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-s1-01758-twin
bookingforge
ctx_compress-s1-01758
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_booking grows linearly instead of doubling. It lives in src/bookingforge/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 bookingforge layout. [tool_use kind=s...
33,847
1,692
63
108
false
synth:2088005380620478515@1.37952
synth
[{"aliases": ["src/bookingforge/limits.py", "limits.py"], "kind": "file", "value": "src/bookingforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/bookingforge/cli.py", "cli.py"], "kind": "file", "value": "src/bookingforge/cli.py...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/bookingforge/limits....
ctx_compress-s1-01759
refunddesk
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_tariff returns the upper bound for every value instead of clamping into the range. It lives in src/refunddesk/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 refu...
99,069
3,072
133
154
false
synth:1764815298080200210@2.31727
synth
[{"aliases": ["src/refunddesk/io.py", "io.py"], "kind": "file", "value": "src/refunddesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/refunddesk/store.py", "store.py"], "kind": "file", "value": "src/refunddesk/store.py"}, {"aliases": ["src/refu...
[{"answers": ["src/refunddesk/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 rece...
ctx_compress-s1-01759-twin
refunddesk
ctx_compress-s1-01759
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_tariff returns the upper bound for every value instead of clamping into the range. It lives in src/refunddesk/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 refu...
63,519
3,072
89
154
false
synth:1764815298080200210@2.31727
synth
[{"aliases": ["src/refunddesk/io.py", "io.py"], "kind": "file", "value": "src/refunddesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/refunddesk/store.py", "store.py"], "kind": "file", "value": "src/refunddesk/store.py"}, {"aliases": ["src/refu...
[{"answers": ["src/refunddesk/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 rece...
ctx_compress-s1-01760
quotakit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_window_once raises KeyError for a missing key instead of returning the default. It lives in src/quotakit/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 quot...
23,670
1,183
58
63
false
synth:7264248800647030966@1
synth
[{"aliases": ["src/quotakit/limits.py", "limits.py"], "kind": "file", "value": "src/quotakit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/quota_2.csv", "quota_2.csv"], "kind": "file", "value": "data/quota_2.csv"}, {"aliases": [...
[{"answers": ["src/quotakit/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 mos...
ctx_compress-s1-01760-twin
quotakit
ctx_compress-s1-01760
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_window_once raises KeyError for a missing key instead of returning the default. It lives in src/quotakit/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 quot...
11,758
1,000
31
63
false
synth:7264248800647030966@1
synth
[{"aliases": ["src/quotakit/limits.py", "limits.py"], "kind": "file", "value": "src/quotakit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/quota_2.csv", "quota_2.csv"], "kind": "file", "value": "data/quota_2.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": ["limits.py", "src/quotakit/limits.py"]...
ctx_compress-s1-01761
slugdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_stem drops override values of 0, False and empty string. It lives in src/slugdesk/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 slugdesk layout. [tool_use kind=s...
110,162
3,072
112
123
false
synth:5673197346014184139@2.06902
synth
[{"aliases": ["src/slugdesk/io.py", "io.py"], "kind": "file", "value": "src/slugdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/snippet_5.csv", "snippet_5.csv"], "kind": "file", "value": "data/snippet_5.csv"}, {"aliases": ["data/cue_0.csv", ...
[{"answers": ["src/slugdesk/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-s1-01761-twin
slugdesk
ctx_compress-s1-01761
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_stem drops override values of 0, False and empty string. It lives in src/slugdesk/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 slugdesk layout. [tool_use kind=s...
48,925
2,446
60
123
false
synth:5673197346014184139@2.06902
synth
[{"aliases": ["src/slugdesk/io.py", "io.py"], "kind": "file", "value": "src/slugdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/snippet_5.csv", "snippet_5.csv"], "kind": "file", "value": "data/snippet_5.csv"}, {"aliases": ["data/cue_0.csv", ...
[{"answers": ["src/slugdesk/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-s1-01762
warehouseledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_pallet raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/warehouseledger/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...
46,268
2,313
76
85
false
synth:2613794018265131976@1
synth
[{"aliases": ["src/warehouseledger/format.py", "format.py"], "kind": "file", "value": "src/warehouseledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/warehouse_1.csv", "warehouse_1.csv"], "kind": "file", "value": "data/wareho...
[{"answers": ["src/warehouseledger/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": ["pass"], "question": "Did the agen...
ctx_compress-s1-01762-twin
warehouseledger
ctx_compress-s1-01762
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_pallet raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/warehouseledger/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...
26,975
1,348
48
85
false
synth:2613794018265131976@1
synth
[{"aliases": ["src/warehouseledger/format.py", "format.py"], "kind": "file", "value": "src/warehouseledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/warehouse_1.csv", "warehouse_1.csv"], "kind": "file", "value": "data/wareho...
[{"answers": ["src/warehouseledger/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": ["pass"], "question": "Did the agen...
ctx_compress-s1-01763
slugkit
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_cue_strict returns the upper bound for every value instead of clamping into the range. It lives in src/slugkit/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 th...
79,974
3,072
127
138
false
synth:2660405105051813832@1.5225
synth
[{"aliases": ["src/slugkit/store.py", "store.py"], "kind": "file", "value": "src/slugkit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/slug_7.csv", "slug_7.csv"], "kind": "file", "value": "data/slug_7.csv"}, {"aliases": ["src/slugki...
[{"answers": ["src/slugkit/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 r...
ctx_compress-s1-01763-twin
slugkit
ctx_compress-s1-01763
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_cue_strict returns the upper bound for every value instead of clamping into the range. It lives in src/slugkit/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 th...
38,656
1,932
75
138
false
synth:2660405105051813832@1.5225
synth
[{"aliases": ["src/slugkit/store.py", "store.py"], "kind": "file", "value": "src/slugkit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/slug_7.csv", "slug_7.csv"], "kind": "file", "value": "data/slug_7.csv"}, {"aliases": ["src/slugki...
[{"answers": ["src/slugkit/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-01764
draftdesk
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_margin returns one item too few on every page. It lives in src/draftdesk/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 draftdesk layout. [tool_use kind=sear...
88,723
3,072
89
102
false
synth:5566897059159768627@1.7658
synth
[{"aliases": ["src/draftdesk/rules.py", "rules.py"], "kind": "file", "value": "src/draftdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/margin_6.csv", "margin_6.csv"], "kind": "file", "value": "data/margin_6.csv"}, {"aliases": ["...
[{"answers": ["src/draftdesk/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-s1-01764-twin
draftdesk
ctx_compress-s1-01764
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_margin returns one item too few on every page. It lives in src/draftdesk/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 draftdesk layout. [tool_use kind=sear...
57,737
2,886
59
102
false
synth:5566897059159768627@1.7658
synth
[{"aliases": ["src/draftdesk/rules.py", "rules.py"], "kind": "file", "value": "src/draftdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/margin_6.csv", "margin_6.csv"], "kind": "file", "value": "data/margin_6.csv"}, {"aliases": ["...
[{"answers": ["src/draftdesk/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-s1-01765
receiptworks
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_credit grows linearly instead of doubling. It lives in src/receiptworks/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 receiptworks layout. [tool_use kind=s...
88,318
3,072
103
112
false
synth:1435860981785756069@1.76485
synth
[{"aliases": ["src/receiptworks/limits.py", "limits.py"], "kind": "file", "value": "src/receiptworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/credit_1.csv", "credit_1.csv"], "kind": "file", "value": "data/credit_1.csv"}, {"...
[{"answers": ["src/receiptworks/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-s1-01765-twin
receiptworks
ctx_compress-s1-01765
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_credit grows linearly instead of doubling. It lives in src/receiptworks/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 receiptworks layout. [tool_use kind=s...
46,132
2,306
51
112
false
synth:1435860981785756069@1.76485
synth
[{"aliases": ["src/receiptworks/limits.py", "limits.py"], "kind": "file", "value": "src/receiptworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/credit_1.csv", "credit_1.csv"], "kind": "file", "value": "data/credit_1.csv"}, {"...
[{"answers": ["src/receiptworks/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-s1-01766
marginledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_clause double counts carried amounts when the period closes on a weekend. It lives in src/marginledger/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 marginle...
39,318
1,965
78
83
false
synth:5147295514860647280@1
synth
[{"aliases": ["src/marginledger/format.py", "format.py"], "kind": "file", "value": "src/marginledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/marginledger/cli.py", "cli.py"], "kind": "file", "value": "src/marginledger/cli.py...
[{"answers": ["src/marginledger/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-s1-01766-twin
marginledger
ctx_compress-s1-01766
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_clause double counts carried amounts when the period closes on a weekend. It lives in src/marginledger/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 marginle...
25,382
1,269
48
83
false
synth:5147295514860647280@1
synth
[{"aliases": ["src/marginledger/format.py", "format.py"], "kind": "file", "value": "src/marginledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/marginledger/cli.py", "cli.py"], "kind": "file", "value": "src/marginledger/cli.py...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_rules.py", "tests/test_rules.py"...
ctx_compress-s1-01767
marginkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_footnote rejects a range where lo equals hi, which is a valid single point range. It lives in src/marginkit/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 margi...
71,167
3,072
92
107
false
synth:1763555969668876125@1.9054
synth
[{"aliases": ["src/marginkit/cli.py", "cli.py"], "kind": "file", "value": "src/marginkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/marginkit/limits.py", "limits.py"], "kind": "file", "value": "src/marginkit/limits.py"}, {"aliases": ["doc...
[{"answers": ["src/marginkit/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-s1-01767-twin
marginkit
ctx_compress-s1-01767
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_footnote rejects a range where lo equals hi, which is a valid single point range. It lives in src/marginkit/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 margi...
39,012
1,950
53
107
false
synth:1763555969668876125@1.9054
synth
[{"aliases": ["src/marginkit/cli.py", "cli.py"], "kind": "file", "value": "src/marginkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/marginkit/limits.py", "limits.py"], "kind": "file", "value": "src/marginkit/limits.py"}, {"aliases": ["doc...
[{"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-s1-01768
captiondesk
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] normalise_chunk double counts carried amounts when the period closes on a weekend. It lives in src/captiondesk/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 caption...
56,197
2,809
92
107
false
synth:2077053801170387025@1.26362
synth
[{"aliases": ["src/captiondesk/audit.py", "audit.py"], "kind": "file", "value": "src/captiondesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/captiondesk/compat.py", "compat.py"], "kind": "file", "value": "src/captiondesk/compat.py"...
[{"answers": ["src/captiondesk/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-s1-01768-twin
captiondesk
ctx_compress-s1-01768
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] normalise_chunk double counts carried amounts when the period closes on a weekend. It lives in src/captiondesk/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 caption...
36,128
1,806
62
107
false
synth:2077053801170387025@1.26362
synth
[{"aliases": ["src/captiondesk/audit.py", "audit.py"], "kind": "file", "value": "src/captiondesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/captiondesk/compat.py", "compat.py"], "kind": "file", "value": "src/captiondesk/compat.py"...
[{"answers": ["src/captiondesk/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-s1-01769
cadencedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_slot produces doubled dashes for consecutive separators. It lives in src/cadencedesk/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 cadencedesk layout. [too...
23,657
1,182
56
65
false
synth:6487527950627128318@1
synth
[{"aliases": ["src/cadencedesk/format.py", "format.py"], "kind": "file", "value": "src/cadencedesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/cadencedesk/io.py", "io.py"], "kind": "file", "value": "src/cadencedesk/io.py"}, {"a...
[{"answers": ["src/cadencedesk/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": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01769-twin
cadencedesk
ctx_compress-s1-01769
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_slot produces doubled dashes for consecutive separators. It lives in src/cadencedesk/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 cadencedesk layout. [too...
12,231
1,000
30
65
false
synth:6487527950627128318@1
synth
[{"aliases": ["src/cadencedesk/format.py", "format.py"], "kind": "file", "value": "src/cadencedesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/cadencedesk/io.py", "io.py"], "kind": "file", "value": "src/cadencedesk/io.py"}, {"a...
[{"answers": ["src/cadencedesk/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-s1-01770
scopeforge
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] narrow_grant_batch rejects a range where lo equals hi, which is a valid single point range. It lives in src/scopeforge/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...
107,527
3,072
125
148
false
synth:7373059177158104153@2.29355
synth
[{"aliases": ["src/scopeforge/rules.py", "rules.py"], "kind": "file", "value": "src/scopeforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/principal_12.csv", "principal_12.csv"], "kind": "file", "value": "data/principal_12.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": ["data/token_8.csv", "token_8.csv"], "k...
ctx_compress-s1-01770-twin
scopeforge
ctx_compress-s1-01770
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] narrow_grant_batch rejects a range where lo equals hi, which is a valid single point range. It lives in src/scopeforge/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...
62,085
3,072
67
148
false
synth:7373059177158104153@2.29355
synth
[{"aliases": ["src/scopeforge/rules.py", "rules.py"], "kind": "file", "value": "src/scopeforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/principal_12.csv", "principal_12.csv"], "kind": "file", "value": "data/principal_12.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": ["test_io.py", "tests/test_io.py"], "ki...
ctx_compress-s1-01771
ledgerflow
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_receipt skips the first page; page 1 returns what should be page 2. It lives in src/ledgerflow/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 ledgerflow layout....
48,694
2,434
70
89
false
synth:1224234677807208135@1
synth
[{"aliases": ["src/ledgerflow/core.py", "core.py"], "kind": "file", "value": "src/ledgerflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/invoice_0.csv", "invoice_0.csv"], "kind": "file", "value": "data/invoice_0.csv"}, {"aliases": ["s...
[{"answers": ["src/ledgerflow/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 ...
ctx_compress-s1-01771-twin
ledgerflow
ctx_compress-s1-01771
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_receipt skips the first page; page 1 returns what should be page 2. It lives in src/ledgerflow/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 ledgerflow layout....
33,426
1,671
52
89
false
synth:1224234677807208135@1
synth
[{"aliases": ["src/ledgerflow/core.py", "core.py"], "kind": "file", "value": "src/ledgerflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/invoice_0.csv", "invoice_0.csv"], "kind": "file", "value": "data/invoice_0.csv"}, {"aliases": ["s...
[{"answers": ["src/ledgerflow/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 ...
ctx_compress-s1-01772
seriesforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series raises TraceShapeError for bytes input in strict mode instead of decoding it. It lives in src/seriesforge/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 mapp...
75,338
3,072
88
109
false
synth:7415452718193432858@1.76705
synth
[{"aliases": ["src/seriesforge/limits.py", "limits.py"], "kind": "file", "value": "src/seriesforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/seriesforge/schema.py", "schema.py"], "kind": "file", "value": "src/seriesforge/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": ["data/gauge_8.csv", "gauge_8.csv"], "k...
ctx_compress-s1-01772-twin
seriesforge
ctx_compress-s1-01772
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series raises TraceShapeError for bytes input in strict mode instead of decoding it. It lives in src/seriesforge/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 mapp...
47,870
2,393
64
109
false
synth:7415452718193432858@1.76705
synth
[{"aliases": ["src/seriesforge/limits.py", "limits.py"], "kind": "file", "value": "src/seriesforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/seriesforge/schema.py", "schema.py"], "kind": "file", "value": "src/seriesforge/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": ["test_io.py", "tests/test_io.py"], "ki...
ctx_compress-s1-01773
quotakit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_quota_safe produces doubled dashes for consecutive separators. It lives in src/quotakit/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the quotakit layout. [to...
41,414
2,070
84
101
false
synth:8115942815423507837@1.0489
synth
[{"aliases": ["src/quotakit/format.py", "format.py"], "kind": "file", "value": "src/quotakit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/quotakit/cli.py", "cli.py"], "kind": "file", "value": "src/quotakit/cli.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/quotakit/tables.py", "tables.py"]...
ctx_compress-s1-01773-twin
quotakit
ctx_compress-s1-01773
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_quota_safe produces doubled dashes for consecutive separators. It lives in src/quotakit/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the quotakit layout. [to...
26,407
1,320
54
101
false
synth:8115942815423507837@1.0489
synth
[{"aliases": ["src/quotakit/format.py", "format.py"], "kind": "file", "value": "src/quotakit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/quotakit/cli.py", "cli.py"], "kind": "file", "value": "src/quotakit/cli.py"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/quotakit/audit.py"], ...
ctx_compress-s1-01774
revisionforge
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_heading_safe raises FootnoteStateError for bytes input in strict mode instead of decoding it. It lives in src/revisionforge/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 ma...
78,345
3,072
88
105
false
synth:7983267436686683795@2.30733
synth
[{"aliases": ["src/revisionforge/cli.py", "cli.py"], "kind": "file", "value": "src/revisionforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/revisionforge/store.py", "store.py"], "kind": "file", "value": "src/revisionforge/store.py"}, {"al...
[{"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/revisionforge/format...
ctx_compress-s1-01774-twin
revisionforge
ctx_compress-s1-01774
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_heading_safe raises FootnoteStateError for bytes input in strict mode instead of decoding it. It lives in src/revisionforge/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 ma...
43,661
2,183
55
105
false
synth:7983267436686683795@2.30733
synth
[{"aliases": ["src/revisionforge/cli.py", "cli.py"], "kind": "file", "value": "src/revisionforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/revisionforge/store.py", "store.py"], "kind": "file", "value": "src/revisionforge/store.py"}, {"al...
[{"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/draft_1.csv", "draft_1.csv"], "k...
ctx_compress-s1-01775
tokenworks
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_token_once double counts carried amounts when the period closes on a weekend. It lives in src/tokenworks/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 tokenwor...
40,715
2,035
76
97
false
synth:3335562032469068686@1.21665
synth
[{"aliases": ["src/tokenworks/core.py", "core.py"], "kind": "file", "value": "src/tokenworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/tokenworks/compat.py", "compat.py"], "kind": "file", "value": "src/tokenworks/compat.py"}, {"alias...
[{"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/session_1.csv", "session_1.csv"]...
ctx_compress-s1-01775-twin
tokenworks
ctx_compress-s1-01775
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_token_once double counts carried amounts when the period closes on a weekend. It lives in src/tokenworks/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 tokenwor...
21,342
1,067
47
97
false
synth:3335562032469068686@1.21665
synth
[{"aliases": ["src/tokenworks/core.py", "core.py"], "kind": "file", "value": "src/tokenworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/tokenworks/compat.py", "compat.py"], "kind": "file", "value": "src/tokenworks/compat.py"}, {"alias...
[{"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/tokenworks/limits.py...
ctx_compress-s1-01776
shipmentledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_reorder double counts carried amounts when the period closes on a weekend. It lives in src/shipmentledger/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 shipmen...
97,761
3,072
106
133
false
synth:5470842069998118162@2.07468
synth
[{"aliases": ["src/shipmentledger/rules.py", "rules.py"], "kind": "file", "value": "src/shipmentledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/stock_4.csv", "stock_4.csv"], "kind": "file", "value": "data/stock_4.csv"}, {"alias...
[{"answers": ["src/shipmentledger/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...
ctx_compress-s1-01776-twin
shipmentledger
ctx_compress-s1-01776
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_reorder double counts carried amounts when the period closes on a weekend. It lives in src/shipmentledger/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 shipmen...
49,527
2,476
62
133
false
synth:5470842069998118162@2.07468
synth
[{"aliases": ["src/shipmentledger/rules.py", "rules.py"], "kind": "file", "value": "src/shipmentledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/stock_4.csv", "stock_4.csv"], "kind": "file", "value": "data/stock_4.csv"}, {"alias...
[{"answers": ["src/shipmentledger/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...
ctx_compress-s1-01777
tilekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_corridor skips the first page; page 1 returns what should be page 2. It lives in src/tilekit/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tilekit lay...
76,654
3,072
97
120
false
synth:5621848233127650133@1.89068
synth
[{"aliases": ["src/tilekit/limits.py", "limits.py"], "kind": "file", "value": "src/tilekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/tilekit/cli.py", "cli.py"], "kind": "file", "value": "src/tilekit/cli.py"}, {"aliases": ["da...
[{"answers": ["src/tilekit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01777-twin
tilekit
ctx_compress-s1-01777
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_corridor skips the first page; page 1 returns what should be page 2. It lives in src/tilekit/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tilekit lay...
38,480
1,924
55
120
false
synth:5621848233127650133@1.89068
synth
[{"aliases": ["src/tilekit/limits.py", "limits.py"], "kind": "file", "value": "src/tilekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/tilekit/cli.py", "cli.py"], "kind": "file", "value": "src/tilekit/cli.py"}, {"aliases": ["da...
[{"answers": ["src/tilekit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01778
balanceworks
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_credit_batch raises InvoiceShapeError for bytes input in strict mode instead of decoding it. It lives in src/balanceworks/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 ...
111,853
3,072
127
138
false
synth:7872595171593044563@2.1458
synth
[{"aliases": ["src/balanceworks/limits.py", "limits.py"], "kind": "file", "value": "src/balanceworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/balanceworks/io.py", "io.py"], "kind": "file", "value": "src/balanceworks/io.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": ["limits.py", "src/balanceworks/limits....
ctx_compress-s1-01778-twin
balanceworks
ctx_compress-s1-01778
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_credit_batch raises InvoiceShapeError for bytes input in strict mode instead of decoding it. It lives in src/balanceworks/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 ...
61,072
3,053
65
138
false
synth:7872595171593044563@2.1458
synth
[{"aliases": ["src/balanceworks/limits.py", "limits.py"], "kind": "file", "value": "src/balanceworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/balanceworks/io.py", "io.py"], "kind": "file", "value": "src/balanceworks/io.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_cli.py", "tests/test_cli.py"], "...
ctx_compress-s1-01779
cadencedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_booking lets defaults overwrite explicit overrides. It lives in src/cadencedesk/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 cadencedesk layout. [tool_use k...
35,587
1,779
89
112
false
synth:5326308980668349630@1.07577
synth
[{"aliases": ["src/cadencedesk/rules.py", "rules.py"], "kind": "file", "value": "src/cadencedesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/cadencedesk/compat.py", "compat.py"], "kind": "file", "value": "src/cadencedesk/compat.py"...
[{"answers": ["src/cadencedesk/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 mo...
ctx_compress-s1-01779-twin
cadencedesk
ctx_compress-s1-01779
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_booking lets defaults overwrite explicit overrides. It lives in src/cadencedesk/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 cadencedesk layout. [tool_use k...
25,492
1,274
59
112
false
synth:5326308980668349630@1.07577
synth
[{"aliases": ["src/cadencedesk/rules.py", "rules.py"], "kind": "file", "value": "src/cadencedesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/cadencedesk/compat.py", "compat.py"], "kind": "file", "value": "src/cadencedesk/compat.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_compat.py", "tests/test_compat.p...
ctx_compress-s1-01780
chunkworks
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 produces doubled dashes for consecutive separators. It lives in src/chunkworks/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 chunkworks layout. [tool_use k...
62,428
3,072
99
110
false
synth:1099572687223783501@1.36633
synth
[{"aliases": ["src/chunkworks/io.py", "io.py"], "kind": "file", "value": "src/chunkworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/chunkworks/tables.py", "tables.py"], "kind": "file", "value": "src/chunkworks/tables.py"}, {"aliases": ["src/c...
[{"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/chunkworks/io.py"], "kin...
ctx_compress-s1-01780-twin
chunkworks
ctx_compress-s1-01780
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 produces doubled dashes for consecutive separators. It lives in src/chunkworks/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 chunkworks layout. [tool_use k...
27,578
1,378
55
110
false
synth:1099572687223783501@1.36633
synth
[{"aliases": ["src/chunkworks/io.py", "io.py"], "kind": "file", "value": "src/chunkworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/chunkworks/tables.py", "tables.py"], "kind": "file", "value": "src/chunkworks/tables.py"}, {"aliases": ["src/c...
[{"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/chunkworks/tables.py", "tables.py...
ctx_compress-s1-01781
zoneforge
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_safe skips the first page; page 1 returns what should be page 2. It lives in src/zoneforge/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 zoneforge layo...
93,533
3,072
117
136
false
synth:642422192390941557@1.99148
synth
[{"aliases": ["src/zoneforge/compat.py", "compat.py"], "kind": "file", "value": "src/zoneforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/zone_5.csv", "zone_5.csv"], "kind": "file", "value": "data/zone_5.csv"}, {"aliases": ["...
[{"answers": ["src/zoneforge/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-s1-01781-twin
zoneforge
ctx_compress-s1-01781
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_safe skips the first page; page 1 returns what should be page 2. It lives in src/zoneforge/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 zoneforge layo...
45,822
2,291
65
136
false
synth:642422192390941557@1.99148
synth
[{"aliases": ["src/zoneforge/compat.py", "compat.py"], "kind": "file", "value": "src/zoneforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/zone_5.csv", "zone_5.csv"], "kind": "file", "value": "data/zone_5.csv"}, {"aliases": ["...
[{"answers": ["src/zoneforge/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 mo...
ctx_compress-s1-01782
policydesk
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_scope raises PolicyBoundError for bytes input in strict mode instead of decoding it. It lives in src/policydesk/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...
44,515
2,225
86
109
false
synth:7499628295196556858@1
synth
[{"aliases": ["src/policydesk/rules.py", "rules.py"], "kind": "file", "value": "src/policydesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/policydesk/format.py", "format.py"], "kind": "file", "value": "src/policydesk/format.py"}, {...
[{"answers": ["src/policydesk/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01782-twin
policydesk
ctx_compress-s1-01782
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_scope raises PolicyBoundError for bytes input in strict mode instead of decoding it. It lives in src/policydesk/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...
32,927
1,646
62
109
false
synth:7499628295196556858@1
synth
[{"aliases": ["src/policydesk/rules.py", "rules.py"], "kind": "file", "value": "src/policydesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/policydesk/format.py", "format.py"], "kind": "file", "value": "src/policydesk/format.py"}, {...
[{"answers": ["src/policydesk/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01783
reorderdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_bin_strict keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/reorderdesk/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 reorde...
59,012
2,950
92
109
false
synth:1094258932137415902@1.16118
synth
[{"aliases": ["src/reorderdesk/store.py", "store.py"], "kind": "file", "value": "src/reorderdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/reorderdesk/schema.py", "schema.py"], "kind": "file", "value": "src/reorderdesk/schema.py"...
[{"answers": ["src/reorderdesk/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 mo...
ctx_compress-s1-01783-twin
reorderdesk
ctx_compress-s1-01783
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_bin_strict keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/reorderdesk/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 reorde...
30,290
1,514
54
109
false
synth:1094258932137415902@1.16118
synth
[{"aliases": ["src/reorderdesk/store.py", "store.py"], "kind": "file", "value": "src/reorderdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/reorderdesk/schema.py", "schema.py"], "kind": "file", "value": "src/reorderdesk/schema.py"...
[{"answers": ["src/reorderdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01784
ledgerkit
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_payout raises ValueError when a value itself contains an equals sign. It lives in src/ledgerkit/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 ledgerkit lay...
69,098
3,072
108
117
false
synth:2406048068570003097@1.721
synth
[{"aliases": ["src/ledgerkit/report.py", "report.py"], "kind": "file", "value": "src/ledgerkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/ledgerkit/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerkit/limits.py"},...
[{"answers": ["src/ledgerkit/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-s1-01784-twin
ledgerkit
ctx_compress-s1-01784
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_payout raises ValueError when a value itself contains an equals sign. It lives in src/ledgerkit/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 ledgerkit lay...
39,834
1,991
68
117
false
synth:2406048068570003097@1.721
synth
[{"aliases": ["src/ledgerkit/report.py", "report.py"], "kind": "file", "value": "src/ledgerkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/ledgerkit/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerkit/limits.py"},...
[{"answers": ["src/ledgerkit/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-s1-01785
slugkit
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] normalise_snippet raises StemStateError for bytes input in strict mode instead of decoding it. It lives in src/slugkit/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 t...
37,472
1,873
56
69
false
synth:9130179261384640915@1
synth
[{"aliases": ["src/slugkit/report.py", "report.py"], "kind": "file", "value": "src/slugkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/slugkit/audit.py", "audit.py"], "kind": "file", "value": "src/slugkit/audit.py"}, {"aliases"...
[{"answers": ["src/slugkit/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01785-twin
slugkit
ctx_compress-s1-01785
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] normalise_snippet raises StemStateError for bytes input in strict mode instead of decoding it. It lives in src/slugkit/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 t...
18,284
1,000
36
69
false
synth:9130179261384640915@1
synth
[{"aliases": ["src/slugkit/report.py", "report.py"], "kind": "file", "value": "src/slugkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/slugkit/audit.py", "audit.py"], "kind": "file", "value": "src/slugkit/audit.py"}, {"aliases"...
[{"answers": ["src/slugkit/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01786
corridorkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_leg_once produces doubled dashes for consecutive separators. It lives in src/corridorkit/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 corridorkit layout. [...
69,999
3,072
108
115
false
synth:6824210353893605493@2.01385
synth
[{"aliases": ["src/corridorkit/audit.py", "audit.py"], "kind": "file", "value": "src/corridorkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/waypoint_1.csv", "waypoint_1.csv"], "kind": "file", "value": "data/waypoint_1.csv"}, {"al...
[{"answers": ["src/corridorkit/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-s1-01786-twin
corridorkit
ctx_compress-s1-01786
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_leg_once produces doubled dashes for consecutive separators. It lives in src/corridorkit/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 corridorkit layout. [...
45,709
2,285
66
115
false
synth:6824210353893605493@2.01385
synth
[{"aliases": ["src/corridorkit/audit.py", "audit.py"], "kind": "file", "value": "src/corridorkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/waypoint_1.csv", "waypoint_1.csv"], "kind": "file", "value": "data/waypoint_1.csv"}, {"al...
[{"answers": ["src/corridorkit/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-s1-01787
rosterkit
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_slot misses keys that differ only in case. It lives in src/rosterkit/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 rosterkit layout. [tool_use kind=search name=Ba...
86,928
3,072
119
130
false
synth:201932343895579245@2.26832
synth
[{"aliases": ["src/rosterkit/io.py", "io.py"], "kind": "file", "value": "src/rosterkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/rosterkit/cli.py", "cli.py"], "kind": "file", "value": "src/rosterkit/cli.py"}, {"aliases": ["data/roster_4.csv"...
[{"answers": ["src/rosterkit/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-01787-twin
rosterkit
ctx_compress-s1-01787
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_slot misses keys that differ only in case. It lives in src/rosterkit/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 rosterkit layout. [tool_use kind=search name=Ba...
60,134
3,006
77
130
false
synth:201932343895579245@2.26832
synth
[{"aliases": ["src/rosterkit/io.py", "io.py"], "kind": "file", "value": "src/rosterkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/rosterkit/cli.py", "cli.py"], "kind": "file", "value": "src/rosterkit/cli.py"}, {"aliases": ["data/roster_4.csv"...
[{"answers": ["src/rosterkit/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-01788
calendarflow
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_cadence skips the first page; page 1 returns what should be page 2. It lives in src/calendarflow/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 calendarflo...
113,217
3,072
130
139
false
synth:4554204695119246717@2.20542
synth
[{"aliases": ["src/calendarflow/format.py", "format.py"], "kind": "file", "value": "src/calendarflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/roster_1.csv", "roster_1.csv"], "kind": "file", "value": "data/roster_1.csv"}, {"...
[{"answers": ["src/calendarflow/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": ["pass"], "question": "Did the agent's...
ctx_compress-s1-01788-twin
calendarflow
ctx_compress-s1-01788
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_cadence skips the first page; page 1 returns what should be page 2. It lives in src/calendarflow/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 calendarflo...
55,181
2,759
68
139
false
synth:4554204695119246717@2.20542
synth
[{"aliases": ["src/calendarflow/format.py", "format.py"], "kind": "file", "value": "src/calendarflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/roster_1.csv", "roster_1.csv"], "kind": "file", "value": "data/roster_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": ["data/roster_1.csv", "roster_1.csv"], ...
ctx_compress-s1-01789
sampleforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_sample_safe skips the first page; page 1 returns what should be page 2. It lives in src/sampleforge/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 sampleforge...
43,102
2,155
94
107
false
synth:1679386074221560101@1.37773
synth
[{"aliases": ["src/sampleforge/format.py", "format.py"], "kind": "file", "value": "src/sampleforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/trace_1.csv", "trace_1.csv"], "kind": "file", "value": "data/trace_1.csv"}, {"alias...
[{"answers": ["src/sampleforge/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": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01789-twin
sampleforge
ctx_compress-s1-01789
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_sample_safe skips the first page; page 1 returns what should be page 2. It lives in src/sampleforge/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 sampleforge...
19,911
1,000
51
107
false
synth:1679386074221560101@1.37773
synth
[{"aliases": ["src/sampleforge/format.py", "format.py"], "kind": "file", "value": "src/sampleforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/trace_1.csv", "trace_1.csv"], "kind": "file", "value": "data/trace_1.csv"}, {"alias...
[{"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/trace_1.csv", "trace_1.csv"], "k...
ctx_compress-s1-01790
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] smooth_series_window double counts carried amounts when the period closes on a weekend. It lives in src/spanflow/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 spa...
94,657
3,072
151
166
false
synth:7686857135452520508@2.2845
synth
[{"aliases": ["src/spanflow/compat.py", "compat.py"], "kind": "file", "value": "src/spanflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/series_7.csv", "series_7.csv"], "kind": "file", "value": "data/series_7.csv"}, {"aliases"...
[{"answers": ["src/spanflow/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 mos...
ctx_compress-s1-01790-twin
spanflow
ctx_compress-s1-01790
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_series_window double counts carried amounts when the period closes on a weekend. It lives in src/spanflow/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 spa...
66,145
3,072
99
166
false
synth:7686857135452520508@2.2845
synth
[{"aliases": ["src/spanflow/compat.py", "compat.py"], "kind": "file", "value": "src/spanflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/series_7.csv", "series_7.csv"], "kind": "file", "value": "data/series_7.csv"}, {"aliases"...
[{"answers": ["src/spanflow/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 mos...
ctx_compress-s1-01791
bucketdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_probe misses keys that differ only in case. It lives in src/bucketdesk/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 bucketdesk layout. [tool_use kind=search...
98,599
3,072
104
131
false
synth:7548099288907657716@2.14245
synth
[{"aliases": ["src/bucketdesk/format.py", "format.py"], "kind": "file", "value": "src/bucketdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/bucket_0.csv", "bucket_0.csv"], "kind": "file", "value": "data/bucket_0.csv"}, {"alia...
[{"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/trace_3.csv", "trace_3.csv"], "k...
ctx_compress-s1-01791-twin
bucketdesk
ctx_compress-s1-01791
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_probe misses keys that differ only in case. It lives in src/bucketdesk/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 bucketdesk layout. [tool_use kind=search...
48,969
2,448
60
131
false
synth:7548099288907657716@2.14245
synth
[{"aliases": ["src/bucketdesk/format.py", "format.py"], "kind": "file", "value": "src/bucketdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/bucket_0.csv", "bucket_0.csv"], "kind": "file", "value": "data/bucket_0.csv"}, {"alia...
[{"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_report.py", "tests/test_report.p...
ctx_compress-s1-01792
corridorworks
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] project_tile returns the upper bound for every value instead of clamping into the range. It lives in src/corridorworks/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...
63,545
3,072
97
112
false
synth:6192578664478232228@1.16708
synth
[{"aliases": ["src/corridorworks/store.py", "store.py"], "kind": "file", "value": "src/corridorworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/corridorworks/limits.py", "limits.py"], "kind": "file", "value": "src/corridorworks/li...
[{"answers": ["src/corridorworks/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-s1-01792-twin
corridorworks
ctx_compress-s1-01792
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] project_tile returns the upper bound for every value instead of clamping into the range. It lives in src/corridorworks/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...
29,477
1,473
53
112
false
synth:6192578664478232228@1.16708
synth
[{"aliases": ["src/corridorworks/store.py", "store.py"], "kind": "file", "value": "src/corridorworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/corridorworks/limits.py", "limits.py"], "kind": "file", "value": "src/corridorworks/li...
[{"answers": ["src/corridorworks/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01793
counterworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_series raises TraceRangeError for bytes input in strict mode instead of decoding it. It lives in src/counterworks/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 t...
74,908
3,072
94
101
false
synth:4028614240208793910@1.5326
synth
[{"aliases": ["src/counterworks/limits.py", "limits.py"], "kind": "file", "value": "src/counterworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/counterworks/report.py", "report.py"], "kind": "file", "value": "src/counterworks/...
[{"answers": ["src/counterworks/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-s1-01793-twin
counterworks
ctx_compress-s1-01793
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_series raises TraceRangeError for bytes input in strict mode instead of decoding it. It lives in src/counterworks/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 t...
39,862
1,993
48
101
false
synth:4028614240208793910@1.5326
synth
[{"aliases": ["src/counterworks/limits.py", "limits.py"], "kind": "file", "value": "src/counterworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/counterworks/report.py", "report.py"], "kind": "file", "value": "src/counterworks/...
[{"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/counterworks/report....
ctx_compress-s1-01794
lexemeflow
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_lexeme_batch drops override values of 0, False and empty string. It lives in src/lexemeflow/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 lexemeflow layo...
24,656
1,232
39
50
false
synth:3884692091665235568@1
synth
[{"aliases": ["src/lexemeflow/format.py", "format.py"], "kind": "file", "value": "src/lexemeflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"alia...
[{"answers": ["src/lexemeflow/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 m...
ctx_compress-s1-01795
sectionworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_heading double counts carried amounts when the period closes on a weekend. It lives in src/sectionworks/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 sectionworks ...
33,197
1,659
64
69
false
synth:3357831184320883640@1
synth
[{"aliases": ["src/sectionworks/io.py", "io.py"], "kind": "file", "value": "src/sectionworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/sectionworks/tables.py", "tables.py"], "kind": "file", "value": "src/sectionworks/tables.py"}, {"aliases":...
[{"answers": ["src/sectionworks/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 re...
ctx_compress-s1-01795-twin
sectionworks
ctx_compress-s1-01795
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_heading double counts carried amounts when the period closes on a weekend. It lives in src/sectionworks/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 sectionworks ...
14,922
1,000
36
69
false
synth:3357831184320883640@1
synth
[{"aliases": ["src/sectionworks/io.py", "io.py"], "kind": "file", "value": "src/sectionworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/sectionworks/tables.py", "tables.py"], "kind": "file", "value": "src/sectionworks/tables.py"}, {"aliases":...
[{"answers": ["src/sectionworks/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 re...
ctx_compress-s1-01796
traceledger
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 double counts carried amounts when the period closes on a weekend. It lives in src/traceledger/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 traceledger...
115,350
3,072
129
144
false
synth:4738555942064466778@2.18998
synth
[{"aliases": ["src/traceledger/audit.py", "audit.py"], "kind": "file", "value": "src/traceledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/counter_7.csv", "counter_7.csv"], "kind": "file", "value": "data/counter_7.csv"}, {"alias...
[{"answers": ["src/traceledger/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-s1-01796-twin
traceledger
ctx_compress-s1-01796
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 double counts carried amounts when the period closes on a weekend. It lives in src/traceledger/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 traceledger...
58,185
2,909
73
144
false
synth:4738555942064466778@2.18998
synth
[{"aliases": ["src/traceledger/audit.py", "audit.py"], "kind": "file", "value": "src/traceledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/counter_7.csv", "counter_7.csv"], "kind": "file", "value": "data/counter_7.csv"}, {"alias...
[{"answers": ["src/traceledger/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-s1-01797
nonceledger
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 returns the upper bound for every value instead of clamping into the range. It lives in src/nonceledger/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 non...
91,137
3,072
117
126
false
synth:8660437754074295705@1.81492
synth
[{"aliases": ["src/nonceledger/core.py", "core.py"], "kind": "file", "value": "src/nonceledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/scope_2.csv", "scope_2.csv"], "kind": "file", "value": "data/scope_2.csv"}, {"aliases": ["data/...
[{"answers": ["src/nonceledger/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...
ctx_compress-s1-01797-twin
nonceledger
ctx_compress-s1-01797
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 returns the upper bound for every value instead of clamping into the range. It lives in src/nonceledger/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 non...
63,665
3,072
75
126
false
synth:8660437754074295705@1.81492
synth
[{"aliases": ["src/nonceledger/core.py", "core.py"], "kind": "file", "value": "src/nonceledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/scope_2.csv", "scope_2.csv"], "kind": "file", "value": "data/scope_2.csv"}, {"aliases": ["data/...
[{"answers": ["src/nonceledger/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...
ctx_compress-s1-01798
slotflow
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_slot_window drops override values of 0, False and empty string. It lives in src/slotflow/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 slotflow layout. [too...
97,516
3,072
108
127
false
synth:8413878054802483996@1.92083
synth
[{"aliases": ["src/slotflow/audit.py", "audit.py"], "kind": "file", "value": "src/slotflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/slotflow/schema.py", "schema.py"], "kind": "file", "value": "src/slotflow/schema.py"}, {"aliases...
[{"answers": ["src/slotflow/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-s1-01798-twin
slotflow
ctx_compress-s1-01798
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_slot_window drops override values of 0, False and empty string. It lives in src/slotflow/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 slotflow layout. [too...
55,037
2,751
72
127
false
synth:8413878054802483996@1.92083
synth
[{"aliases": ["src/slotflow/audit.py", "audit.py"], "kind": "file", "value": "src/slotflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/slotflow/schema.py", "schema.py"], "kind": "file", "value": "src/slotflow/schema.py"}, {"aliases...
[{"answers": ["src/slotflow/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-s1-01799
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] settle_refund raises KeyError for a missing key instead of returning the default. It lives in src/balanceforge/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 balan...
90,450
3,072
112
127
false
synth:491571369295497391@1.87367
synth
[{"aliases": ["src/balanceforge/compat.py", "compat.py"], "kind": "file", "value": "src/balanceforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/balanceforge/report.py", "report.py"], "kind": "file", "value": "src/balanceforge/...
[{"answers": ["src/balanceforge/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-s1-01799-twin
balanceforge
ctx_compress-s1-01799
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_refund raises KeyError for a missing key instead of returning the default. It lives in src/balanceforge/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 balan...
40,388
2,019
60
127
false
synth:491571369295497391@1.87367
synth
[{"aliases": ["src/balanceforge/compat.py", "compat.py"], "kind": "file", "value": "src/balanceforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/balanceforge/report.py", "report.py"], "kind": "file", "value": "src/balanceforge/...
[{"answers": ["src/balanceforge/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-s1-01800
stockworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_pallet ignores the cap, so late attempts wait for minutes. It lives in src/stockworks/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 stockworks layout. [tool...
95,137
3,072
108
129
false
synth:8441924224496067748@2.3087
synth
[{"aliases": ["src/stockworks/report.py", "report.py"], "kind": "file", "value": "src/stockworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/sku_1.csv", "sku_1.csv"], "kind": "file", "value": "data/sku_1.csv"}, {"aliases": ["d...
[{"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-s1-01800-twin
stockworks
ctx_compress-s1-01800
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_pallet ignores the cap, so late attempts wait for minutes. It lives in src/stockworks/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 stockworks layout. [tool...
56,951
2,847
67
129
false
synth:8441924224496067748@2.3087
synth
[{"aliases": ["src/stockworks/report.py", "report.py"], "kind": "file", "value": "src/stockworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/sku_1.csv", "sku_1.csv"], "kind": "file", "value": "data/sku_1.csv"}, {"aliases": ["d...
[{"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": ["cli.py", "src/stockworks/cli.py"], "k...
ctx_compress-s1-01801
receiptdesk
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_receipt double counts carried amounts when the period closes on a weekend. It lives in src/receiptdesk/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 rece...
54,372
2,718
98
119
false
synth:3833282334232481257@1.2955
synth
[{"aliases": ["src/receiptdesk/report.py", "report.py"], "kind": "file", "value": "src/receiptdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/invoice_2.csv", "invoice_2.csv"], "kind": "file", "value": "data/invoice_2.csv"}, {...
[{"answers": ["src/receiptdesk/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-s1-01801-twin
receiptdesk
ctx_compress-s1-01801
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_receipt double counts carried amounts when the period closes on a weekend. It lives in src/receiptdesk/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 rece...
32,268
1,613
55
119
false
synth:3833282334232481257@1.2955
synth
[{"aliases": ["src/receiptdesk/report.py", "report.py"], "kind": "file", "value": "src/receiptdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/invoice_2.csv", "invoice_2.csv"], "kind": "file", "value": "data/invoice_2.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": ["report.py", "src/receiptdesk/report.p...
ctx_compress-s1-01802
zoneledger
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_fix produces doubled dashes for consecutive separators. It lives in src/zoneledger/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 zoneledger layout. [tool_use k...
46,208
2,310
76
91
false
synth:7676491404704701218@1.18815
synth
[{"aliases": ["src/zoneledger/rules.py", "rules.py"], "kind": "file", "value": "src/zoneledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/corridor_0.csv", "corridor_0.csv"], "kind": "file", "value": "data/corridor_0.csv"}, {"alia...
[{"answers": ["src/zoneledger/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01802-twin
zoneledger
ctx_compress-s1-01802
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_fix produces doubled dashes for consecutive separators. It lives in src/zoneledger/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 zoneledger layout. [tool_use k...
27,893
1,394
54
91
false
synth:7676491404704701218@1.18815
synth
[{"aliases": ["src/zoneledger/rules.py", "rules.py"], "kind": "file", "value": "src/zoneledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/corridor_0.csv", "corridor_0.csv"], "kind": "file", "value": "data/corridor_0.csv"}, {"alia...
[{"answers": ["src/zoneledger/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-s1-01803
rosterflow
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_safe grows linearly instead of doubling. It lives in src/rosterflow/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 rosterflow layout. [tool_use kind=...
72,271
3,072
86
95
false
synth:2508305625497527654@1.65825
synth
[{"aliases": ["src/rosterflow/store.py", "store.py"], "kind": "file", "value": "src/rosterflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/rosterflow/tables.py", "tables.py"], "kind": "file", "value": "src/rosterflow/tables.py"}, {...
[{"answers": ["src/rosterflow/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 mos...
ctx_compress-s1-01803-twin
rosterflow
ctx_compress-s1-01803
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_safe grows linearly instead of doubling. It lives in src/rosterflow/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 rosterflow layout. [tool_use kind=...
33,974
1,698
50
95
false
synth:2508305625497527654@1.65825
synth
[{"aliases": ["src/rosterflow/store.py", "store.py"], "kind": "file", "value": "src/rosterflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/rosterflow/tables.py", "tables.py"], "kind": "file", "value": "src/rosterflow/tables.py"}, {...
[{"answers": ["src/rosterflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01804
spanworks
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] merge_span raises KeyError for a missing key instead of returning the default. It lives in src/spanworks/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 spanworks l...
70,010
3,072
98
121
false
synth:7572365325780831656@1.8738
synth
[{"aliases": ["src/spanworks/report.py", "report.py"], "kind": "file", "value": "src/spanworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/spanworks/audit.py", "audit.py"], "kind": "file", "value": "src/spanworks/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": ["core.py", "src/spanworks/core.py"], "...
ctx_compress-s1-01804-twin
spanworks
ctx_compress-s1-01804
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] merge_span raises KeyError for a missing key instead of returning the default. It lives in src/spanworks/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 spanworks l...
40,146
2,007
67
121
false
synth:7572365325780831656@1.8738
synth
[{"aliases": ["src/spanworks/report.py", "report.py"], "kind": "file", "value": "src/spanworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/spanworks/audit.py", "audit.py"], "kind": "file", "value": "src/spanworks/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": ["test_compat.py", "tests/test_compat.p...
ctx_compress-s1-01805
sectionflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_margin_strict lets defaults overwrite explicit overrides. It lives in src/sectionflow/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 sectionflow layout. [to...
69,403
3,072
92
111
false
synth:6652118548048247861@1.25158
synth
[{"aliases": ["src/sectionflow/report.py", "report.py"], "kind": "file", "value": "src/sectionflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/sectionflow/io.py", "io.py"], "kind": "file", "value": "src/sectionflow/io.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["schema.py", "src/sectionflow/schema.p...
ctx_compress-s1-01805-twin
sectionflow
ctx_compress-s1-01805
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_margin_strict lets defaults overwrite explicit overrides. It lives in src/sectionflow/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 sectionflow layout. [to...
40,963
2,048
58
111
false
synth:6652118548048247861@1.25158
synth
[{"aliases": ["src/sectionflow/report.py", "report.py"], "kind": "file", "value": "src/sectionflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/sectionflow/io.py", "io.py"], "kind": "file", "value": "src/sectionflow/io.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_audit.py", "tests/test_audit.py"...
ctx_compress-s1-01806
headingdesk
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_revision raises KeyError for a missing key instead of returning the default. It lives in src/headingdesk/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 hea...
71,698
3,072
98
123
false
synth:8187301455761785434@1.6146
synth
[{"aliases": ["src/headingdesk/compat.py", "compat.py"], "kind": "file", "value": "src/headingdesk/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/revision_1.csv", "revision_1.csv"], "kind": "file", "value": "data/revision_1.csv"}...
[{"answers": ["src/headingdesk/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-s1-01806-twin
headingdesk
ctx_compress-s1-01806
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_revision raises KeyError for a missing key instead of returning the default. It lives in src/headingdesk/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 hea...
50,063
2,503
66
123
false
synth:8187301455761785434@1.6146
synth
[{"aliases": ["src/headingdesk/compat.py", "compat.py"], "kind": "file", "value": "src/headingdesk/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/revision_1.csv", "revision_1.csv"], "kind": "file", "value": "data/revision_1.csv"}...
[{"answers": ["src/headingdesk/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-s1-01807
sampleflow
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] merge_sample rejects a range where lo equals hi, which is a valid single point range. It lives in src/sampleflow/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 sam...
26,130
1,306
61
72
false
synth:6036302288495527804@1
synth
[{"aliases": ["src/sampleflow/limits.py", "limits.py"], "kind": "file", "value": "src/sampleflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/sampleflow/schema.py", "schema.py"], "kind": "file", "value": "src/sampleflow/schema.p...
[{"answers": ["src/sampleflow/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-s1-01807-twin
sampleflow
ctx_compress-s1-01807
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] merge_sample rejects a range where lo equals hi, which is a valid single point range. It lives in src/sampleflow/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 sam...
13,451
1,000
35
72
false
synth:6036302288495527804@1
synth
[{"aliases": ["src/sampleflow/limits.py", "limits.py"], "kind": "file", "value": "src/sampleflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/sampleflow/schema.py", "schema.py"], "kind": "file", "value": "src/sampleflow/schema.p...
[{"answers": ["src/sampleflow/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-s1-01808
ledgerworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_tariff lets defaults overwrite explicit overrides. It lives in src/ledgerworks/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 ledgerworks layout. [tool_use kind=sea...
108,882
3,072
103
118
false
synth:2136171342685904758@2.30155
synth
[{"aliases": ["src/ledgerworks/io.py", "io.py"], "kind": "file", "value": "src/ledgerworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/ledgerworks/report.py", "report.py"], "kind": "file", "value": "src/ledgerworks/report.py"}, {"aliases": ["d...
[{"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/ledgerworks/rules.py"...
ctx_compress-s1-01808-twin
ledgerworks
ctx_compress-s1-01808
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_tariff lets defaults overwrite explicit overrides. It lives in src/ledgerworks/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 ledgerworks layout. [tool_use kind=sea...
59,141
2,957
59
118
false
synth:2136171342685904758@2.30155
synth
[{"aliases": ["src/ledgerworks/io.py", "io.py"], "kind": "file", "value": "src/ledgerworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/ledgerworks/report.py", "report.py"], "kind": "file", "value": "src/ledgerworks/report.py"}, {"aliases": ["d...
[{"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/ledgerworks/report.p...