Datasets:
microtensor-io
/
agent-sessions

Dataset card Data Studio Files
xet
 Community
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-s1-01809
marginworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_once raises ValueError when a value itself contains an equals sign. It lives in src/marginworks/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 marginwo...
74,606
3,072
108
117
false
synth:3964428302603970878@1.23755
synth
[{"aliases": ["src/marginworks/core.py", "core.py"], "kind": "file", "value": "src/marginworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/marginworks/rules.py", "rules.py"], "kind": "file", "value": "src/marginworks/rules.py"}, {"alia...
[{"answers": ["src/marginworks/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01809-twin
marginworks
ctx_compress-s1-01809
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_once raises ValueError when a value itself contains an equals sign. It lives in src/marginworks/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 marginwo...
37,640
1,882
60
117
false
synth:3964428302603970878@1.23755
synth
[{"aliases": ["src/marginworks/core.py", "core.py"], "kind": "file", "value": "src/marginworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/marginworks/rules.py", "rules.py"], "kind": "file", "value": "src/marginworks/rules.py"}, {"alia...
[{"answers": ["src/marginworks/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01810
cadencekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/cadencekit/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 cadencekit layout. [tool_use kind=searc...
34,963
1,748
56
61
false
synth:2522291574923700747@1
synth
[{"aliases": ["src/cadencekit/store.py", "store.py"], "kind": "file", "value": "src/cadencekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/slot_3.csv", "slot_3.csv"], "kind": "file", "value": "data/slot_3.csv"}, {"aliases": ["data...
[{"answers": ["src/cadencekit/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-s1-01810-twin
cadencekit
ctx_compress-s1-01810
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/cadencekit/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 cadencekit layout. [tool_use kind=searc...
16,715
1,000
29
61
false
synth:2522291574923700747@1
synth
[{"aliases": ["src/cadencekit/store.py", "store.py"], "kind": "file", "value": "src/cadencekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/slot_3.csv", "slot_3.csv"], "kind": "file", "value": "data/slot_3.csv"}, {"aliases": ["data...
[{"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/cadencekit/schema.py...
ctx_compress-s1-01811
rosterforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_quota_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/rosterforge/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 ro...
56,147
2,807
109
132
false
synth:3602989319796475681@1.42073
synth
[{"aliases": ["src/rosterforge/rules.py", "rules.py"], "kind": "file", "value": "src/rosterforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/rosterforge/io.py", "io.py"], "kind": "file", "value": "src/rosterforge/io.py"}, {"aliases...
[{"answers": ["src/rosterforge/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-s1-01811-twin
rosterforge
ctx_compress-s1-01811
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_quota_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/rosterforge/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 ro...
37,105
1,855
75
132
false
synth:3602989319796475681@1.42073
synth
[{"aliases": ["src/rosterforge/rules.py", "rules.py"], "kind": "file", "value": "src/rosterforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/rosterforge/io.py", "io.py"], "kind": "file", "value": "src/rosterforge/io.py"}, {"aliases...
[{"answers": ["src/rosterforge/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-s1-01812
tokenledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_session_strict raises PolicyStateError for bytes input in strict mode instead of decoding it. It lives in src/tokenledger/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...
105,368
3,072
112
129
false
synth:7121957117310426445@2.22107
synth
[{"aliases": ["src/tokenledger/format.py", "format.py"], "kind": "file", "value": "src/tokenledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/scope_11.csv", "scope_11.csv"], "kind": "file", "value": "data/scope_11.csv"}, {"al...
[{"answers": ["src/tokenledger/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-s1-01812-twin
tokenledger
ctx_compress-s1-01812
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_session_strict raises PolicyStateError for bytes input in strict mode instead of decoding it. It lives in src/tokenledger/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...
54,609
2,730
64
129
false
synth:7121957117310426445@2.22107
synth
[{"aliases": ["src/tokenledger/format.py", "format.py"], "kind": "file", "value": "src/tokenledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/scope_11.csv", "scope_11.csv"], "kind": "file", "value": "data/scope_11.csv"}, {"al...
[{"answers": ["src/tokenledger/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-s1-01813
probeledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe double counts carried amounts when the period closes on a weekend. It lives in src/probeledger/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 probeledger ...
118,522
3,072
130
151
false
synth:2099891118080854689@2.34207
synth
[{"aliases": ["src/probeledger/io.py", "io.py"], "kind": "file", "value": "src/probeledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/probeledger/tables.py", "tables.py"], "kind": "file", "value": "src/probeledger/tables.py"}, {"aliases": ["d...
[{"answers": ["src/probeledger/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-01813-twin
probeledger
ctx_compress-s1-01813
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe double counts carried amounts when the period closes on a weekend. It lives in src/probeledger/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 probeledger ...
82,510
3,072
88
151
false
synth:2099891118080854689@2.34207
synth
[{"aliases": ["src/probeledger/io.py", "io.py"], "kind": "file", "value": "src/probeledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/probeledger/tables.py", "tables.py"], "kind": "file", "value": "src/probeledger/tables.py"}, {"aliases": ["d...
[{"answers": ["src/probeledger/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-01814
counterledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe averages one value too many once the window is full. It lives in src/counterledger/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 counterledger layout. [...
54,294
2,714
103
118
false
synth:2631924441518047046@1.21035
synth
[{"aliases": ["src/counterledger/store.py", "store.py"], "kind": "file", "value": "src/counterledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/counterledger/limits.py", "limits.py"], "kind": "file", "value": "src/counterledger/li...
[{"answers": ["src/counterledger/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-s1-01814-twin
counterledger
ctx_compress-s1-01814
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe averages one value too many once the window is full. It lives in src/counterledger/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 counterledger layout. [...
29,389
1,469
67
118
false
synth:2631924441518047046@1.21035
synth
[{"aliases": ["src/counterledger/store.py", "store.py"], "kind": "file", "value": "src/counterledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/counterledger/limits.py", "limits.py"], "kind": "file", "value": "src/counterledger/li...
[{"answers": ["src/counterledger/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-s1-01815
bookingflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_booking produces doubled dashes for consecutive separators. It lives in src/bookingflow/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 bookingflow layout. [tool_us...
44,128
2,206
72
81
false
synth:1085470268785504694@1
synth
[{"aliases": ["src/bookingflow/io.py", "io.py"], "kind": "file", "value": "src/bookingflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/bookingflow/schema.py", "schema.py"], "kind": "file", "value": "src/bookingflow/schema.py"}, {"aliases": ["s...
[{"answers": ["src/bookingflow/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rec...
ctx_compress-s1-01815-twin
bookingflow
ctx_compress-s1-01815
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_booking produces doubled dashes for consecutive separators. It lives in src/bookingflow/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 bookingflow layout. [tool_us...
19,873
1,000
44
81
false
synth:1085470268785504694@1
synth
[{"aliases": ["src/bookingflow/io.py", "io.py"], "kind": "file", "value": "src/bookingflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/bookingflow/schema.py", "schema.py"], "kind": "file", "value": "src/bookingflow/schema.py"}, {"aliases": ["s...
[{"answers": ["src/bookingflow/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rec...
ctx_compress-s1-01816
stockkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_pallet raises KeyError for a missing key instead of returning the default. It lives in src/stockkit/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 stockkit lay...
35,991
1,799
92
113
false
synth:1941767649094519254@1
synth
[{"aliases": ["src/stockkit/format.py", "format.py"], "kind": "file", "value": "src/stockkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/lot_2.csv", "lot_2.csv"], "kind": "file", "value": "data/lot_2.csv"}, {"aliases": ["src/s...
[{"answers": ["src/stockkit/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 mos...
ctx_compress-s1-01816-twin
stockkit
ctx_compress-s1-01816
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_pallet raises KeyError for a missing key instead of returning the default. It lives in src/stockkit/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 stockkit lay...
23,676
1,183
60
113
false
synth:1941767649094519254@1
synth
[{"aliases": ["src/stockkit/format.py", "format.py"], "kind": "file", "value": "src/stockkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/lot_2.csv", "lot_2.csv"], "kind": "file", "value": "data/lot_2.csv"}, {"aliases": ["src/s...
[{"answers": ["src/stockkit/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01817
nonceflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_policy_all averages one value too many once the window is full. It lives in src/nonceflow/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the nonceflow layout. [tool...
58,817
2,940
86
109
false
synth:2127231522957414347@1.42542
synth
[{"aliases": ["src/nonceflow/core.py", "core.py"], "kind": "file", "value": "src/nonceflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/claim_4.csv", "claim_4.csv"], "kind": "file", "value": "data/claim_4.csv"}, {"aliases": ["data/sess...
[{"answers": ["src/nonceflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most r...
ctx_compress-s1-01817-twin
nonceflow
ctx_compress-s1-01817
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_policy_all averages one value too many once the window is full. It lives in src/nonceflow/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the nonceflow layout. [tool...
39,739
1,986
58
109
false
synth:2127231522957414347@1.42542
synth
[{"aliases": ["src/nonceflow/core.py", "core.py"], "kind": "file", "value": "src/nonceflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/claim_4.csv", "claim_4.csv"], "kind": "file", "value": "data/claim_4.csv"}, {"aliases": ["data/sess...
[{"answers": ["src/nonceflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-01818
skuforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse_all raises SkuShapeError for bytes input in strict mode instead of decoding it. It lives in src/skuforge/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...
27,008
1,350
49
56
false
synth:6712363179329002575@1
synth
[{"aliases": ["src/skuforge/store.py", "store.py"], "kind": "file", "value": "src/skuforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/skuforge/schema.py", "schema.py"], "kind": "file", "value": "src/skuforge/schema.py"}, {"aliases...
[{"answers": ["src/skuforge/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 ...
ctx_compress-s1-01818-twin
skuforge
ctx_compress-s1-01818
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse_all raises SkuShapeError for bytes input in strict mode instead of decoding it. It lives in src/skuforge/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...
11,413
1,000
29
56
false
synth:6712363179329002575@1
synth
[{"aliases": ["src/skuforge/store.py", "store.py"], "kind": "file", "value": "src/skuforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/skuforge/schema.py", "schema.py"], "kind": "file", "value": "src/skuforge/schema.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": ["test_rules.py", "tests/test_rules.py"...
ctx_compress-s1-01819
revisiondesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_margin lets defaults overwrite explicit overrides. It lives in src/revisiondesk/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 revisiondesk layout. [tool_use ki...
26,133
1,306
41
48
false
synth:5728051284414373631@1
synth
[{"aliases": ["src/revisiondesk/store.py", "store.py"], "kind": "file", "value": "src/revisiondesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/clause_0.csv", "clause_0.csv"], "kind": "file", "value": "data/clause_0.csv"}, {"aliase...
[{"answers": ["src/revisiondesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01819-twin
revisiondesk
ctx_compress-s1-01819
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_margin lets defaults overwrite explicit overrides. It lives in src/revisiondesk/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 revisiondesk layout. [tool_use ki...
16,757
1,000
23
48
false
synth:5728051284414373631@1
synth
[{"aliases": ["src/revisiondesk/store.py", "store.py"], "kind": "file", "value": "src/revisiondesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/clause_0.csv", "clause_0.csv"], "kind": "file", "value": "data/clause_0.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["core.py", "src/revisiondesk/core.py"]...
ctx_compress-s1-01820
tokendesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_principal ignores the cap, so late attempts wait for minutes. It lives in src/tokendesk/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 tokendesk layout. [tool...
47,774
2,388
78
83
false
synth:98061086701834182@1
synth
[{"aliases": ["src/tokendesk/audit.py", "audit.py"], "kind": "file", "value": "src/tokendesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/session_0.csv", "session_0.csv"], "kind": "file", "value": "data/session_0.csv"}, {"aliases":...
[{"answers": ["src/tokendesk/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01820-twin
tokendesk
ctx_compress-s1-01820
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_principal ignores the cap, so late attempts wait for minutes. It lives in src/tokendesk/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 tokendesk layout. [tool...
28,594
1,429
48
83
false
synth:98061086701834182@1
synth
[{"aliases": ["src/tokendesk/audit.py", "audit.py"], "kind": "file", "value": "src/tokendesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/session_0.csv", "session_0.csv"], "kind": "file", "value": "data/session_0.csv"}, {"aliases":...
[{"answers": ["src/tokendesk/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-s1-01821
cueworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_stem skips the first page; page 1 returns what should be page 2. It lives in src/cueworks/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 cueworks layout. [t...
60,141
3,007
102
121
false
synth:7834791076252067346@1.1456
synth
[{"aliases": ["src/cueworks/report.py", "report.py"], "kind": "file", "value": "src/cueworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/cueworks/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 mos...
ctx_compress-s1-01821-twin
cueworks
ctx_compress-s1-01821
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_stem skips the first page; page 1 returns what should be page 2. It lives in src/cueworks/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 cueworks layout. [t...
36,496
1,824
68
121
false
synth:7834791076252067346@1.1456
synth
[{"aliases": ["src/cueworks/report.py", "report.py"], "kind": "file", "value": "src/cueworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/cueworks/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 mos...
ctx_compress-s1-01822
tariffdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_balance raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/tariffdesk/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 tari...
92,524
3,072
99
112
false
synth:6960022575857219384@1.70325
synth
[{"aliases": ["src/tariffdesk/store.py", "store.py"], "kind": "file", "value": "src/tariffdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/tariffdesk/schema.py", "schema.py"], "kind": "file", "value": "src/tariffdesk/schema.py"}, {...
[{"answers": ["src/tariffdesk/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-s1-01822-twin
tariffdesk
ctx_compress-s1-01822
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_balance raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/tariffdesk/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 tari...
49,583
2,479
59
112
false
synth:6960022575857219384@1.70325
synth
[{"aliases": ["src/tariffdesk/store.py", "store.py"], "kind": "file", "value": "src/tariffdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/tariffdesk/schema.py", "schema.py"], "kind": "file", "value": "src/tariffdesk/schema.py"}, {...
[{"answers": ["src/tariffdesk/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-s1-01823
balanceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_credit raises BalanceStateError for bytes input in strict mode instead of decoding it. 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 map...
55,033
2,751
97
122
false
synth:210569254913392158@1
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": ["data/balance_3.csv", "balance_3.csv"], "kind": "file", "value": "data/balance_3.csv"},...
[{"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-s1-01823-twin
balanceforge
ctx_compress-s1-01823
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 raises BalanceStateError for bytes input in strict mode instead of decoding it. 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 map...
36,790
1,839
69
122
false
synth:210569254913392158@1
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": ["data/balance_3.csv", "balance_3.csv"], "kind": "file", "value": "data/balance_3.csv"},...
[{"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-s1-01824
tariffworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_payout produces doubled dashes for consecutive separators. It lives in src/tariffworks/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 tariffworks layout. [t...
40,397
2,019
74
79
false
synth:7916352135919167647@1
synth
[{"aliases": ["src/tariffworks/format.py", "format.py"], "kind": "file", "value": "src/tariffworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/credit_0.csv", "credit_0.csv"], "kind": "file", "value": "data/credit_0.csv"}, {"al...
[{"answers": ["src/tariffworks/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-s1-01824-twin
tariffworks
ctx_compress-s1-01824
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_payout produces doubled dashes for consecutive separators. It lives in src/tariffworks/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 tariffworks layout. [t...
24,947
1,247
44
79
false
synth:7916352135919167647@1
synth
[{"aliases": ["src/tariffworks/format.py", "format.py"], "kind": "file", "value": "src/tariffworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/credit_0.csv", "credit_0.csv"], "kind": "file", "value": "data/credit_0.csv"}, {"al...
[{"answers": ["src/tariffworks/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-s1-01825
receiptledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 rejects a range where lo equals hi, which is a valid single point range. It lives in src/receiptledger/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...
45,988
2,299
80
97
false
synth:2531590650706314303@1
synth
[{"aliases": ["src/receiptledger/format.py", "format.py"], "kind": "file", "value": "src/receiptledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/tariff_0.csv", "tariff_0.csv"], "kind": "file", "value": "data/tariff_0.csv"}, ...
[{"answers": ["src/receiptledger/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'...
ctx_compress-s1-01825-twin
receiptledger
ctx_compress-s1-01825
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 rejects a range where lo equals hi, which is a valid single point range. It lives in src/receiptledger/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...
35,634
1,781
56
97
false
synth:2531590650706314303@1
synth
[{"aliases": ["src/receiptledger/format.py", "format.py"], "kind": "file", "value": "src/receiptledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/tariff_0.csv", "tariff_0.csv"], "kind": "file", "value": "data/tariff_0.csv"}, ...
[{"answers": ["src/receiptledger/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'...
ctx_compress-s1-01826
binworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_pallet_all drops override values of 0, False and empty string. It lives in src/binworks/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 binworks layout. [tool_u...
62,676
3,072
110
121
false
synth:2647411976810793336@1.50012
synth
[{"aliases": ["src/binworks/limits.py", "limits.py"], "kind": "file", "value": "src/binworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/binworks/store.py", "store.py"], "kind": "file", "value": "src/binworks/store.py"}, {"alia...
[{"answers": ["src/binworks/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01826-twin
binworks
ctx_compress-s1-01826
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_pallet_all drops override values of 0, False and empty string. It lives in src/binworks/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 binworks layout. [tool_u...
35,276
1,763
58
121
false
synth:2647411976810793336@1.50012
synth
[{"aliases": ["src/binworks/limits.py", "limits.py"], "kind": "file", "value": "src/binworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/binworks/store.py", "store.py"], "kind": "file", "value": "src/binworks/store.py"}, {"alia...
[{"answers": ["src/binworks/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01827
balanceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_tariff_window double counts carried amounts when the period closes on a weekend. It lives in src/balanceforge/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 balan...
68,597
3,072
113
122
false
synth:4571895925698736470@1.18952
synth
[{"aliases": ["src/balanceforge/io.py", "io.py"], "kind": "file", "value": "src/balanceforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/balanceforge/core.py", "core.py"], "kind": "file", "value": "src/balanceforge/core.py"}, {"aliases": ["src...
[{"answers": ["src/balanceforge/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-s1-01827-twin
balanceforge
ctx_compress-s1-01827
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_tariff_window double counts carried amounts when the period closes on a weekend. It lives in src/balanceforge/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 balan...
32,248
1,612
63
122
false
synth:4571895925698736470@1.18952
synth
[{"aliases": ["src/balanceforge/io.py", "io.py"], "kind": "file", "value": "src/balanceforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/balanceforge/core.py", "core.py"], "kind": "file", "value": "src/balanceforge/core.py"}, {"aliases": ["src...
[{"answers": ["src/balanceforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-01828
sessionflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_principal_all raises PolicyShapeError for bytes input in strict mode instead of decoding it. It lives in src/sessionflow/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 ...
46,529
2,326
92
115
false
synth:6055705350853687455@1.44775
synth
[{"aliases": ["src/sessionflow/report.py", "report.py"], "kind": "file", "value": "src/sessionflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/scope_0.csv", "scope_0.csv"], "kind": "file", "value": "data/scope_0.csv"}, {"alias...
[{"answers": ["src/sessionflow/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-s1-01828-twin
sessionflow
ctx_compress-s1-01828
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_principal_all raises PolicyShapeError for bytes input in strict mode instead of decoding it. It lives in src/sessionflow/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 ...
31,692
1,584
59
115
false
synth:6055705350853687455@1.44775
synth
[{"aliases": ["src/sessionflow/report.py", "report.py"], "kind": "file", "value": "src/sessionflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/scope_0.csv", "scope_0.csv"], "kind": "file", "value": "data/scope_0.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": ["src/sessionflow/store.py", "store.py"...
ctx_compress-s1-01829
tariffledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_ledger_safe returns one item too few on every page. It lives in src/tariffledger/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 tariffledger layout. [tool_use k...
94,821
3,072
108
135
false
synth:2040735439840141823@1.85783
synth
[{"aliases": ["src/tariffledger/audit.py", "audit.py"], "kind": "file", "value": "src/tariffledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/invoice_12.csv", "invoice_12.csv"], "kind": "file", "value": "data/invoice_12.csv"}, {"...
[{"answers": ["src/tariffledger/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01829-twin
tariffledger
ctx_compress-s1-01829
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_ledger_safe returns one item too few on every page. It lives in src/tariffledger/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 tariffledger layout. [tool_use k...
62,653
3,072
74
135
false
synth:2040735439840141823@1.85783
synth
[{"aliases": ["src/tariffledger/audit.py", "audit.py"], "kind": "file", "value": "src/tariffledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/invoice_12.csv", "invoice_12.csv"], "kind": "file", "value": "data/invoice_12.csv"}, {"...
[{"answers": ["src/tariffledger/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01830
headingworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_draft averages one value too many once the window is full. It lives in src/headingworks/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 headingworks layout. ...
95,303
3,072
94
103
false
synth:3894778603409789130@1.75445
synth
[{"aliases": ["src/headingworks/limits.py", "limits.py"], "kind": "file", "value": "src/headingworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/clause_2.csv", "clause_2.csv"], "kind": "file", "value": "data/clause_2.csv"}, {"...
[{"answers": ["src/headingworks/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-s1-01830-twin
headingworks
ctx_compress-s1-01830
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_draft averages one value too many once the window is full. It lives in src/headingworks/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 headingworks layout. ...
44,106
2,205
49
103
false
synth:3894778603409789130@1.75445
synth
[{"aliases": ["src/headingworks/limits.py", "limits.py"], "kind": "file", "value": "src/headingworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/clause_2.csv", "clause_2.csv"], "kind": "file", "value": "data/clause_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": ["clause_2.csv", "data/clause_2.csv"], ...
ctx_compress-s1-01831
headingflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_index skips the first page; page 1 returns what should be page 2. It lives in src/headingflow/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 headingflow layout. ...
51,572
2,578
71
90
false
synth:3736887309683218813@1.01345
synth
[{"aliases": ["src/headingflow/cli.py", "cli.py"], "kind": "file", "value": "src/headingflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/revision_3.csv", "revision_3.csv"], "kind": "file", "value": "data/revision_3.csv"}, {"aliases": ["da...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/headingflow/compat.p...
ctx_compress-s1-01831-twin
headingflow
ctx_compress-s1-01831
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_index skips the first page; page 1 returns what should be page 2. It lives in src/headingflow/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 headingflow layout. ...
32,773
1,638
51
90
false
synth:3736887309683218813@1.01345
synth
[{"aliases": ["src/headingflow/cli.py", "cli.py"], "kind": "file", "value": "src/headingflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/revision_3.csv", "revision_3.csv"], "kind": "file", "value": "data/revision_3.csv"}, {"aliases": ["da...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/revision_4.csv", "revision_4.csv...
ctx_compress-s1-01832
lexemeledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_page drops override values of 0, False and empty string. It lives in src/lexemeledger/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 lexemeledger layout...
67,885
3,072
95
102
false
synth:8353798649590401574@1.93885
synth
[{"aliases": ["src/lexemeledger/core.py", "core.py"], "kind": "file", "value": "src/lexemeledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/cue_2.csv", "cue_2.csv"], "kind": "file", "value": "data/cue_2.csv"}, {"aliases": ["data/chun...
[{"answers": ["src/lexemeledger/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01832-twin
lexemeledger
ctx_compress-s1-01832
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_page drops override values of 0, False and empty string. It lives in src/lexemeledger/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 lexemeledger layout...
40,158
2,007
53
102
false
synth:8353798649590401574@1.93885
synth
[{"aliases": ["src/lexemeledger/core.py", "core.py"], "kind": "file", "value": "src/lexemeledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/cue_2.csv", "cue_2.csv"], "kind": "file", "value": "data/cue_2.csv"}, {"aliases": ["data/chun...
[{"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": ["cue_5.csv", "data/cue_5.csv"], "kind"...
ctx_compress-s1-01833
stockledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_reorder drops override values of 0, False and empty string. It lives in src/stockledger/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 stockledger layout. [tool_use ...
80,154
3,072
116
143
false
synth:5643882780552555965@2.11247
synth
[{"aliases": ["src/stockledger/cli.py", "cli.py"], "kind": "file", "value": "src/stockledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/stockledger/report.py", "report.py"], "kind": "file", "value": "src/stockledger/report.py"}, {"aliases...
[{"answers": ["src/stockledger/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most r...
ctx_compress-s1-01833-twin
stockledger
ctx_compress-s1-01833
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_reorder drops override values of 0, False and empty string. It lives in src/stockledger/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 stockledger layout. [tool_use ...
44,497
2,224
71
143
false
synth:5643882780552555965@2.11247
synth
[{"aliases": ["src/stockledger/cli.py", "cli.py"], "kind": "file", "value": "src/stockledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/stockledger/report.py", "report.py"], "kind": "file", "value": "src/stockledger/report.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": ["test_report.py", "tests/test_report.p...
ctx_compress-s1-01834
draftforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_clause_page raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/draftforge/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 draf...
64,939
3,072
88
101
false
synth:1519539234682449632@1.4071
synth
[{"aliases": ["src/draftforge/core.py", "core.py"], "kind": "file", "value": "src/draftforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/revision_0.csv", "revision_0.csv"], "kind": "file", "value": "data/revision_0.csv"}, {"aliases": ...
[{"answers": ["src/draftforge/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-s1-01834-twin
draftforge
ctx_compress-s1-01834
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_page raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/draftforge/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 draf...
41,420
2,071
56
101
false
synth:1519539234682449632@1.4071
synth
[{"aliases": ["src/draftforge/core.py", "core.py"], "kind": "file", "value": "src/draftforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/revision_0.csv", "revision_0.csv"], "kind": "file", "value": "data/revision_0.csv"}, {"aliases": ...
[{"answers": ["src/draftforge/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-01835
bookingdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_cadence_window returns one item too few on every page. It lives in src/bookingdesk/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 bookingdesk layout. [tool_use ...
25,935
1,296
56
65
false
synth:3014526114974779439@1
synth
[{"aliases": ["src/bookingdesk/core.py", "core.py"], "kind": "file", "value": "src/bookingdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/slot_1.csv", "slot_1.csv"], "kind": "file", "value": "data/slot_1.csv"}, {"aliases": ["src/book...
[{"answers": ["src/bookingdesk/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-s1-01835-twin
bookingdesk
ctx_compress-s1-01835
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cadence_window returns one item too few on every page. It lives in src/bookingdesk/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 bookingdesk layout. [tool_use ...
12,300
1,000
34
65
false
synth:3014526114974779439@1
synth
[{"aliases": ["src/bookingdesk/core.py", "core.py"], "kind": "file", "value": "src/bookingdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/slot_1.csv", "slot_1.csv"], "kind": "file", "value": "data/slot_1.csv"}, {"aliases": ["src/book...
[{"answers": ["src/bookingdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01836
seriesworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_series returns the upper bound for every value instead of clamping into the range. It lives in src/seriesworks/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 ser...
85,464
3,072
112
131
false
synth:3569266061941546386@1.8573
synth
[{"aliases": ["src/seriesworks/store.py", "store.py"], "kind": "file", "value": "src/seriesworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/gauge_8.csv", "gauge_8.csv"], "kind": "file", "value": "data/gauge_8.csv"}, {"aliases": [...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/trace_3.csv", "trace_3.csv"], "k...
ctx_compress-s1-01836-twin
seriesworks
ctx_compress-s1-01836
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_series returns the upper bound for every value instead of clamping into the range. It lives in src/seriesworks/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 ser...
52,081
2,604
78
131
false
synth:3569266061941546386@1.8573
synth
[{"aliases": ["src/seriesworks/store.py", "store.py"], "kind": "file", "value": "src/seriesworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/gauge_8.csv", "gauge_8.csv"], "kind": "file", "value": "data/gauge_8.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": ["core.py", "src/seriesworks/core.py"],...
ctx_compress-s1-01837
counterkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter ignores the cap, so late attempts wait for minutes. It lives in src/counterkit/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 counterkit layout. [to...
54,442
2,722
94
113
false
synth:317444618449363686@1.4294
synth
[{"aliases": ["src/counterkit/format.py", "format.py"], "kind": "file", "value": "src/counterkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/sample_4.csv", "sample_4.csv"], "kind": "file", "value": "data/sample_4.csv"}, {"alia...
[{"answers": ["src/counterkit/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 m...
ctx_compress-s1-01837-twin
counterkit
ctx_compress-s1-01837
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter ignores the cap, so late attempts wait for minutes. It lives in src/counterkit/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 counterkit layout. [to...
29,966
1,498
51
113
false
synth:317444618449363686@1.4294
synth
[{"aliases": ["src/counterkit/format.py", "format.py"], "kind": "file", "value": "src/counterkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/sample_4.csv", "sample_4.csv"], "kind": "file", "value": "data/sample_4.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_cli.py", "tests/test_cli.py"], "...
ctx_compress-s1-01838
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] paginate_section raises KeyError for a missing key instead of returning the default. It lives in src/headingdesk/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 headi...
38,805
1,940
67
82
false
synth:6407520747064291695@1
synth
[{"aliases": ["src/headingdesk/store.py", "store.py"], "kind": "file", "value": "src/headingdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/headingdesk/cli.py", "cli.py"], "kind": "file", "value": "src/headingdesk/cli.py"}, {"alia...
[{"answers": ["src/headingdesk/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-s1-01838-twin
headingdesk
ctx_compress-s1-01838
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_section raises KeyError for a missing key instead of returning the default. It lives in src/headingdesk/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 headi...
18,668
1,000
41
82
false
synth:6407520747064291695@1
synth
[{"aliases": ["src/headingdesk/store.py", "store.py"], "kind": "file", "value": "src/headingdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/headingdesk/cli.py", "cli.py"], "kind": "file", "value": "src/headingdesk/cli.py"}, {"alia...
[{"answers": ["src/headingdesk/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-s1-01839
sessionkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/sessionkit/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 sessio...
73,929
3,072
110
123
false
synth:1151868895669448234@1.34193
synth
[{"aliases": ["src/sessionkit/audit.py", "audit.py"], "kind": "file", "value": "src/sessionkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/sessionkit/io.py", "io.py"], "kind": "file", "value": "src/sessionkit/io.py"}, {"aliases": [...
[{"answers": ["src/sessionkit/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-01839-twin
sessionkit
ctx_compress-s1-01839
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/sessionkit/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 sessio...
32,423
1,621
57
123
false
synth:1151868895669448234@1.34193
synth
[{"aliases": ["src/sessionkit/audit.py", "audit.py"], "kind": "file", "value": "src/sessionkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/sessionkit/io.py", "io.py"], "kind": "file", "value": "src/sessionkit/io.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/sessionkit/audit.py"]...
ctx_compress-s1-01840
skudesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_reorder double counts carried amounts when the period closes on a weekend. It lives in src/skudesk/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 skudesk layout. ...
54,389
2,719
98
111
false
synth:5628601813957791449@1.4139
synth
[{"aliases": ["src/skudesk/io.py", "io.py"], "kind": "file", "value": "src/skudesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/skudesk/schema.py", "schema.py"], "kind": "file", "value": "src/skudesk/schema.py"}, {"aliases": ["src/skudesk/table...
[{"answers": ["src/skudesk/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-s1-01840-twin
skudesk
ctx_compress-s1-01840
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_reorder double counts carried amounts when the period closes on a weekend. It lives in src/skudesk/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 skudesk layout. ...
28,558
1,427
51
111
false
synth:5628601813957791449@1.4139
synth
[{"aliases": ["src/skudesk/io.py", "io.py"], "kind": "file", "value": "src/skudesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/skudesk/schema.py", "schema.py"], "kind": "file", "value": "src/skudesk/schema.py"}, {"aliases": ["src/skudesk/table...
[{"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/skudesk/schema.py"],...
ctx_compress-s1-01841
tariffledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/tariffledger/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 tariff...
84,607
3,072
102
125
false
synth:1746308403337505309@1.93207
synth
[{"aliases": ["src/tariffledger/core.py", "core.py"], "kind": "file", "value": "src/tariffledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/tariffledger/format.py", "format.py"], "kind": "file", "value": "src/tariffledger/format.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/tariffledger/store.py", "store.py...
ctx_compress-s1-01841-twin
tariffledger
ctx_compress-s1-01841
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/tariffledger/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 tariff...
47,384
2,369
64
125
false
synth:1746308403337505309@1.93207
synth
[{"aliases": ["src/tariffledger/core.py", "core.py"], "kind": "file", "value": "src/tariffledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/tariffledger/format.py", "format.py"], "kind": "file", "value": "src/tariffledger/format.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_format.py", "tests/test_format.p...
ctx_compress-s1-01842
headingworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_index_page raises FootnoteShapeError for bytes input in strict mode instead of decoding it. It lives in src/headingworks/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 mappin...
95,966
3,072
90
105
false
synth:5220115591502771107@1.54985
synth
[{"aliases": ["src/headingworks/io.py", "io.py"], "kind": "file", "value": "src/headingworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/headingworks/audit.py", "audit.py"], "kind": "file", "value": "src/headingworks/audit.py"}, {"aliases": ["...
[{"answers": ["src/headingworks/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-s1-01842-twin
headingworks
ctx_compress-s1-01842
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_index_page raises FootnoteShapeError for bytes input in strict mode instead of decoding it. It lives in src/headingworks/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 mappin...
54,981
2,749
54
105
false
synth:5220115591502771107@1.54985
synth
[{"aliases": ["src/headingworks/io.py", "io.py"], "kind": "file", "value": "src/headingworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/headingworks/audit.py", "audit.py"], "kind": "file", "value": "src/headingworks/audit.py"}, {"aliases": ["...
[{"answers": ["src/headingworks/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-s1-01843
creditflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_payout raises TariffBoundError for bytes input in strict mode instead of decoding it. It lives in src/creditflow/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 th...
55,149
2,757
84
105
false
synth:1322840254220557262@1.0198
synth
[{"aliases": ["src/creditflow/core.py", "core.py"], "kind": "file", "value": "src/creditflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/creditflow/rules.py", "rules.py"], "kind": "file", "value": "src/creditflow/rules.py"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/creditflow/store.py", "store.py"]...
ctx_compress-s1-01843-twin
creditflow
ctx_compress-s1-01843
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_payout raises TariffBoundError for bytes input in strict mode instead of decoding it. It lives in src/creditflow/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 th...
34,257
1,712
56
105
false
synth:1322840254220557262@1.0198
synth
[{"aliases": ["src/creditflow/core.py", "core.py"], "kind": "file", "value": "src/creditflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/creditflow/rules.py", "rules.py"], "kind": "file", "value": "src/creditflow/rules.py"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_report.py", "tests/test_report.p...
ctx_compress-s1-01844
gaugekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 raises TraceStateError for bytes input in strict mode instead of decoding it. It lives in src/gaugekit/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 gaugek...
28,012
1,400
44
51
false
synth:7562685427169795414@1
synth
[{"aliases": ["src/gaugekit/core.py", "core.py"], "kind": "file", "value": "src/gaugekit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/gaugekit/limits.py", "limits.py"], "kind": "file", "value": "src/gaugekit/limits.py"}, {"aliases": ["s...
[{"answers": ["src/gaugekit/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most re...
ctx_compress-s1-01844-twin
gaugekit
ctx_compress-s1-01844
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 raises TraceStateError for bytes input in strict mode instead of decoding it. It lives in src/gaugekit/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 gaugek...
13,987
1,000
27
51
false
synth:7562685427169795414@1
synth
[{"aliases": ["src/gaugekit/core.py", "core.py"], "kind": "file", "value": "src/gaugekit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/gaugekit/limits.py", "limits.py"], "kind": "file", "value": "src/gaugekit/limits.py"}, {"aliases": ["s...
[{"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/gaugekit/tables.py", "tables.py"]...
ctx_compress-s1-01845
stockforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 misses keys that differ only in case. It lives in src/stockforge/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 stockforge layout. [tool_use kind=search name=...
56,343
2,817
71
86
false
synth:1768570786802312481@1.0609
synth
[{"aliases": ["src/stockforge/io.py", "io.py"], "kind": "file", "value": "src/stockforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/stockforge/rules.py", "rules.py"], "kind": "file", "value": "src/stockforge/rules.py"}, {"aliases": ["docs/cha...
[{"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/sku_4.csv", "sku_4.csv"], "kind"...
ctx_compress-s1-01845-twin
stockforge
ctx_compress-s1-01845
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 misses keys that differ only in case. It lives in src/stockforge/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 stockforge layout. [tool_use kind=search name=...
26,429
1,321
39
86
false
synth:1768570786802312481@1.0609
synth
[{"aliases": ["src/stockforge/io.py", "io.py"], "kind": "file", "value": "src/stockforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/stockforge/rules.py", "rules.py"], "kind": "file", "value": "src/stockforge/rules.py"}, {"aliases": ["docs/cha...
[{"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/sku_4.csv", "sku_4.csv"], "kind"...
ctx_compress-s1-01846
windowdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_strict grows linearly instead of doubling. It lives in src/windowdesk/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 windowdesk layout. [tool_use ki...
98,296
3,072
98
125
false
synth:2186970875851496611@1.99505
synth
[{"aliases": ["src/windowdesk/report.py", "report.py"], "kind": "file", "value": "src/windowdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/windowdesk/cli.py", "cli.py"], "kind": "file", "value": "src/windowdesk/cli.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": ["test_format.py", "tests/test_format.p...
ctx_compress-s1-01846-twin
windowdesk
ctx_compress-s1-01846
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_strict grows linearly instead of doubling. It lives in src/windowdesk/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 windowdesk layout. [tool_use ki...
60,357
3,017
70
125
false
synth:2186970875851496611@1.99505
synth
[{"aliases": ["src/windowdesk/report.py", "report.py"], "kind": "file", "value": "src/windowdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/windowdesk/cli.py", "cli.py"], "kind": "file", "value": "src/windowdesk/cli.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": ["schema.py", "src/windowdesk/schema.py...
ctx_compress-s1-01847
sampledesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_sample leaves a trailing dash when the title ends in punctuation. It lives in src/sampledesk/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 sampledesk l...
94,610
3,072
117
136
false
synth:8819768766556240336@1.88563
synth
[{"aliases": ["src/sampledesk/format.py", "format.py"], "kind": "file", "value": "src/sampledesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/sampledesk/tables.py", "tables.py"], "kind": "file", "value": "src/sampledesk/tables.p...
[{"answers": ["src/sampledesk/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 m...
ctx_compress-s1-01847-twin
sampledesk
ctx_compress-s1-01847
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_sample leaves a trailing dash when the title ends in punctuation. It lives in src/sampledesk/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 sampledesk l...
62,513
3,072
81
136
false
synth:8819768766556240336@1.88563
synth
[{"aliases": ["src/sampledesk/format.py", "format.py"], "kind": "file", "value": "src/sampledesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/sampledesk/tables.py", "tables.py"], "kind": "file", "value": "src/sampledesk/tables.p...
[{"answers": ["src/sampledesk/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-s1-01848
corridorflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 misses keys that differ only in case. It lives in src/corridorflow/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 corridorflow layout. [tool_use kind=sea...
32,941
1,647
57
66
false
synth:8780268817197656623@1
synth
[{"aliases": ["src/corridorflow/store.py", "store.py"], "kind": "file", "value": "src/corridorflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/corridorflow/rules.py", "rules.py"], "kind": "file", "value": "src/corridorflow/rules.py...
[{"answers": ["src/corridorflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01848-twin
corridorflow
ctx_compress-s1-01848
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 misses keys that differ only in case. It lives in src/corridorflow/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 corridorflow layout. [tool_use kind=sea...
17,491
1,000
37
66
false
synth:8780268817197656623@1
synth
[{"aliases": ["src/corridorflow/store.py", "store.py"], "kind": "file", "value": "src/corridorflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/corridorflow/rules.py", "rules.py"], "kind": "file", "value": "src/corridorflow/rules.py...
[{"answers": ["src/corridorflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01849
warehouseworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_lot_page returns one item too few on every page. It lives in src/warehouseworks/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 warehouseworks layout. [tool_...
98,526
3,072
108
125
false
synth:7190735662502054518@1.86563
synth
[{"aliases": ["src/warehouseworks/store.py", "store.py"], "kind": "file", "value": "src/warehouseworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/lot_7.csv", "lot_7.csv"], "kind": "file", "value": "data/lot_7.csv"}, {"aliases": [...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/lot_0.csv", "lot_0.csv"], "kind"...
ctx_compress-s1-01849-twin
warehouseworks
ctx_compress-s1-01849
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_lot_page returns one item too few on every page. It lives in src/warehouseworks/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 warehouseworks layout. [tool_...
68,501
3,072
66
125
false
synth:7190735662502054518@1.86563
synth
[{"aliases": ["src/warehouseworks/store.py", "store.py"], "kind": "file", "value": "src/warehouseworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/lot_7.csv", "lot_7.csv"], "kind": "file", "value": "data/lot_7.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": ["compat.py", "src/warehouseworks/compa...
ctx_compress-s1-01850
sectionkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_revision_safe misses keys that differ only in case. It lives in src/sectionkit/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 sectionkit layout. [tool_use k...
122,704
3,072
121
138
false
synth:6103589857682642851@2.26067
synth
[{"aliases": ["src/sectionkit/report.py", "report.py"], "kind": "file", "value": "src/sectionkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/revision_6.csv", "revision_6.csv"], "kind": "file", "value": "data/revision_6.csv"}, ...
[{"answers": ["src/sectionkit/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01850-twin
sectionkit
ctx_compress-s1-01850
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_revision_safe misses keys that differ only in case. It lives in src/sectionkit/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 sectionkit layout. [tool_use k...
75,212
3,072
81
138
false
synth:6103589857682642851@2.26067
synth
[{"aliases": ["src/sectionkit/report.py", "report.py"], "kind": "file", "value": "src/sectionkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/revision_6.csv", "revision_6.csv"], "kind": "file", "value": "data/revision_6.csv"}, ...
[{"answers": ["src/sectionkit/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01851
calendarledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window lets defaults overwrite explicit overrides. It lives in src/calendarledger/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 calendarledger layout. [to...
36,834
1,841
73
88
false
synth:2791949411321412871@1
synth
[{"aliases": ["src/calendarledger/compat.py", "compat.py"], "kind": "file", "value": "src/calendarledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/calendarledger/audit.py", "audit.py"], "kind": "file", "value": "src/calendarl...
[{"answers": ["src/calendarledger/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...
ctx_compress-s1-01851-twin
calendarledger
ctx_compress-s1-01851
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window lets defaults overwrite explicit overrides. It lives in src/calendarledger/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 calendarledger layout. [to...
22,406
1,120
45
88
false
synth:2791949411321412871@1
synth
[{"aliases": ["src/calendarledger/compat.py", "compat.py"], "kind": "file", "value": "src/calendarledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/calendarledger/audit.py", "audit.py"], "kind": "file", "value": "src/calendarl...
[{"answers": ["src/calendarledger/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...
ctx_compress-s1-01852
revisiondesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_draft raises RevisionBoundError for bytes input in strict mode instead of decoding it. It lives in src/revisiondesk/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 ...
36,608
1,830
70
79
false
synth:2279543610783917410@1
synth
[{"aliases": ["src/revisiondesk/io.py", "io.py"], "kind": "file", "value": "src/revisiondesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/revisiondesk/rules.py", "rules.py"], "kind": "file", "value": "src/revisiondesk/rules.py"}, {"aliases": ["...
[{"answers": ["src/revisiondesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-01852-twin
revisiondesk
ctx_compress-s1-01852
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_draft raises RevisionBoundError for bytes input in strict mode instead of decoding it. It lives in src/revisiondesk/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 ...
18,888
1,000
36
79
false
synth:2279543610783917410@1
synth
[{"aliases": ["src/revisiondesk/io.py", "io.py"], "kind": "file", "value": "src/revisiondesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/revisiondesk/rules.py", "rules.py"], "kind": "file", "value": "src/revisiondesk/rules.py"}, {"aliases": ["...
[{"answers": ["src/revisiondesk/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-s1-01853
creditledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_credit raises ReceiptDriftError for bytes input in strict mode instead of decoding it. It lives in src/creditledger/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping...
39,018
1,950
94
105
false
synth:2434878338658673579@1
synth
[{"aliases": ["src/creditledger/audit.py", "audit.py"], "kind": "file", "value": "src/creditledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/creditledger/compat.py", "compat.py"], "kind": "file", "value": "src/creditledger/compat...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/creditledger/audit.py...
ctx_compress-s1-01853-twin
creditledger
ctx_compress-s1-01853
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_credit raises ReceiptDriftError for bytes input in strict mode instead of decoding it. It lives in src/creditledger/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping...
24,560
1,228
52
105
false
synth:2434878338658673579@1
synth
[{"aliases": ["src/creditledger/audit.py", "audit.py"], "kind": "file", "value": "src/creditledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/creditledger/compat.py", "compat.py"], "kind": "file", "value": "src/creditledger/compat...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/refund_2.csv", "refund_2.csv"], ...
ctx_compress-s1-01854
windowflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_roster produces doubled dashes for consecutive separators. It lives in src/windowflow/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 windowflow layout. [too...
94,099
3,072
106
117
false
synth:1115299031157911132@2.01653
synth
[{"aliases": ["src/windowflow/format.py", "format.py"], "kind": "file", "value": "src/windowflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/slot_8.csv", "slot_8.csv"], "kind": "file", "value": "data/slot_8.csv"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/windowflow/format.py...
ctx_compress-s1-01854-twin
windowflow
ctx_compress-s1-01854
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_roster produces doubled dashes for consecutive separators. It lives in src/windowflow/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 windowflow layout. [too...
53,587
2,679
64
117
false
synth:1115299031157911132@2.01653
synth
[{"aliases": ["src/windowflow/format.py", "format.py"], "kind": "file", "value": "src/windowflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/slot_8.csv", "slot_8.csv"], "kind": "file", "value": "data/slot_8.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": ["report.py", "src/windowflow/report.py...
ctx_compress-s1-01855
indexledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_footnote_page averages one value too many once the window is full. It lives in src/indexledger/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 indexledger la...
101,253
3,072
119
140
false
synth:4371836757817995817@1.87885
synth
[{"aliases": ["src/indexledger/format.py", "format.py"], "kind": "file", "value": "src/indexledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/indexledger/core.py", "core.py"], "kind": "file", "value": "src/indexledger/core.py"...
[{"answers": ["src/indexledger/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-s1-01855-twin
indexledger
ctx_compress-s1-01855
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_footnote_page averages one value too many once the window is full. It lives in src/indexledger/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 indexledger la...
46,020
2,301
63
140
false
synth:4371836757817995817@1.87885
synth
[{"aliases": ["src/indexledger/format.py", "format.py"], "kind": "file", "value": "src/indexledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/indexledger/core.py", "core.py"], "kind": "file", "value": "src/indexledger/core.py"...
[{"answers": ["src/indexledger/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-s1-01856
lotworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_pallet_once leaves a trailing dash when the title ends in punctuation. It lives in src/lotworks/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 lotworks layout. [too...
84,597
3,072
104
129
false
synth:1121414702587977701@2.0263
synth
[{"aliases": ["src/lotworks/cli.py", "cli.py"], "kind": "file", "value": "src/lotworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/lotworks/l...
[{"answers": ["src/lotworks/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-01856-twin
lotworks
ctx_compress-s1-01856
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_pallet_once leaves a trailing dash when the title ends in punctuation. It lives in src/lotworks/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 lotworks layout. [too...
53,563
2,678
76
129
false
synth:1121414702587977701@2.0263
synth
[{"aliases": ["src/lotworks/cli.py", "cli.py"], "kind": "file", "value": "src/lotworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/lotworks/l...
[{"answers": ["src/lotworks/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rece...
ctx_compress-s1-01857
probeflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_trace lets defaults overwrite explicit overrides. It lives in src/probeflow/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 probeflow layout. [tool_use kind=sear...
42,340
2,117
76
97
false
synth:9192634638992326971@1.36965
synth
[{"aliases": ["src/probeflow/rules.py", "rules.py"], "kind": "file", "value": "src/probeflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/probeflow/io.py", "io.py"], "kind": "file", "value": "src/probeflow/io.py"}, {"aliases": ["src...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/probeflow/compat.py"...
ctx_compress-s1-01857-twin
probeflow
ctx_compress-s1-01857
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_trace lets defaults overwrite explicit overrides. It lives in src/probeflow/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 probeflow layout. [tool_use kind=sear...
28,848
1,442
47
97
false
synth:9192634638992326971@1.36965
synth
[{"aliases": ["src/probeflow/rules.py", "rules.py"], "kind": "file", "value": "src/probeflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/probeflow/io.py", "io.py"], "kind": "file", "value": "src/probeflow/io.py"}, {"aliases": ["src...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["io.py", "src/probeflow/io.py"], "kind...
ctx_compress-s1-01858
beaconworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_zone ignores the cap, so late attempts wait for minutes. It lives in src/beaconworks/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 beaconworks layout. [to...
81,746
3,072
107
114
false
synth:5549689221866200731@1.6358
synth
[{"aliases": ["src/beaconworks/limits.py", "limits.py"], "kind": "file", "value": "src/beaconworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/bearing_5.csv", "bearing_5.csv"], "kind": "file", "value": "data/bearing_5.csv"}, {...
[{"answers": ["src/beaconworks/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-s1-01858-twin
beaconworks
ctx_compress-s1-01858
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_zone ignores the cap, so late attempts wait for minutes. It lives in src/beaconworks/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 beaconworks layout. [to...
48,790
2,439
59
114
false
synth:5549689221866200731@1.6358
synth
[{"aliases": ["src/beaconworks/limits.py", "limits.py"], "kind": "file", "value": "src/beaconworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/bearing_5.csv", "bearing_5.csv"], "kind": "file", "value": "data/bearing_5.csv"}, {...
[{"answers": ["src/beaconworks/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 ...