/format.py", "format.py"], "kind": "file", "value": "src/slotkit/format.py"}, {"aliases": [...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/slotkit/tables.py", "tables.py"],...
ctx_compress-s1-01978
sampleworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series returns the upper bound for every value instead of clamping into the range. It lives in src/sampleworks/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappin...
71,666
3,072
91
100
false
synth:5260132557915658733@2.30812
synth
[{"aliases": ["src/sampleworks/compat.py", "compat.py"], "kind": "file", "value": "src/sampleworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/sampleworks/cli.py", "cli.py"], "kind": "file", "value": "src/sampleworks/cli.py"}, ...
[{"answers": ["src/sampleworks/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01978-twin
sampleworks
ctx_compress-s1-01978
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series returns the upper bound for every value instead of clamping into the range. It lives in src/sampleworks/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappin...
39,434
1,971
55
100
false
synth:5260132557915658733@2.30812
synth
[{"aliases": ["src/sampleworks/compat.py", "compat.py"], "kind": "file", "value": "src/sampleworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/sampleworks/cli.py", "cli.py"], "kind": "file", "value": "src/sampleworks/cli.py"}, ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_format.py", "tests/test_format.p...
ctx_compress-s1-01979
lotworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_sku returns one item too few on every page. It lives in src/lotworks/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the lotworks layout. [tool_use kind=search...
19,394
1,000
46
55
false
synth:7447786588682557632@1
synth
[{"aliases": ["src/lotworks/limits.py", "limits.py"], "kind": "file", "value": "src/lotworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/lotworks/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01979-twin
lotworks
ctx_compress-s1-01979
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_sku returns one item too few on every page. It lives in src/lotworks/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the lotworks layout. [tool_use kind=search...
10,749
1,000
30
55
false
synth:7447786588682557632@1
synth
[{"aliases": ["src/lotworks/limits.py", "limits.py"], "kind": "file", "value": "src/lotworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/lotworks/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01980
slugdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_slug raises SnippetRangeError for bytes input in strict mode instead of decoding it. It lives in src/slugdesk/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slugdes...
34,992
1,749
68
81
false
synth:5461235977116818349@1
synth
[{"aliases": ["src/slugdesk/cli.py", "cli.py"], "kind": "file", "value": "src/slugdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/slugdesk/rules.py", "rules.py"], "kind": "file", "value": "src/slugdesk/rules.py"}, {"aliases": ["src/slugde...
[{"answers": ["src/slugdesk/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-01980-twin
slugdesk
ctx_compress-s1-01980
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_slug raises SnippetRangeError for bytes input in strict mode instead of decoding it. It lives in src/slugdesk/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slugdes...
18,996
1,000
42
81
false
synth:5461235977116818349@1
synth
[{"aliases": ["src/slugdesk/cli.py", "cli.py"], "kind": "file", "value": "src/slugdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/slugdesk/rules.py", "rules.py"], "kind": "file", "value": "src/slugdesk/rules.py"}, {"aliases": ["src/slugde...
[{"answers": ["src/slugdesk/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-01981
clauseforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_revision produces doubled dashes for consecutive separators. It lives in src/clauseforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the clauseforge layout. [t...
29,020
1,451
56
63
false
synth:6511490014992728364@1
synth
[{"aliases": ["src/clauseforge/store.py", "store.py"], "kind": "file", "value": "src/clauseforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/clauseforge/core.py", "core.py"], "kind": "file", "value": "src/clauseforge/core.py"}, {"a...
[{"answers": ["src/clauseforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01981-twin
clauseforge
ctx_compress-s1-01981
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_revision produces doubled dashes for consecutive separators. It lives in src/clauseforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the clauseforge layout. [t...
10,642
1,000
32
63
false
synth:6511490014992728364@1
synth
[{"aliases": ["src/clauseforge/store.py", "store.py"], "kind": "file", "value": "src/clauseforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/clauseforge/core.py", "core.py"], "kind": "file", "value": "src/clauseforge/core.py"}, {"a...
[{"answers": ["src/clauseforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01982
legflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_leg raises TileShapeError for bytes input in strict mode instead of decoding it. It lives in src/legflow/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the legfl...
46,622
2,331
85
96
false
synth:504790270335145937@1
synth
[{"aliases": ["src/legflow/report.py", "report.py"], "kind": "file", "value": "src/legflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/legflow/schema.py", "schema.py"], "kind": "file", "value": "src/legflow/schema.py"}, {"alias...
[{"answers": ["src/legflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01982-twin
legflow
ctx_compress-s1-01982
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_leg raises TileShapeError for bytes input in strict mode instead of decoding it. It lives in src/legflow/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the legfl...
21,481
1,074
47
96
false
synth:504790270335145937@1
synth
[{"aliases": ["src/legflow/report.py", "report.py"], "kind": "file", "value": "src/legflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/legflow/schema.py", "schema.py"], "kind": "file", "value": "src/legflow/schema.py"}, {"alias...
[{"answers": ["src/legflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01983
lotledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_lot ignores the cap, so late attempts wait for minutes. It lives in src/lotledger/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the lotledger layout. [tool_use kin...
63,147
3,072
108
123
false
synth:8696191849127155129@1.97748
synth
[{"aliases": ["src/lotledger/rules.py", "rules.py"], "kind": "file", "value": "src/lotledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/lotledger/compat.py", "compat.py"], "kind": "file", "value": "src/lotledger/compat.py"}, {"ali...
[{"answers": ["src/lotledger/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01983-twin
lotledger
ctx_compress-s1-01983
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_lot ignores the cap, so late attempts wait for minutes. It lives in src/lotledger/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the lotledger layout. [tool_use kin...
32,291
1,614
57
123
false
synth:8696191849127155129@1.97748
synth
[{"aliases": ["src/lotledger/rules.py", "rules.py"], "kind": "file", "value": "src/lotledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/lotledger/compat.py", "compat.py"], "kind": "file", "value": "src/lotledger/compat.py"}, {"ali...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/lotledger/compat.py"...
ctx_compress-s1-01984
cadencedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_shift double counts carried amounts when the period closes on a weekend. It lives in src/cadencedesk/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the cadencedesk la...
51,483
2,574
77
96
false
synth:7900491567787859395@1
synth
[{"aliases": ["src/cadencedesk/cli.py", "cli.py"], "kind": "file", "value": "src/cadencedesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/cadencedesk/limits.py", "limits.py"], "kind": "file", "value": "src/cadencedesk/limits.py"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/cadencedesk/audit.py"...
ctx_compress-s1-01984-twin
cadencedesk
ctx_compress-s1-01984
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_shift double counts carried amounts when the period closes on a weekend. It lives in src/cadencedesk/cli.py; `python -m pytest tests/test_cli.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the cadencedesk la...
30,931
1,546
49
96
false
synth:7900491567787859395@1
synth
[{"aliases": ["src/cadencedesk/cli.py", "cli.py"], "kind": "file", "value": "src/cadencedesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/cadencedesk/limits.py", "limits.py"], "kind": "file", "value": "src/cadencedesk/limits.py"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["cadence_2.csv", "data/cadence_2.csv"]...
ctx_compress-s1-01985
shipmentworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reserve_bin raises KeyError for a missing key instead of returning the default. It lives in src/shipmentworks/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the shipme...
35,752
1,787
74
93
false
synth:118370731292202380@1
synth
[{"aliases": ["src/shipmentworks/limits.py", "limits.py"], "kind": "file", "value": "src/shipmentworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/shipmentworks/audit.py", "audit.py"], "kind": "file", "value": "src/shipmentwork...
[{"answers": ["src/shipmentworks/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent'...
ctx_compress-s1-01985-twin
shipmentworks
ctx_compress-s1-01985
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reserve_bin raises KeyError for a missing key instead of returning the default. It lives in src/shipmentworks/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the shipme...
23,347
1,167
54
93
false
synth:118370731292202380@1
synth
[{"aliases": ["src/shipmentworks/limits.py", "limits.py"], "kind": "file", "value": "src/shipmentworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/shipmentworks/audit.py", "audit.py"], "kind": "file", "value": "src/shipmentwork...
[{"answers": ["src/shipmentworks/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent'...
ctx_compress-s1-01986
windowworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_booking raises ShiftBoundError for bytes input in strict mode instead of decoding it. It lives in src/windowworks/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping...
45,061
2,253
93
100
false
synth:5218033241533536989@1.2173
synth
[{"aliases": ["src/windowworks/format.py", "format.py"], "kind": "file", "value": "src/windowworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/windowworks/io.py", "io.py"], "kind": "file", "value": "src/windowworks/io.py"}, {"a...
[{"answers": ["src/windowworks/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01986-twin
windowworks
ctx_compress-s1-01986
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_booking raises ShiftBoundError for bytes input in strict mode instead of decoding it. It lives in src/windowworks/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping...
25,005
1,250
51
100
false
synth:5218033241533536989@1.2173
synth
[{"aliases": ["src/windowworks/format.py", "format.py"], "kind": "file", "value": "src/windowworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/windowworks/io.py", "io.py"], "kind": "file", "value": "src/windowworks/io.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-01987
glyphledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_glyph double counts carried amounts when the period closes on a weekend. It lives in src/glyphledger/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the glyph...
58,029
2,901
107
118
false
synth:1504376440001378142@1.2592
synth
[{"aliases": ["src/glyphledger/report.py", "report.py"], "kind": "file", "value": "src/glyphledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/stem_0.csv", "stem_0.csv"], "kind": "file", "value": "data/stem_0.csv"}, {"aliases"...
[{"answers": ["src/glyphledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01987-twin
glyphledger
ctx_compress-s1-01987
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_glyph double counts carried amounts when the period closes on a weekend. It lives in src/glyphledger/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the glyph...
29,634
1,481
63
118
false
synth:1504376440001378142@1.2592
synth
[{"aliases": ["src/glyphledger/report.py", "report.py"], "kind": "file", "value": "src/glyphledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/stem_0.csv", "stem_0.csv"], "kind": "file", "value": "data/stem_0.csv"}, {"aliases"...
[{"answers": ["src/glyphledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01988
stockdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_shipment_page raises BinRangeError for bytes input in strict mode instead of decoding it. It lives in src/stockdesk/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by map...
102,481
3,072
102
129
false
synth:2833168189949573295@1.77245
synth
[{"aliases": ["src/stockdesk/limits.py", "limits.py"], "kind": "file", "value": "src/stockdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/sku_2.csv", "sku_2.csv"], "kind": "file", "value": "data/sku_2.csv"}, {"aliases": ["dat...
[{"answers": ["src/stockdesk/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01988-twin
stockdesk
ctx_compress-s1-01988
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_shipment_page raises BinRangeError for bytes input in strict mode instead of decoding it. It lives in src/stockdesk/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by map...
55,572
2,778
68
129
false
synth:2833168189949573295@1.77245
synth
[{"aliases": ["src/stockdesk/limits.py", "limits.py"], "kind": "file", "value": "src/stockdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/sku_2.csv", "sku_2.csv"], "kind": "file", "value": "data/sku_2.csv"}, {"aliases": ["dat...
[{"answers": ["src/stockdesk/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01989
lexemeforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_cue leaves a trailing dash when the title ends in punctuation. It lives in src/lexemeforge/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the lexemeforge layo...
89,760
3,072
110
127
false
synth:5272035502384417402@1.5837
synth
[{"aliases": ["src/lexemeforge/compat.py", "compat.py"], "kind": "file", "value": "src/lexemeforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/lexemeforge/report.py", "report.py"], "kind": "file", "value": "src/lexemeforge/repo...
[{"answers": ["src/lexemeforge/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01989-twin
lexemeforge
ctx_compress-s1-01989
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_cue leaves a trailing dash when the title ends in punctuation. It lives in src/lexemeforge/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the lexemeforge layo...
42,371
2,118
59
127
false
synth:5272035502384417402@1.5837
synth
[{"aliases": ["src/lexemeforge/compat.py", "compat.py"], "kind": "file", "value": "src/lexemeforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/lexemeforge/report.py", "report.py"], "kind": "file", "value": "src/lexemeforge/repo...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/lexemeforge/compat.p...
ctx_compress-s1-01990
slugdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_caption double counts carried amounts when the period closes on a weekend. It lives in src/slugdesk/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slugdesk l...
116,095
3,072
124
133
false
synth:239006260841794126@1.97128
synth
[{"aliases": ["src/slugdesk/core.py", "core.py"], "kind": "file", "value": "src/slugdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/snippet_13.csv", "snippet_13.csv"], "kind": "file", "value": "data/snippet_13.csv"}, {"aliases": ["da...
[{"answers": ["src/slugdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most re...
ctx_compress-s1-01990-twin
slugdesk
ctx_compress-s1-01990
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_caption double counts carried amounts when the period closes on a weekend. It lives in src/slugdesk/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slugdesk l...
65,925
3,072
78
133
false
synth:239006260841794126@1.97128
synth
[{"aliases": ["src/slugdesk/core.py", "core.py"], "kind": "file", "value": "src/slugdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/snippet_13.csv", "snippet_13.csv"], "kind": "file", "value": "data/snippet_13.csv"}, {"aliases": ["da...
[{"answers": ["src/slugdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-01991
indexforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_revision_window raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/indexforge/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping ...
94,429
3,072
98
107
false
synth:5148752281318869528@1.54855
synth
[{"aliases": ["src/indexforge/audit.py", "audit.py"], "kind": "file", "value": "src/indexforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/footnote_10.csv", "footnote_10.csv"], "kind": "file", "value": "data/footnote_10.csv"}, {"a...
[{"answers": ["src/indexforge/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01991-twin
indexforge
ctx_compress-s1-01991
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_revision_window raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/indexforge/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping ...
49,361
2,468
62
107
false
synth:5148752281318869528@1.54855
synth
[{"aliases": ["src/indexforge/audit.py", "audit.py"], "kind": "file", "value": "src/indexforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/footnote_10.csv", "footnote_10.csv"], "kind": "file", "value": "data/footnote_10.csv"}, {"a...
[{"answers": ["src/indexforge/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01992
revisionkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_section rejects a range where lo equals hi, which is a valid single point range. It lives in src/revisionkit/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the re...
36,123
1,806
58
65
false
synth:7054897166154674835@1
synth
[{"aliases": ["src/revisionkit/core.py", "core.py"], "kind": "file", "value": "src/revisionkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/footnote_0.csv", "footnote_0.csv"], "kind": "file", "value": "data/footnote_0.csv"}, {"aliases"...
[{"answers": ["src/revisionkit/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01992-twin
revisionkit
ctx_compress-s1-01992
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_section rejects a range where lo equals hi, which is a valid single point range. It lives in src/revisionkit/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the re...
18,847
1,000
36
65
false
synth:7054897166154674835@1
synth
[{"aliases": ["src/revisionkit/core.py", "core.py"], "kind": "file", "value": "src/revisionkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/footnote_0.csv", "footnote_0.csv"], "kind": "file", "value": "data/footnote_0.csv"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-01993
probekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_bucket raises ValueError when a value itself contains an equals sign. It lives in src/probekit/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the probekit layout. [...
65,961
3,072
103
118
false
synth:8128199090628096352@2.17868
synth
[{"aliases": ["src/probekit/store.py", "store.py"], "kind": "file", "value": "src/probekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/probekit/audit.py", "audit.py"], "kind": "file", "value": "src/probekit/audit.py"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/sample_4.csv", "sample_4.csv"], ...
ctx_compress-s1-01993-twin
probekit
ctx_compress-s1-01993
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_bucket raises ValueError when a value itself contains an equals sign. It lives in src/probekit/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the probekit layout. [...
40,301
2,015
69
118
false
synth:8128199090628096352@2.17868
synth
[{"aliases": ["src/probekit/store.py", "store.py"], "kind": "file", "value": "src/probekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/probekit/audit.py", "audit.py"], "kind": "file", "value": "src/probekit/audit.py"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/span_2.csv", "span_2.csv"], "kin...
ctx_compress-s1-01994
bookingkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_booking grows linearly instead of doubling. It lives in src/bookingkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bookingkit layout. [tool_use kind=search...
99,812
3,072
115
138
false
synth:5682150737616798459@1.87027
synth
[{"aliases": ["src/bookingkit/rules.py", "rules.py"], "kind": "file", "value": "src/bookingkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/bookingkit/schema.py", "schema.py"], "kind": "file", "value": "src/bookingkit/schema.py"}, {...
[{"answers": ["src/bookingkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01994-twin
bookingkit
ctx_compress-s1-01994
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_booking grows linearly instead of doubling. It lives in src/bookingkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bookingkit layout. [tool_use kind=search...
65,117
3,072
75
138
false
synth:5682150737616798459@1.87027
synth
[{"aliases": ["src/bookingkit/rules.py", "rules.py"], "kind": "file", "value": "src/bookingkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/bookingkit/schema.py", "schema.py"], "kind": "file", "value": "src/bookingkit/schema.py"}, {...
[{"answers": ["src/bookingkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01995
waypointdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_tile returns one item too few on every page. It lives in src/waypointdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the waypointdesk layout. [tool_use kind=...
80,868
3,072
116
131
false
synth:8533662787362095677@2.16082
synth
[{"aliases": ["src/waypointdesk/store.py", "store.py"], "kind": "file", "value": "src/waypointdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/waypointdesk/report.py", "report.py"], "kind": "file", "value": "src/waypointdesk/report...
[{"answers": ["src/waypointdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01995-twin
waypointdesk
ctx_compress-s1-01995
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_tile returns one item too few on every page. It lives in src/waypointdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the waypointdesk layout. [tool_use kind=...
46,162
2,308
68
131
false
synth:8533662787362095677@2.16082
synth
[{"aliases": ["src/waypointdesk/store.py", "store.py"], "kind": "file", "value": "src/waypointdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/waypointdesk/report.py", "report.py"], "kind": "file", "value": "src/waypointdesk/report...
[{"answers": ["src/waypointdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01996
headingforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_clause rejects a range where lo equals hi, which is a valid single point range. It lives in src/headingforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the ...
40,885
2,044
74
93
false
synth:59990052163199689@1
synth
[{"aliases": ["src/headingforge/store.py", "store.py"], "kind": "file", "value": "src/headingforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliase...
[{"answers": ["src/headingforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01996-twin
headingforge
ctx_compress-s1-01996
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_clause rejects a range where lo equals hi, which is a valid single point range. It lives in src/headingforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the ...
22,734
1,136
42
93
false
synth:59990052163199689@1
synth
[{"aliases": ["src/headingforge/store.py", "store.py"], "kind": "file", "value": "src/headingforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliase...
[{"answers": ["src/headingforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01997
draftforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_heading drops override values of 0, False and empty string. It lives in src/draftforge/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the draftforge layout. [tool...
78,809
3,072
87
106
false
synth:496463749908475746@2.05158
synth
[{"aliases": ["src/draftforge/audit.py", "audit.py"], "kind": "file", "value": "src/draftforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/margin_5.csv", "margin_5.csv"], "kind": "file", "value": "data/margin_5.csv"}, {"aliases": ...
[{"answers": ["src/draftforge/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01997-twin
draftforge
ctx_compress-s1-01997
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_heading drops override values of 0, False and empty string. It lives in src/draftforge/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the draftforge layout. [tool...
47,223
2,361
59
106
false
synth:496463749908475746@2.05158
synth
[{"aliases": ["src/draftforge/audit.py", "audit.py"], "kind": "file", "value": "src/draftforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/margin_5.csv", "margin_5.csv"], "kind": "file", "value": "data/margin_5.csv"}, {"aliases": ...
[{"answers": ["src/draftforge/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01998
grantdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_principal rejects a range where lo equals hi, which is a valid single point range. It lives in src/grantdesk/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the gran...
88,437
3,072
98
111
false
synth:2731500467687397475@1.98625
synth
[{"aliases": ["src/grantdesk/core.py", "core.py"], "kind": "file", "value": "src/grantdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/grant_7.csv", "grant_7.csv"], "kind": "file", "value": "data/grant_7.csv"}, {"aliases": ["src/grant...
[{"answers": ["src/grantdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most r...
ctx_compress-s1-01998-twin
grantdesk
ctx_compress-s1-01998
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_principal rejects a range where lo equals hi, which is a valid single point range. It lives in src/grantdesk/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the gran...
51,580
2,579
62
111
false
synth:2731500467687397475@1.98625
synth
[{"aliases": ["src/grantdesk/core.py", "core.py"], "kind": "file", "value": "src/grantdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/grant_7.csv", "grant_7.csv"], "kind": "file", "value": "data/grant_7.csv"}, {"aliases": ["src/grant...
[{"answers": ["src/grantdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most r...
ctx_compress-s1-01999
grantdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_nonce skips the first page; page 1 returns what should be page 2. It lives in src/grantdesk/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the grantdesk layout. [tool_u...
86,079
3,072
105
118
false
synth:4780446650631614919@1.75875
synth
[{"aliases": ["src/grantdesk/io.py", "io.py"], "kind": "file", "value": "src/grantdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/grantdesk/limi...
[{"answers": ["src/grantdesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-01999-twin
grantdesk
ctx_compress-s1-01999
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_nonce skips the first page; page 1 returns what should be page 2. It lives in src/grantdesk/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the grantdesk layout. [tool_u...
38,386
1,919
57
118
false
synth:4780446650631614919@1.75875
synth
[{"aliases": ["src/grantdesk/io.py", "io.py"], "kind": "file", "value": "src/grantdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/grantdesk/limi...
[{"answers": ["src/grantdesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...