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-s1-01005-twin
tokendesk
ctx_compress-s1-01005
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_policy_batch skips the first page; page 1 returns what should be page 2. It lives in src/tokendesk/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 tokendesk ...
39,174
1,958
62
119
false
synth:8777427019572643191@1.6746
synth
[{"aliases": ["src/tokendesk/limits.py", "limits.py"], "kind": "file", "value": "src/tokendesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/scope_4.csv", "scope_4.csv"], "kind": "file", "value": "data/scope_4.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": ["data/nonce_5.csv", "nonce_5.csv"], "k...
ctx_compress-s1-01006
headingkit
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_draft_safe grows linearly instead of doubling. It lives in src/headingkit/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 headingkit layout. [tool_use kind...
41,632
2,081
75
88
false
synth:6080922191537727625@1
synth
[{"aliases": ["src/headingkit/format.py", "format.py"], "kind": "file", "value": "src/headingkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/headingkit/audit.py", "audit.py"], "kind": "file", "value": "src/headingkit/audit.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": ["clause_2.csv", "data/clause_2.csv"], ...
ctx_compress-s1-01006-twin
headingkit
ctx_compress-s1-01006
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_draft_safe grows linearly instead of doubling. It lives in src/headingkit/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 headingkit layout. [tool_use kind...
25,251
1,262
45
88
false
synth:6080922191537727625@1
synth
[{"aliases": ["src/headingkit/format.py", "format.py"], "kind": "file", "value": "src/headingkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/headingkit/audit.py", "audit.py"], "kind": "file", "value": "src/headingkit/audit.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": ["src/headingkit/store.py", "store.py"]...
ctx_compress-s1-01007
creditflow
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_invoice returns one item too few on every page. It lives in src/creditflow/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 creditflow layout. [tool_use kind=se...
76,693
3,072
116
139
false
synth:8679702137076173993@1.69595
synth
[{"aliases": ["src/creditflow/report.py", "report.py"], "kind": "file", "value": "src/creditflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/creditflow/schema.py", "schema.py"], "kind": "file", "value": "src/creditflow/schema.p...
[{"answers": ["src/creditflow/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 m...
ctx_compress-s1-01007-twin
creditflow
ctx_compress-s1-01007
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_invoice returns one item too few on every page. It lives in src/creditflow/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 creditflow layout. [tool_use kind=se...
40,578
2,028
64
139
false
synth:8679702137076173993@1.69595
synth
[{"aliases": ["src/creditflow/report.py", "report.py"], "kind": "file", "value": "src/creditflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/creditflow/schema.py", "schema.py"], "kind": "file", "value": "src/creditflow/schema.p...
[{"answers": ["src/creditflow/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 m...
ctx_compress-s1-01008
creditkit
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_refund_once produces doubled dashes for consecutive separators. It lives in src/creditkit/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 creditkit layout. [too...
41,277
2,063
84
103
false
synth:3139829404114371199@1.48628
synth
[{"aliases": ["src/creditkit/compat.py", "compat.py"], "kind": "file", "value": "src/creditkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliase...
[{"answers": ["src/creditkit/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 mo...
ctx_compress-s1-01008-twin
creditkit
ctx_compress-s1-01008
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_refund_once produces doubled dashes for consecutive separators. It lives in src/creditkit/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 creditkit layout. [too...
30,636
1,531
51
103
false
synth:3139829404114371199@1.48628
synth
[{"aliases": ["src/creditkit/compat.py", "compat.py"], "kind": "file", "value": "src/creditkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"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": ["data/refund_3.csv", "refund_3.csv"], ...
ctx_compress-s1-01009
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] anchor_revision raises KeyError for a missing key instead of returning the default. It lives in src/clausekit/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 clause...
76,470
3,072
101
124
false
synth:6583089645106133421@1.64538
synth
[{"aliases": ["src/clausekit/report.py", "report.py"], "kind": "file", "value": "src/clausekit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/heading_5.csv", "heading_5.csv"], "kind": "file", "value": "data/heading_5.csv"}, {"ali...
[{"answers": ["src/clausekit/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 mo...
ctx_compress-s1-01009-twin
clausekit
ctx_compress-s1-01009
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 raises KeyError for a missing key instead of returning the default. It lives in src/clausekit/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 clause...
49,503
2,475
65
124
false
synth:6583089645106133421@1.64538
synth
[{"aliases": ["src/clausekit/report.py", "report.py"], "kind": "file", "value": "src/clausekit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/heading_5.csv", "heading_5.csv"], "kind": "file", "value": "data/heading_5.csv"}, {"ali...
[{"answers": ["src/clausekit/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 mo...
ctx_compress-s1-01010
stemkit
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_slug raises KeyError for a missing key instead of returning the default. It lives in src/stemkit/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 stemkit layou...
99,897
3,072
117
148
false
synth:5948761599273388026@1.8546
synth
[{"aliases": ["src/stemkit/core.py", "core.py"], "kind": "file", "value": "src/stemkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/glyph_1.csv", "glyph_1.csv"], "kind": "file", "value": "data/glyph_1.csv"}, {"aliases": ["data/cue_10.c...
[{"answers": ["src/stemkit/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 rec...
ctx_compress-s1-01010-twin
stemkit
ctx_compress-s1-01010
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_slug raises KeyError for a missing key instead of returning the default. It lives in src/stemkit/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 stemkit layou...
58,836
2,941
69
148
false
synth:5948761599273388026@1.8546
synth
[{"aliases": ["src/stemkit/core.py", "core.py"], "kind": "file", "value": "src/stemkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/glyph_1.csv", "glyph_1.csv"], "kind": "file", "value": "data/glyph_1.csv"}, {"aliases": ["data/cue_10.c...
[{"answers": ["src/stemkit/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 rec...
ctx_compress-s1-01011
sessiondesk
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_policy raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/sessiondesk/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 se...
48,962
2,448
70
85
false
synth:5974996805534481387@1
synth
[{"aliases": ["src/sessiondesk/limits.py", "limits.py"], "kind": "file", "value": "src/sessiondesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/sessiondesk/report.py", "report.py"], "kind": "file", "value": "src/sessiondesk/repo...
[{"answers": ["src/sessiondesk/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-s1-01011-twin
sessiondesk
ctx_compress-s1-01011
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_policy raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/sessiondesk/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 se...
23,276
1,163
40
85
false
synth:5974996805534481387@1
synth
[{"aliases": ["src/sessiondesk/limits.py", "limits.py"], "kind": "file", "value": "src/sessiondesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/sessiondesk/report.py", "report.py"], "kind": "file", "value": "src/sessiondesk/repo...
[{"answers": ["src/sessiondesk/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 ...
ctx_compress-s1-01012
traceworks
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] smooth_gauge_strict keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/traceworks/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 trace...
54,637
2,731
79
88
false
synth:7479005543410392283@1
synth
[{"aliases": ["src/traceworks/limits.py", "limits.py"], "kind": "file", "value": "src/traceworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/traceworks/tables.py", "tables.py"], "kind": "file", "value": "src/traceworks/tables.p...
[{"answers": ["src/traceworks/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-s1-01012-twin
traceworks
ctx_compress-s1-01012
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] smooth_gauge_strict keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/traceworks/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 trace...
24,905
1,245
41
88
false
synth:7479005543410392283@1
synth
[{"aliases": ["src/traceworks/limits.py", "limits.py"], "kind": "file", "value": "src/traceworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/traceworks/tables.py", "tables.py"], "kind": "file", "value": "src/traceworks/tables.p...
[{"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": ["rules.py", "src/traceworks/rules.py"]...
ctx_compress-s1-01013
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] balance_cadence_batch raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/bookingkit/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...
87,552
3,072
113
128
false
synth:6936347983759131603@2.03262
synth
[{"aliases": ["src/bookingkit/compat.py", "compat.py"], "kind": "file", "value": "src/bookingkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/bookingkit/limits.py", "limits.py"], "kind": "file", "value": "src/bookingkit/limits.p...
[{"answers": ["src/bookingkit/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-s1-01013-twin
bookingkit
ctx_compress-s1-01013
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_cadence_batch raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/bookingkit/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...
43,381
2,169
59
128
false
synth:6936347983759131603@2.03262
synth
[{"aliases": ["src/bookingkit/compat.py", "compat.py"], "kind": "file", "value": "src/bookingkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/bookingkit/limits.py", "limits.py"], "kind": "file", "value": "src/bookingkit/limits.p...
[{"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_3.csv", "data/cadence_3.csv"]...
ctx_compress-s1-01014
shiftdesk
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_shift grows linearly instead of doubling. It lives in src/shiftdesk/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 shiftdesk layout. [tool_use kind=search name=B...
90,744
3,072
111
138
false
synth:6063199465743703423@2.04853
synth
[{"aliases": ["src/shiftdesk/cli.py", "cli.py"], "kind": "file", "value": "src/shiftdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/shiftdesk/format.py", "format.py"], "kind": "file", "value": "src/shiftdesk/format.py"}, {"aliases": ["dat...
[{"answers": ["src/shiftdesk/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 rec...
ctx_compress-s1-01014-twin
shiftdesk
ctx_compress-s1-01014
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_shift grows linearly instead of doubling. It lives in src/shiftdesk/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 shiftdesk layout. [tool_use kind=search name=B...
50,201
2,510
63
138
false
synth:6063199465743703423@2.04853
synth
[{"aliases": ["src/shiftdesk/cli.py", "cli.py"], "kind": "file", "value": "src/shiftdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/shiftdesk/format.py", "format.py"], "kind": "file", "value": "src/shiftdesk/format.py"}, {"aliases": ["dat...
[{"answers": ["src/shiftdesk/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 rec...
ctx_compress-s1-01015
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_clause double counts carried amounts when the period closes on a weekend. It lives in src/clauseflow/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 clause...
26,351
1,317
42
53
false
synth:691177456982894223@1
synth
[{"aliases": ["src/clauseflow/format.py", "format.py"], "kind": "file", "value": "src/clauseflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/index_3.csv", "index_3.csv"], "kind": "file", "value": "data/index_3.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": ["format.py", "src/clauseflow/format.py...
ctx_compress-s1-01015-twin
clauseflow
ctx_compress-s1-01015
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 double counts carried amounts when the period closes on a weekend. It lives in src/clauseflow/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 clause...
12,708
1,000
24
53
false
synth:691177456982894223@1
synth
[{"aliases": ["src/clauseflow/format.py", "format.py"], "kind": "file", "value": "src/clauseflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/index_3.csv", "index_3.csv"], "kind": "file", "value": "data/index_3.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": ["test_io.py", "tests/test_io.py"], "ki...
ctx_compress-s1-01016
slugledger
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] dedupe_snippet returns the upper bound for every value instead of clamping into the range. It lives in src/slugledger/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 ...
60,199
3,009
87
98
false
synth:5164607138338483819@1.1502
synth
[{"aliases": ["src/slugledger/rules.py", "rules.py"], "kind": "file", "value": "src/slugledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slugledger/audit.py", "audit.py"], "kind": "file", "value": "src/slugledger/audit.py"}, {"al...
[{"answers": ["src/slugledger/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-s1-01016-twin
slugledger
ctx_compress-s1-01016
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] dedupe_snippet returns the upper bound for every value instead of clamping into the range. It lives in src/slugledger/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 ...
30,865
1,543
49
98
false
synth:5164607138338483819@1.1502
synth
[{"aliases": ["src/slugledger/rules.py", "rules.py"], "kind": "file", "value": "src/slugledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slugledger/audit.py", "audit.py"], "kind": "file", "value": "src/slugledger/audit.py"}, {"al...
[{"answers": ["src/slugledger/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-s1-01017
quotaledger
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_calendar raises QuotaShapeError for bytes input in strict mode instead of decoding it. It lives in src/quotaledger/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 qu...
27,654
1,382
74
79
false
synth:6477898575101905655@1
synth
[{"aliases": ["src/quotaledger/io.py", "io.py"], "kind": "file", "value": "src/quotaledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/quota_0.csv", "quota_0.csv"], "kind": "file", "value": "data/quota_0.csv"}, {"aliases": ["src/quotaledger/l...
[{"answers": ["src/quotaledger/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-s1-01017-twin
quotaledger
ctx_compress-s1-01017
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_calendar raises QuotaShapeError for bytes input in strict mode instead of decoding it. It lives in src/quotaledger/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 qu...
13,986
1,000
36
79
false
synth:6477898575101905655@1
synth
[{"aliases": ["src/quotaledger/io.py", "io.py"], "kind": "file", "value": "src/quotaledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/quota_0.csv", "quota_0.csv"], "kind": "file", "value": "data/quota_0.csv"}, {"aliases": ["src/quotaledger/l...
[{"answers": ["src/quotaledger/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-s1-01018
balanceforge
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] allocate_balance_page returns one item too few on every page. It lives in src/balanceforge/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 balanceforge layout. [too...
75,969
3,072
121
130
false
synth:1025526150683488933@1.85462
synth
[{"aliases": ["src/balanceforge/report.py", "report.py"], "kind": "file", "value": "src/balanceforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/balanceforge/tables.py", "tables.py"], "kind": "file", "value": "src/balanceforge/...
[{"answers": ["src/balanceforge/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-s1-01018-twin
balanceforge
ctx_compress-s1-01018
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] allocate_balance_page returns one item too few on every page. It lives in src/balanceforge/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 balanceforge layout. [too...
38,134
1,906
59
130
false
synth:1025526150683488933@1.85462
synth
[{"aliases": ["src/balanceforge/report.py", "report.py"], "kind": "file", "value": "src/balanceforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/balanceforge/tables.py", "tables.py"], "kind": "file", "value": "src/balanceforge/...
[{"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/balanceforge/store.py", "store.py...
ctx_compress-s1-01019
tokenflow
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_nonce_safe rejects a range where lo equals hi, which is a valid single point range. It lives in src/tokenflow/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...
91,266
3,072
121
134
false
synth:6213040985689057642@2.35885
synth
[{"aliases": ["src/tokenflow/compat.py", "compat.py"], "kind": "file", "value": "src/tokenflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/tokenflow/cli.py", "cli.py"], "kind": "file", "value": "src/tokenflow/cli.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": ["report.py", "src/tokenflow/report.py"...
ctx_compress-s1-01019-twin
tokenflow
ctx_compress-s1-01019
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_nonce_safe rejects a range where lo equals hi, which is a valid single point range. It lives in src/tokenflow/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...
50,351
2,517
67
134
false
synth:6213040985689057642@2.35885
synth
[{"aliases": ["src/tokenflow/compat.py", "compat.py"], "kind": "file", "value": "src/tokenflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/tokenflow/cli.py", "cli.py"], "kind": "file", "value": "src/tokenflow/cli.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": ["rules.py", "src/tokenflow/rules.py"],...
ctx_compress-s1-01020
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] collate_section keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/marginforge/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 marginforge la...
72,376
3,072
114
127
false
synth:7548928695929984409@1.24545
synth
[{"aliases": ["src/marginforge/cli.py", "cli.py"], "kind": "file", "value": "src/marginforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/marginforge/report.py", "report.py"], "kind": "file", "value": "src/marginforge/report.py"}, {"aliases...
[{"answers": ["src/marginforge/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 r...
ctx_compress-s1-01020-twin
marginforge
ctx_compress-s1-01020
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] collate_section keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/marginforge/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 marginforge la...
38,264
1,913
64
127
false
synth:7548928695929984409@1.24545
synth
[{"aliases": ["src/marginforge/cli.py", "cli.py"], "kind": "file", "value": "src/marginforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/marginforge/report.py", "report.py"], "kind": "file", "value": "src/marginforge/report.py"}, {"aliases...
[{"answers": ["src/marginforge/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 r...
ctx_compress-s1-01021
samplekit
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] clip_sample skips the first page; page 1 returns what should be page 2. It lives in src/samplekit/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 samplekit layout. ...
20,377
1,018
43
50
false
synth:4792629204190769126@1
synth
[{"aliases": ["src/samplekit/limits.py", "limits.py"], "kind": "file", "value": "src/samplekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/counter_0.csv", "counter_0.csv"], "kind": "file", "value": "data/counter_0.csv"}, {"ali...
[{"answers": ["src/samplekit/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-s1-01021-twin
samplekit
ctx_compress-s1-01021
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] clip_sample skips the first page; page 1 returns what should be page 2. It lives in src/samplekit/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 samplekit layout. ...
9,092
1,000
27
50
false
synth:4792629204190769126@1
synth
[{"aliases": ["src/samplekit/limits.py", "limits.py"], "kind": "file", "value": "src/samplekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/counter_0.csv", "counter_0.csv"], "kind": "file", "value": "data/counter_0.csv"}, {"ali...
[{"answers": ["src/samplekit/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-s1-01022
captionkit
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] truncate_glyph skips the first page; page 1 returns what should be page 2. It lives in src/captionkit/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 captionkit layou...
43,011
2,150
73
82
false
synth:3352149043273147475@1
synth
[{"aliases": ["src/captionkit/audit.py", "audit.py"], "kind": "file", "value": "src/captionkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ...
[{"answers": ["src/captionkit/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-s1-01022-twin
captionkit
ctx_compress-s1-01022
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] truncate_glyph skips the first page; page 1 returns what should be page 2. It lives in src/captionkit/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 captionkit layou...
23,262
1,163
45
82
false
synth:3352149043273147475@1
synth
[{"aliases": ["src/captionkit/audit.py", "audit.py"], "kind": "file", "value": "src/captionkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ...
[{"answers": ["src/captionkit/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-s1-01023
probeworks
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] bin_sample_batch averages one value too many once the window is full. It lives in src/probeworks/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 probeworks layout. [t...
22,881
1,144
62
69
false
synth:7420646102845612994@1
synth
[{"aliases": ["src/probeworks/audit.py", "audit.py"], "kind": "file", "value": "src/probeworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/probeworks/schema.py", "schema.py"], "kind": "file", "value": "src/probeworks/schema.py"}, {...
[{"answers": ["src/probeworks/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-s1-01023-twin
probeworks
ctx_compress-s1-01023
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] bin_sample_batch averages one value too many once the window is full. It lives in src/probeworks/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 probeworks layout. [t...
10,522
1,000
38
69
false
synth:7420646102845612994@1
synth
[{"aliases": ["src/probeworks/audit.py", "audit.py"], "kind": "file", "value": "src/probeworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/probeworks/schema.py", "schema.py"], "kind": "file", "value": "src/probeworks/schema.py"}, {...
[{"answers": ["src/probeworks/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-s1-01024
nonceflow
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_claim_safe raises KeyError for a missing key instead of returning the default. It lives in src/nonceflow/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 nonceflo...
53,186
2,659
99
118
false
synth:8911076009061862856@1.32472
synth
[{"aliases": ["src/nonceflow/core.py", "core.py"], "kind": "file", "value": "src/nonceflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/nonceflow/tables.py", "tables.py"], "kind": "file", "value": "src/nonceflow/tables.py"}, {"aliases":...
[{"answers": ["src/nonceflow/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-s1-01024-twin
nonceflow
ctx_compress-s1-01024
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_claim_safe raises KeyError for a missing key instead of returning the default. It lives in src/nonceflow/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 nonceflo...
29,036
1,451
59
118
false
synth:8911076009061862856@1.32472
synth
[{"aliases": ["src/nonceflow/core.py", "core.py"], "kind": "file", "value": "src/nonceflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/nonceflow/tables.py", "tables.py"], "kind": "file", "value": "src/nonceflow/tables.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/nonceflow/tables.py", "tables.py"...
ctx_compress-s1-01025
clauseworks
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] collate_clause_page double counts carried amounts when the period closes on a weekend. It lives in src/clauseworks/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 clausewor...
65,159
3,072
90
107
false
synth:7952620793224991067@1.2206
synth
[{"aliases": ["src/clauseworks/io.py", "io.py"], "kind": "file", "value": "src/clauseworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/clauseworks/store.py", "store.py"], "kind": "file", "value": "src/clauseworks/store.py"}, {"aliases": ["src/...
[{"answers": ["src/clauseworks/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-s1-01025-twin
clauseworks
ctx_compress-s1-01025
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] collate_clause_page double counts carried amounts when the period closes on a weekend. It lives in src/clauseworks/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 clausewor...
40,313
2,015
62
107
false
synth:7952620793224991067@1.2206
synth
[{"aliases": ["src/clauseworks/io.py", "io.py"], "kind": "file", "value": "src/clauseworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/clauseworks/store.py", "store.py"], "kind": "file", "value": "src/clauseworks/store.py"}, {"aliases": ["src/...
[{"answers": ["src/clauseworks/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-s1-01026
invoiceflow
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_refund_batch returns one item too few on every page. It lives in src/invoiceflow/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 invoiceflow layout. [tool_use kind=s...
26,872
1,343
33
42
false
synth:1768266384903620649@1
synth
[{"aliases": ["src/invoiceflow/cli.py", "cli.py"], "kind": "file", "value": "src/invoiceflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/invoiceflow/report.py", "report.py"], "kind": "file", "value": "src/invoiceflow/report.py"}, {"aliases...
[{"answers": ["src/invoiceflow/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 r...
ctx_compress-s1-01027
quotadesk
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] split_roster double counts carried amounts when the period closes on a weekend. It lives in src/quotadesk/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 quotadesk ...
62,720
3,072
100
107
false
synth:637021990252864816@1.44613
synth
[{"aliases": ["src/quotadesk/report.py", "report.py"], "kind": "file", "value": "src/quotadesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/quotadesk/schema.py", "schema.py"], "kind": "file", "value": "src/quotadesk/schema.py"},...
[{"answers": ["src/quotadesk/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 mo...
ctx_compress-s1-01027-twin
quotadesk
ctx_compress-s1-01027
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] split_roster double counts carried amounts when the period closes on a weekend. It lives in src/quotadesk/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 quotadesk ...
35,575
1,778
64
107
false
synth:637021990252864816@1.44613
synth
[{"aliases": ["src/quotadesk/report.py", "report.py"], "kind": "file", "value": "src/quotadesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/quotadesk/schema.py", "schema.py"], "kind": "file", "value": "src/quotadesk/schema.py"},...
[{"answers": ["src/quotadesk/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 mo...
ctx_compress-s1-01028
nonceforge
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] mint_session returns the upper bound for every value instead of clamping into the range. It lives in src/nonceforge/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 ...
86,071
3,072
114
145
false
synth:4296039987213435061@2.0672
synth
[{"aliases": ["src/nonceforge/limits.py", "limits.py"], "kind": "file", "value": "src/nonceforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/nonceforge/io.py", "io.py"], "kind": "file", "value": "src/nonceforge/io.py"}, {"alias...
[{"answers": ["src/nonceforge/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-s1-01028-twin
nonceforge
ctx_compress-s1-01028
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] mint_session returns the upper bound for every value instead of clamping into the range. It lives in src/nonceforge/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 ...
57,008
2,850
78
145
false
synth:4296039987213435061@2.0672
synth
[{"aliases": ["src/nonceforge/limits.py", "limits.py"], "kind": "file", "value": "src/nonceforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/nonceforge/io.py", "io.py"], "kind": "file", "value": "src/nonceforge/io.py"}, {"alias...
[{"answers": ["src/nonceforge/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 m...
ctx_compress-s1-01029
spanflow
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] clip_counter double counts carried amounts when the period closes on a weekend. It lives in src/spanflow/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 spanflow layout. [t...
69,536
3,072
95
108
false
synth:3451307117447162846@1.44487
synth
[{"aliases": ["src/spanflow/io.py", "io.py"], "kind": "file", "value": "src/spanflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/spanflow/store.py", "store.py"], "kind": "file", "value": "src/spanflow/store.py"}, {"aliases": ["data/trace_4.csv...
[{"answers": ["src/spanflow/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 recent...
ctx_compress-s1-01029-twin
spanflow
ctx_compress-s1-01029
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] clip_counter double counts carried amounts when the period closes on a weekend. It lives in src/spanflow/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 spanflow layout. [t...
34,905
1,745
61
108
false
synth:3451307117447162846@1.44487
synth
[{"aliases": ["src/spanflow/io.py", "io.py"], "kind": "file", "value": "src/spanflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/spanflow/store.py", "store.py"], "kind": "file", "value": "src/spanflow/store.py"}, {"aliases": ["data/trace_4.csv...
[{"answers": ["src/spanflow/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 recent...
ctx_compress-s1-01030
creditledger
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_credit returns one item too few on every page. It lives in src/creditledger/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 creditledger layout. [tool_use kind...
55,613
2,780
93
116
false
synth:7309598614469115080@1.27015
synth
[{"aliases": ["src/creditledger/audit.py", "audit.py"], "kind": "file", "value": "src/creditledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/creditledger/schema.py", "schema.py"], "kind": "file", "value": "src/creditledger/schema...
[{"answers": ["src/creditledger/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 m...
ctx_compress-s1-01030-twin
creditledger
ctx_compress-s1-01030
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_credit returns one item too few on every page. It lives in src/creditledger/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 creditledger layout. [tool_use kind...
33,501
1,675
53
116
false
synth:7309598614469115080@1.27015
synth
[{"aliases": ["src/creditledger/audit.py", "audit.py"], "kind": "file", "value": "src/creditledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/creditledger/schema.py", "schema.py"], "kind": "file", "value": "src/creditledger/schema...
[{"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/creditledger/audit.py...
ctx_compress-s1-01031
quotaflow
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_strict returns the upper bound for every value instead of clamping into the range. It lives in src/quotaflow/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 mapp...
25,561
1,278
42
49
false
synth:2897518960272598266@1
synth
[{"aliases": ["src/quotaflow/format.py", "format.py"], "kind": "file", "value": "src/quotaflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/window_1.csv", "window_1.csv"], "kind": "file", "value": "data/window_1.csv"}, {"aliase...
[{"answers": ["src/quotaflow/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 mo...
ctx_compress-s1-01031-twin
quotaflow
ctx_compress-s1-01031
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_strict returns the upper bound for every value instead of clamping into the range. It lives in src/quotaflow/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 mapp...
23,893
1,194
28
49
false
synth:2897518960272598266@1
synth
[{"aliases": ["src/quotaflow/format.py", "format.py"], "kind": "file", "value": "src/quotaflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/window_1.csv", "window_1.csv"], "kind": "file", "value": "data/window_1.csv"}, {"aliase...
[{"answers": ["no"], "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 recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_compat.py", "tests/test_compat.p...
ctx_compress-s1-01032
creditflow
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] allocate_tariff_page averages one value too many once the window is full. It lives in src/creditflow/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 creditflow layout...
25,491
1,274
44
55
false
synth:6280205603835037883@1
synth
[{"aliases": ["src/creditflow/rules.py", "rules.py"], "kind": "file", "value": "src/creditflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/receipt_1.csv", "receipt_1.csv"], "kind": "file", "value": "data/receipt_1.csv"}, {"aliases...
[{"answers": ["src/creditflow/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-s1-01032-twin
creditflow
ctx_compress-s1-01032
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] allocate_tariff_page averages one value too many once the window is full. It lives in src/creditflow/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 creditflow layout...
11,230
1,000
27
55
false
synth:6280205603835037883@1
synth
[{"aliases": ["src/creditflow/rules.py", "rules.py"], "kind": "file", "value": "src/creditflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/receipt_1.csv", "receipt_1.csv"], "kind": "file", "value": "data/receipt_1.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": ["rules.py", "src/creditflow/rules.py"]...
ctx_compress-s1-01033
bookingforge
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 leaves a trailing dash when the title ends in punctuation. It lives in src/bookingforge/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 bookingforge layout. ...
84,180
3,072
98
117
false
synth:4260946071899851718@2.13548
synth
[{"aliases": ["src/bookingforge/io.py", "io.py"], "kind": "file", "value": "src/bookingforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/bookingforge/rules.py", "rules.py"], "kind": "file", "value": "src/bookingforge/rules.py"}, {"aliases": ["...
[{"answers": ["src/bookingforge/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 re...
ctx_compress-s1-01033-twin
bookingforge
ctx_compress-s1-01033
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 leaves a trailing dash when the title ends in punctuation. It lives in src/bookingforge/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 bookingforge layout. ...
48,845
2,442
59
117
false
synth:4260946071899851718@2.13548
synth
[{"aliases": ["src/bookingforge/io.py", "io.py"], "kind": "file", "value": "src/bookingforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/bookingforge/rules.py", "rules.py"], "kind": "file", "value": "src/bookingforge/rules.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": ["schema.py", "src/bookingforge/schema....
ctx_compress-s1-01034
bearingflow
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] simplify_tile_once raises ValueError when a value itself contains an equals sign. It lives in src/bearingflow/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 bearingf...
62,769
3,072
97
118
false
synth:8136535596451691801@1.51058
synth
[{"aliases": ["src/bearingflow/audit.py", "audit.py"], "kind": "file", "value": "src/bearingflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/fix_3.csv", "fix_3.csv"], "kind": "file", "value": "data/fix_3.csv"}, {"aliases": ["src/b...
[{"answers": ["src/bearingflow/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 mo...
ctx_compress-s1-01034-twin
bearingflow
ctx_compress-s1-01034
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] simplify_tile_once raises ValueError when a value itself contains an equals sign. It lives in src/bearingflow/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 bearingf...
37,009
1,850
63
118
false
synth:8136535596451691801@1.51058
synth
[{"aliases": ["src/bearingflow/audit.py", "audit.py"], "kind": "file", "value": "src/bearingflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/fix_3.csv", "fix_3.csv"], "kind": "file", "value": "data/fix_3.csv"}, {"aliases": ["src/b...
[{"answers": ["src/bearingflow/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 mo...
ctx_compress-s1-01035
tileforge
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] interpolate_zone returns the upper bound for every value instead of clamping into the range. It lives in src/tileforge/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...
37,066
1,853
59
74
false
synth:7656303608688261977@1
synth
[{"aliases": ["src/tileforge/rules.py", "rules.py"], "kind": "file", "value": "src/tileforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/corridor_2.csv", "corridor_2.csv"], "kind": "file", "value": "data/corridor_2.csv"}, {"aliase...
[{"answers": ["src/tileforge/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-s1-01035-twin
tileforge
ctx_compress-s1-01035
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] interpolate_zone returns the upper bound for every value instead of clamping into the range. It lives in src/tileforge/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...
24,960
1,248
35
74
false
synth:7656303608688261977@1
synth
[{"aliases": ["src/tileforge/rules.py", "rules.py"], "kind": "file", "value": "src/tileforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/corridor_2.csv", "corridor_2.csv"], "kind": "file", "value": "data/corridor_2.csv"}, {"aliase...
[{"answers": ["src/tileforge/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-s1-01036
sessionflow
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] mint_session_window leaves a trailing dash when the title ends in punctuation. It lives in src/sessionflow/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 sessionfl...
49,575
2,478
87
98
false
synth:546846986910281840@1
synth
[{"aliases": ["src/sessionflow/format.py", "format.py"], "kind": "file", "value": "src/sessionflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/policy_1.csv", "policy_1.csv"], "kind": "file", "value": "data/policy_1.csv"}, {"al...
[{"answers": ["src/sessionflow/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-s1-01036-twin
sessionflow
ctx_compress-s1-01036
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] mint_session_window leaves a trailing dash when the title ends in punctuation. It lives in src/sessionflow/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 sessionfl...
31,163
1,558
55
98
false
synth:546846986910281840@1
synth
[{"aliases": ["src/sessionflow/format.py", "format.py"], "kind": "file", "value": "src/sessionflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/policy_1.csv", "policy_1.csv"], "kind": "file", "value": "data/policy_1.csv"}, {"al...
[{"answers": ["src/sessionflow/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-s1-01037
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] simplify_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 waypoin...
63,861
3,072
83
100
false
synth:7548079742767903572@1.13478
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": ["src/waypointforge/report.py", "report.py"], "kind": "file", "value": "src/waypointforge/report.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": ["io.py", "src/waypointforge/io.py"], "...
ctx_compress-s1-01037-twin
waypointforge
ctx_compress-s1-01037
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] simplify_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 waypoin...
38,670
1,933
47
100
false
synth:7548079742767903572@1.13478
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": ["src/waypointforge/report.py", "report.py"], "kind": "file", "value": "src/waypointforge/report.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": ["core.py", "src/waypointforge/core.py"...
ctx_compress-s1-01038
bearingdesk
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_strict raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/bearingdesk/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 ...
25,370
1,268
50
55
false
synth:512432266462993978@1
synth
[{"aliases": ["src/bearingdesk/audit.py", "audit.py"], "kind": "file", "value": "src/bearingdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/bearingdesk/report.py", "report.py"], "kind": "file", "value": "src/bearingdesk/report.py"...
[{"answers": ["src/bearingdesk/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 mo...
ctx_compress-s1-01038-twin
bearingdesk
ctx_compress-s1-01038
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_strict raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/bearingdesk/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 ...
9,343
1,000
27
55
false
synth:512432266462993978@1
synth
[{"aliases": ["src/bearingdesk/audit.py", "audit.py"], "kind": "file", "value": "src/bearingdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/bearingdesk/report.py", "report.py"], "kind": "file", "value": "src/bearingdesk/report.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": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-01039
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_leg drops override values of 0, False and empty string. It lives in src/waypointforge/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 waypointforge layout. [t...
27,981
1,399
48
57
false
synth:8581847523672130901@1
synth
[{"aliases": ["src/waypointforge/limits.py", "limits.py"], "kind": "file", "value": "src/waypointforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/waypointforge/cli.py", "cli.py"], "kind": "file", "value": "src/waypointforge/cl...
[{"answers": ["src/waypointforge/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-s1-01039-twin
waypointforge
ctx_compress-s1-01039
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 drops override values of 0, False and empty string. It lives in src/waypointforge/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 waypointforge layout. [t...
12,388
1,000
26
57
false
synth:8581847523672130901@1
synth
[{"aliases": ["src/waypointforge/limits.py", "limits.py"], "kind": "file", "value": "src/waypointforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/waypointforge/cli.py", "cli.py"], "kind": "file", "value": "src/waypointforge/cl...
[{"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": ["core.py", "src/waypointforge/core.py"...
ctx_compress-s1-01040
creditledger
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] settle_balance drops override values of 0, False and empty string. It lives in src/creditledger/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 creditledger layout. [tool...
112,703
3,072
116
137
false
synth:1430104299901902991@2.03705
synth
[{"aliases": ["src/creditledger/cli.py", "cli.py"], "kind": "file", "value": "src/creditledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/creditledger/audit.py", "audit.py"], "kind": "file", "value": "src/creditledger/audit.py"}, {"aliase...
[{"answers": ["src/creditledger/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-s1-01040-twin
creditledger
ctx_compress-s1-01040
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] settle_balance drops override values of 0, False and empty string. It lives in src/creditledger/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 creditledger layout. [tool...
56,419
2,820
72
137
false
synth:1430104299901902991@2.03705
synth
[{"aliases": ["src/creditledger/cli.py", "cli.py"], "kind": "file", "value": "src/creditledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/creditledger/audit.py", "audit.py"], "kind": "file", "value": "src/creditledger/audit.py"}, {"aliase...
[{"answers": ["src/creditledger/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-s1-01041
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] net_balance produces doubled dashes for consecutive separators. 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 balanceflow layout. [too...
42,626
2,131
67
84
false
synth:2630290683466651248@1
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": ["src/balanceflow/compat.py", "compat.py"], "kind": "file", "value": "src/balanceflow/comp...
[{"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-s1-01041-twin
balanceflow
ctx_compress-s1-01041
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_balance produces doubled dashes for consecutive separators. 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 balanceflow layout. [too...
22,613
1,130
47
84
false
synth:2630290683466651248@1
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": ["src/balanceflow/compat.py", "compat.py"], "kind": "file", "value": "src/balanceflow/comp...
[{"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-s1-01042
samplekit
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] clip_trace produces doubled dashes for consecutive separators. It lives in src/samplekit/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 samplekit layout. [tool_use k...
55,371
2,768
87
108
false
synth:5049985865603774037@2.16387
synth
[{"aliases": ["src/samplekit/store.py", "store.py"], "kind": "file", "value": "src/samplekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sample_0.csv", "sample_0.csv"], "kind": "file", "value": "data/sample_0.csv"}, {"aliases": ["...
[{"answers": ["src/samplekit/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-s1-01042-twin
samplekit
ctx_compress-s1-01042
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] clip_trace produces doubled dashes for consecutive separators. It lives in src/samplekit/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 samplekit layout. [tool_use k...
49,372
2,468
63
108
false
synth:5049985865603774037@2.16387
synth
[{"aliases": ["src/samplekit/store.py", "store.py"], "kind": "file", "value": "src/samplekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sample_0.csv", "sample_0.csv"], "kind": "file", "value": "data/sample_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": ["data/probe_4.csv", "probe_4.csv"], "k...
ctx_compress-s1-01043
shipmentforge
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] restock_reorder skips the first page; page 1 returns what should be page 2. It lives in src/shipmentforge/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 shipmentforg...
41,978
2,098
78
95
false
synth:5271022411378493458@1
synth
[{"aliases": ["src/shipmentforge/store.py", "store.py"], "kind": "file", "value": "src/shipmentforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/shipmentforge/tables.py", "tables.py"], "kind": "file", "value": "src/shipmentforge/ta...
[{"answers": ["src/shipmentforge/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 ...
ctx_compress-s1-01043-twin
shipmentforge
ctx_compress-s1-01043
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] restock_reorder skips the first page; page 1 returns what should be page 2. It lives in src/shipmentforge/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 shipmentforg...
24,611
1,230
56
95
false
synth:5271022411378493458@1
synth
[{"aliases": ["src/shipmentforge/store.py", "store.py"], "kind": "file", "value": "src/shipmentforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/shipmentforge/tables.py", "tables.py"], "kind": "file", "value": "src/shipmentforge/ta...
[{"answers": ["src/shipmentforge/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 ...
ctx_compress-s1-01044
indexkit
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] collate_section ignores the cap, so late attempts wait for minutes. It lives in src/indexkit/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 indexkit layout. [tool_use ki...
69,855
3,072
91
108
false
synth:7370235749448403935@1.66823
synth
[{"aliases": ["src/indexkit/cli.py", "cli.py"], "kind": "file", "value": "src/indexkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/indexkit/store.py", "store.py"], "kind": "file", "value": "src/indexkit/store.py"}, {"aliases": ["src/indexk...
[{"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_audit.py", "tests/test_audit.py"...
ctx_compress-s1-01044-twin
indexkit
ctx_compress-s1-01044
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] collate_section ignores the cap, so late attempts wait for minutes. It lives in src/indexkit/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 indexkit layout. [tool_use ki...
40,944
2,047
53
108
false
synth:7370235749448403935@1.66823
synth
[{"aliases": ["src/indexkit/cli.py", "cli.py"], "kind": "file", "value": "src/indexkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/indexkit/store.py", "store.py"], "kind": "file", "value": "src/indexkit/store.py"}, {"aliases": ["src/indexk...
[{"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_limits.py", "tests/test_limits.p...
ctx_compress-s1-01045
bookingdesk
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] split_slot raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/bookingdesk/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 booki...
64,075
3,072
97
120
false
synth:2514279207804104018@2.1911
synth
[{"aliases": ["src/bookingdesk/report.py", "report.py"], "kind": "file", "value": "src/bookingdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/calendar_1.csv", "calendar_1.csv"], "kind": "file", "value": "data/calendar_1.csv"}...
[{"answers": ["src/bookingdesk/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-s1-01045-twin
bookingdesk
ctx_compress-s1-01045
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] split_slot raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/bookingdesk/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 booki...
37,864
1,893
55
120
false
synth:2514279207804104018@2.1911
synth
[{"aliases": ["src/bookingdesk/report.py", "report.py"], "kind": "file", "value": "src/bookingdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/calendar_1.csv", "calendar_1.csv"], "kind": "file", "value": "data/calendar_1.csv"}...
[{"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": ["schema.py", "src/bookingdesk/schema.p...
ctx_compress-s1-01046
tariffflow
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_balance_page misses keys that differ only in case. It lives in src/tariffflow/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 tariffflow layout. [tool_use kind=se...
67,750
3,072
90
107
false
synth:2806305140530860774@1.29693
synth
[{"aliases": ["src/tariffflow/store.py", "store.py"], "kind": "file", "value": "src/tariffflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/tariffflow/format.py", "format.py"], "kind": "file", "value": "src/tariffflow/format.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": ["io.py", "src/tariffflow/io.py"], "kin...
ctx_compress-s1-01046-twin
tariffflow
ctx_compress-s1-01046
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_balance_page misses keys that differ only in case. It lives in src/tariffflow/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 tariffflow layout. [tool_use kind=se...
35,955
1,797
50
107
false
synth:2806305140530860774@1.29693
synth
[{"aliases": ["src/tariffflow/store.py", "store.py"], "kind": "file", "value": "src/tariffflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/tariffflow/format.py", "format.py"], "kind": "file", "value": "src/tariffflow/format.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_io.py", "tests/test_io.py"], "ki...
ctx_compress-s1-01047
shiftledger
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_slot_window misses keys that differ only in case. It lives in src/shiftledger/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 shiftledger layout. [tool_use ...
67,539
3,072
92
113
false
synth:781089477140473956@1.50913
synth
[{"aliases": ["src/shiftledger/limits.py", "limits.py"], "kind": "file", "value": "src/shiftledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/shiftledger/report.py", "report.py"], "kind": "file", "value": "src/shiftledger/repo...
[{"answers": ["src/shiftledger/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-s1-01047-twin
shiftledger
ctx_compress-s1-01047
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_slot_window misses keys that differ only in case. It lives in src/shiftledger/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 shiftledger layout. [tool_use ...
41,072
2,053
54
113
false
synth:781089477140473956@1.50913
synth
[{"aliases": ["src/shiftledger/limits.py", "limits.py"], "kind": "file", "value": "src/shiftledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/shiftledger/report.py", "report.py"], "kind": "file", "value": "src/shiftledger/repo...
[{"answers": ["src/shiftledger/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 ...
ctx_compress-s1-01048
beaconkit
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] interpolate_zone rejects a range where lo equals hi, which is a valid single point range. It lives in src/beaconkit/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 ...
92,769
3,072
100
119
false
synth:4790044320126902727@1.86525
synth
[{"aliases": ["src/beaconkit/report.py", "report.py"], "kind": "file", "value": "src/beaconkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/leg_8.csv", "leg_8.csv"], "kind": "file", "value": "data/leg_8.csv"}, {"aliases": ["src...
[{"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_core.py", "tests/test_core.py"],...
ctx_compress-s1-01048-twin
beaconkit
ctx_compress-s1-01048
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] interpolate_zone rejects a range where lo equals hi, which is a valid single point range. It lives in src/beaconkit/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 ...
53,692
2,684
62
119
false
synth:4790044320126902727@1.86525
synth
[{"aliases": ["src/beaconkit/report.py", "report.py"], "kind": "file", "value": "src/beaconkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/leg_8.csv", "leg_8.csv"], "kind": "file", "value": "data/leg_8.csv"}, {"aliases": ["src...
[{"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_audit.py", "tests/test_audit.py"...
ctx_compress-s1-01049
tileforge
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] interpolate_fix raises WaypointBoundError for bytes input in strict mode instead of decoding it. It lives in src/tileforge/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...
70,864
3,072
108
117
false
synth:9087363123849386140@1.24905
synth
[{"aliases": ["src/tileforge/cli.py", "cli.py"], "kind": "file", "value": "src/tileforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/tileforge/rules.py", "rules.py"], "kind": "file", "value": "src/tileforge/rules.py"}, {"aliases": ["data/f...
[{"answers": ["src/tileforge/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 rec...
ctx_compress-s1-01049-twin
tileforge
ctx_compress-s1-01049
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] interpolate_fix raises WaypointBoundError for bytes input in strict mode instead of decoding it. It lives in src/tileforge/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...
40,241
2,012
66
117
false
synth:9087363123849386140@1.24905
synth
[{"aliases": ["src/tileforge/cli.py", "cli.py"], "kind": "file", "value": "src/tileforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/tileforge/rules.py", "rules.py"], "kind": "file", "value": "src/tileforge/rules.py"}, {"aliases": ["data/f...
[{"answers": ["src/tileforge/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 rec...
ctx_compress-s1-01050
principalkit
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] mint_policy_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/principalkit/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 princ...
67,668
3,072
93
110
false
synth:7089231308300848387@1.09215
synth
[{"aliases": ["src/principalkit/cli.py", "cli.py"], "kind": "file", "value": "src/principalkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/principalkit/schema.py", "schema.py"], "kind": "file", "value": "src/principalkit/schema.py"}, {"ali...
[{"answers": ["src/principalkit/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-s1-01050-twin
principalkit
ctx_compress-s1-01050
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] mint_policy_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/principalkit/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 princ...
33,104
1,655
51
110
false
synth:7089231308300848387@1.09215
synth
[{"aliases": ["src/principalkit/cli.py", "cli.py"], "kind": "file", "value": "src/principalkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/principalkit/schema.py", "schema.py"], "kind": "file", "value": "src/principalkit/schema.py"}, {"ali...
[{"answers": ["src/principalkit/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-s1-01051
fixforge
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] snap_corridor produces doubled dashes for consecutive separators. It lives in src/fixforge/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 fixforge layout. [tool_use ...
60,068
3,003
112
133
false
synth:8068249042509116909@1.68695
synth
[{"aliases": ["src/fixforge/audit.py", "audit.py"], "kind": "file", "value": "src/fixforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/bearing_6.csv", "bearing_6.csv"], "kind": "file", "value": "data/bearing_6.csv"}, {"aliases": [...
[{"answers": ["src/fixforge/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 most ...
ctx_compress-s1-01051-twin
fixforge
ctx_compress-s1-01051
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] snap_corridor produces doubled dashes for consecutive separators. It lives in src/fixforge/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 fixforge layout. [tool_use ...
38,234
1,911
68
133
false
synth:8068249042509116909@1.68695
synth
[{"aliases": ["src/fixforge/audit.py", "audit.py"], "kind": "file", "value": "src/fixforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/bearing_6.csv", "bearing_6.csv"], "kind": "file", "value": "data/bearing_6.csv"}, {"aliases": [...
[{"answers": ["src/fixforge/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 most ...
ctx_compress-s1-01052
captionledger
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_stem double counts carried amounts when the period closes on a weekend. It lives in src/captionledger/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 captionle...
23,685
1,184
55
64
false
synth:9074000213510474384@1
synth
[{"aliases": ["src/captionledger/compat.py", "compat.py"], "kind": "file", "value": "src/captionledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/captionledger/rules.py", "rules.py"], "kind": "file", "value": "src/captionledge...
[{"answers": ["src/captionledger/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'...
ctx_compress-s1-01052-twin
captionledger
ctx_compress-s1-01052
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_stem double counts carried amounts when the period closes on a weekend. It lives in src/captionledger/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 captionle...
12,241
1,000
33
64
false
synth:9074000213510474384@1
synth
[{"aliases": ["src/captionledger/compat.py", "compat.py"], "kind": "file", "value": "src/captionledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/captionledger/rules.py", "rules.py"], "kind": "file", "value": "src/captionledge...
[{"answers": ["src/captionledger/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'...
ctx_compress-s1-01053
scopeworks
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] mint_session raises KeyError for a missing key instead of returning the default. It lives in src/scopeworks/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 scopeworks lay...
45,255
2,262
77
92
false
synth:4513527002191946091@1
synth
[{"aliases": ["src/scopeworks/cli.py", "cli.py"], "kind": "file", "value": "src/scopeworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/claim_6.csv", "claim_6.csv"], "kind": "file", "value": "data/claim_6.csv"}, {"aliases": ["docs/changelo...
[{"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": ["claim_2.csv", "data/claim_2.csv"], "k...
ctx_compress-s1-01053-twin
scopeworks
ctx_compress-s1-01053
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] mint_session raises KeyError for a missing key instead of returning the default. It lives in src/scopeworks/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 scopeworks lay...
33,009
1,650
49
92
false
synth:4513527002191946091@1
synth
[{"aliases": ["src/scopeworks/cli.py", "cli.py"], "kind": "file", "value": "src/scopeworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/claim_6.csv", "claim_6.csv"], "kind": "file", "value": "data/claim_6.csv"}, {"aliases": ["docs/changelo...
[{"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": ["limits.py", "src/scopeworks/limits.py...
ctx_compress-s1-01054
chunkledger
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_lexeme raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/chunkledger/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...
99,984
3,072
98
109
false
synth:634906833706830650@1.6585
synth
[{"aliases": ["src/chunkledger/compat.py", "compat.py"], "kind": "file", "value": "src/chunkledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/chunkledger/limits.py", "limits.py"], "kind": "file", "value": "src/chunkledger/limi...
[{"answers": ["src/chunkledger/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-s1-01054-twin
chunkledger
ctx_compress-s1-01054
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_lexeme raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/chunkledger/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...
52,679
2,633
62
109
false
synth:634906833706830650@1.6585
synth
[{"aliases": ["src/chunkledger/compat.py", "compat.py"], "kind": "file", "value": "src/chunkledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/chunkledger/limits.py", "limits.py"], "kind": "file", "value": "src/chunkledger/limi...
[{"answers": ["src/chunkledger/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 ...
ctx_compress-s1-01055
payoutforge
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_ledger drops override values of 0, False and empty string. It lives in src/payoutforge/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 payoutforge layout....
69,113
3,072
85
106
false
synth:7565067074176464131@2.11558
synth
[{"aliases": ["src/payoutforge/limits.py", "limits.py"], "kind": "file", "value": "src/payoutforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/payoutforge/schema.py", "schema.py"], "kind": "file", "value": "src/payoutforge/sche...
[{"answers": ["src/payoutforge/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-s1-01055-twin
payoutforge
ctx_compress-s1-01055
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_ledger drops override values of 0, False and empty string. It lives in src/payoutforge/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 payoutforge layout....
47,510
2,375
61
106
false
synth:7565067074176464131@2.11558
synth
[{"aliases": ["src/payoutforge/limits.py", "limits.py"], "kind": "file", "value": "src/payoutforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/payoutforge/schema.py", "schema.py"], "kind": "file", "value": "src/payoutforge/sche...
[{"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_1.csv", "data/balance_1.csv"]...