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-s1-00100-twin
payoutkit
ctx_compress-s1-00100
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_ledger leaves a trailing dash when the title ends in punctuation. It lives in src/payoutkit/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 payoutkit layout. [too...
43,349
2,167
51
110
false
synth:4318221935960054779@2.28812
synth
[{"aliases": ["src/payoutkit/io.py", "io.py"], "kind": "file", "value": "src/payoutkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/payout_4.csv", "payout_4.csv"], "kind": "file", "value": "data/payout_4.csv"}, {"aliases": ["src/payoutkit/tabl...
[{"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/payoutkit/compat.py"...
ctx_compress-s1-00101
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] assign_window_strict raises WindowShapeError for bytes input in strict mode instead of decoding it. It lives in src/rosterforge/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 mapp...
50,012
2,500
89
102
false
synth:3268709161119369248@1
synth
[{"aliases": ["src/rosterforge/core.py", "core.py"], "kind": "file", "value": "src/rosterforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/rosterforge/schema.py", "schema.py"], "kind": "file", "value": "src/rosterforge/schema.py"}, {"a...
[{"answers": ["src/rosterforge/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-s1-00101-twin
rosterforge
ctx_compress-s1-00101
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window_strict raises WindowShapeError for bytes input in strict mode instead of decoding it. It lives in src/rosterforge/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 mapp...
21,844
1,092
51
102
false
synth:3268709161119369248@1
synth
[{"aliases": ["src/rosterforge/core.py", "core.py"], "kind": "file", "value": "src/rosterforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/rosterforge/schema.py", "schema.py"], "kind": "file", "value": "src/rosterforge/schema.py"}, {"a...
[{"answers": ["src/rosterforge/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-s1-00102
invoiceflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_receipt_window raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/invoiceflow/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapp...
61,588
3,072
102
123
false
synth:1501961340814128757@1.30638
synth
[{"aliases": ["src/invoiceflow/limits.py", "limits.py"], "kind": "file", "value": "src/invoiceflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/payout_4.csv", "payout_4.csv"], "kind": "file", "value": "data/payout_4.csv"}, {"al...
[{"answers": ["src/invoiceflow/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-s1-00102-twin
invoiceflow
ctx_compress-s1-00102
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_receipt_window raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/invoiceflow/limits.py; `python -m pytest tests/test_limits.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapp...
40,331
2,016
56
123
false
synth:1501961340814128757@1.30638
synth
[{"aliases": ["src/invoiceflow/limits.py", "limits.py"], "kind": "file", "value": "src/invoiceflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/payout_4.csv", "payout_4.csv"], "kind": "file", "value": "data/payout_4.csv"}, {"al...
[{"answers": ["src/invoiceflow/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-s1-00103
noncedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_safe raises ValueError when a value itself contains an equals sign. It lives in src/noncedesk/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 noncedesk layou...
26,980
1,349
52
61
false
synth:2058632875073706706@1
synth
[{"aliases": ["src/noncedesk/io.py", "io.py"], "kind": "file", "value": "src/noncedesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/noncedesk/tables.py", "tables.py"], "kind": "file", "value": "src/noncedesk/tables.py"}, {"aliases": ["src/nonce...
[{"answers": ["src/noncedesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-00103-twin
noncedesk
ctx_compress-s1-00103
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_safe raises ValueError when a value itself contains an equals sign. It lives in src/noncedesk/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 noncedesk layou...
11,687
1,000
28
61
false
synth:2058632875073706706@1
synth
[{"aliases": ["src/noncedesk/io.py", "io.py"], "kind": "file", "value": "src/noncedesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/noncedesk/tables.py", "tables.py"], "kind": "file", "value": "src/noncedesk/tables.py"}, {"aliases": ["src/nonce...
[{"answers": ["src/noncedesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-00104
balanceworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_balance raises KeyError for a missing key instead of returning the default. It lives in src/balanceworks/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 b...
85,352
3,072
82
101
false
synth:4443907268705857431@1.5728
synth
[{"aliases": ["src/balanceworks/report.py", "report.py"], "kind": "file", "value": "src/balanceworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/balanceworks/io.py", "io.py"], "kind": "file", "value": "src/balanceworks/io.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["schema.py", "src/balanceworks/schema....
ctx_compress-s1-00104-twin
balanceworks
ctx_compress-s1-00104
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_balance raises KeyError for a missing key instead of returning the default. It lives in src/balanceworks/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 b...
54,798
2,739
52
101
false
synth:4443907268705857431@1.5728
synth
[{"aliases": ["src/balanceworks/report.py", "report.py"], "kind": "file", "value": "src/balanceworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/balanceworks/io.py", "io.py"], "kind": "file", "value": "src/balanceworks/io.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/balanceworks/audit.py...
ctx_compress-s1-00105
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_probe_batch averages one value too many once the window is full. 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...
58,976
2,948
108
127
false
synth:5817324158631161952@1.12458
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/counter_3.csv", "counter_3.csv"], "kind": "file", "value": "data/counter_3.csv"}, {"a...
[{"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-s1-00105-twin
counterkit
ctx_compress-s1-00105
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe_batch averages one value too many once the window is full. 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...
29,302
1,465
62
127
false
synth:5817324158631161952@1.12458
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/counter_3.csv", "counter_3.csv"], "kind": "file", "value": "data/counter_3.csv"}, {"a...
[{"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-s1-00106
snippetflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_lexeme produces doubled dashes for consecutive separators. It lives in src/snippetflow/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 snippetflow layout. [tool_us...
23,369
1,168
46
49
false
synth:6827057239831857433@1
synth
[{"aliases": ["src/snippetflow/io.py", "io.py"], "kind": "file", "value": "src/snippetflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/snippetflow/audit.py", "audit.py"], "kind": "file", "value": "src/snippetflow/audit.py"}, {"aliases": ["docs...
[{"answers": ["tests/test_io.py", "test_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-s1-00106-twin
snippetflow
ctx_compress-s1-00106
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_lexeme produces doubled dashes for consecutive separators. It lives in src/snippetflow/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 snippetflow layout. [tool_us...
10,448
1,000
27
49
false
synth:6827057239831857433@1
synth
[{"aliases": ["src/snippetflow/io.py", "io.py"], "kind": "file", "value": "src/snippetflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/snippetflow/audit.py", "audit.py"], "kind": "file", "value": "src/snippetflow/audit.py"}, {"aliases": ["docs...
[{"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/snippetflow/tables.py", "tables.p...
ctx_compress-s1-00107
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 returns one item too few on every page. It lives in src/indexledger/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 indexledger layout. [tool_use ki...
94,598
3,072
99
116
false
synth:3362833085079783801@1.7202
synth
[{"aliases": ["src/indexledger/limits.py", "limits.py"], "kind": "file", "value": "src/indexledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/indexledger/schema.py", "schema.py"], "kind": "file", "value": "src/indexledger/sche...
[{"answers": ["src/indexledger/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-s1-00107-twin
indexledger
ctx_compress-s1-00107
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 returns one item too few on every page. It lives in src/indexledger/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 indexledger layout. [tool_use ki...
45,040
2,252
57
116
false
synth:3362833085079783801@1.7202
synth
[{"aliases": ["src/indexledger/limits.py", "limits.py"], "kind": "file", "value": "src/indexledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/indexledger/schema.py", "schema.py"], "kind": "file", "value": "src/indexledger/sche...
[{"answers": ["src/indexledger/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-s1-00108
probekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_sample_window raises SeriesShapeError for bytes input in strict mode instead of decoding it. It lives in src/probekit/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping...
31,949
1,597
66
83
false
synth:8186624598313155087@1
synth
[{"aliases": ["src/probekit/store.py", "store.py"], "kind": "file", "value": "src/probekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/probekit/format.py", "format.py"], "kind": "file", "value": "src/probekit/format.py"}, {"aliases...
[{"answers": ["src/probekit/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-00108-twin
probekit
ctx_compress-s1-00108
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_sample_window raises SeriesShapeError for bytes input in strict mode instead of decoding it. It lives in src/probekit/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping...
19,983
1,000
48
83
false
synth:8186624598313155087@1
synth
[{"aliases": ["src/probekit/store.py", "store.py"], "kind": "file", "value": "src/probekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/probekit/format.py", "format.py"], "kind": "file", "value": "src/probekit/format.py"}, {"aliases...
[{"answers": ["src/probekit/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-s1-00109
tracedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_series raises CounterShapeError for bytes input in strict mode instead of decoding it. It lives in src/tracedesk/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 ...
35,947
1,797
55
68
false
synth:4082516654131657678@1
synth
[{"aliases": ["src/tracedesk/rules.py", "rules.py"], "kind": "file", "value": "src/tracedesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/tracedesk/store.py", "store.py"], "kind": "file", "value": "src/tracedesk/store.py"}, {"aliase...
[{"answers": ["src/tracedesk/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00109-twin
tracedesk
ctx_compress-s1-00109
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_series raises CounterShapeError for bytes input in strict mode instead of decoding it. It lives in src/tracedesk/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 ...
22,659
1,132
39
68
false
synth:4082516654131657678@1
synth
[{"aliases": ["src/tracedesk/rules.py", "rules.py"], "kind": "file", "value": "src/tracedesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/tracedesk/store.py", "store.py"], "kind": "file", "value": "src/tracedesk/store.py"}, {"aliase...
[{"answers": ["src/tracedesk/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00110
receiptkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_invoice skips the first page; page 1 returns what should be page 2. It lives in src/receiptkit/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 receiptkit ...
99,006
3,072
107
128
false
synth:6075052942675787872@1.80553
synth
[{"aliases": ["src/receiptkit/limits.py", "limits.py"], "kind": "file", "value": "src/receiptkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/receiptkit/io.py", "io.py"], "kind": "file", "value": "src/receiptkit/io.py"}, {"alias...
[{"answers": ["src/receiptkit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00110-twin
receiptkit
ctx_compress-s1-00110
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_invoice skips the first page; page 1 returns what should be page 2. It lives in src/receiptkit/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 receiptkit ...
59,262
2,963
73
128
false
synth:6075052942675787872@1.80553
synth
[{"aliases": ["src/receiptkit/limits.py", "limits.py"], "kind": "file", "value": "src/receiptkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/receiptkit/io.py", "io.py"], "kind": "file", "value": "src/receiptkit/io.py"}, {"alias...
[{"answers": ["src/receiptkit/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00111
ledgerforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_ledger raises ValueError when a value itself contains an equals sign. It lives in src/ledgerforge/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 ledgerforge lay...
43,653
2,182
88
107
false
synth:3903043534669071451@1.14017
synth
[{"aliases": ["src/ledgerforge/cli.py", "cli.py"], "kind": "file", "value": "src/ledgerforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/ledgerforge/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerforge/limits.py"}, {"aliases...
[{"answers": ["src/ledgerforge/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-s1-00111-twin
ledgerforge
ctx_compress-s1-00111
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_ledger raises ValueError when a value itself contains an equals sign. It lives in src/ledgerforge/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 ledgerforge lay...
28,125
1,406
49
107
false
synth:3903043534669071451@1.14017
synth
[{"aliases": ["src/ledgerforge/cli.py", "cli.py"], "kind": "file", "value": "src/ledgerforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/ledgerforge/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerforge/limits.py"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/ledgerforge/limits.p...
ctx_compress-s1-00112
bookingkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/bookingkit/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 bookingkit...
24,326
1,216
61
68
false
synth:526495427283742379@1
synth
[{"aliases": ["src/bookingkit/format.py", "format.py"], "kind": "file", "value": "src/bookingkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/shift_2.csv", "shift_2.csv"], "kind": "file", "value": "data/shift_2.csv"}, {"aliases...
[{"answers": ["src/bookingkit/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-s1-00112-twin
bookingkit
ctx_compress-s1-00112
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/bookingkit/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 bookingkit...
13,864
1,000
33
68
false
synth:526495427283742379@1
synth
[{"aliases": ["src/bookingkit/format.py", "format.py"], "kind": "file", "value": "src/bookingkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/shift_2.csv", "shift_2.csv"], "kind": "file", "value": "data/shift_2.csv"}, {"aliases...
[{"answers": ["src/bookingkit/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-s1-00113
tileledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_tile ignores the cap, so late attempts wait for minutes. It lives in src/tileledger/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 tileledger layout. [tool_u...
97,692
3,072
123
144
false
synth:845852078153404940@2.16255
synth
[{"aliases": ["src/tileledger/report.py", "report.py"], "kind": "file", "value": "src/tileledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/beacon_12.csv", "beacon_12.csv"], "kind": "file", "value": "data/beacon_12.csv"}, {"a...
[{"answers": ["src/tileledger/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-s1-00113-twin
tileledger
ctx_compress-s1-00113
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_tile ignores the cap, so late attempts wait for minutes. It lives in src/tileledger/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 tileledger layout. [tool_u...
45,665
2,283
67
144
false
synth:845852078153404940@2.16255
synth
[{"aliases": ["src/tileledger/report.py", "report.py"], "kind": "file", "value": "src/tileledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/beacon_12.csv", "beacon_12.csv"], "kind": "file", "value": "data/beacon_12.csv"}, {"a...
[{"answers": ["src/tileledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-00114
quotakit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_calendar misses keys that differ only in case. It lives in src/quotakit/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 quotakit layout. [tool_use kind=searc...
55,906
2,795
94
111
false
synth:2806356018281022358@1.266
synth
[{"aliases": ["src/quotakit/compat.py", "compat.py"], "kind": "file", "value": "src/quotakit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/quotakit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00114-twin
quotakit
ctx_compress-s1-00114
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_calendar misses keys that differ only in case. It lives in src/quotakit/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 quotakit layout. [tool_use kind=searc...
33,842
1,692
62
111
false
synth:2806356018281022358@1.266
synth
[{"aliases": ["src/quotakit/compat.py", "compat.py"], "kind": "file", "value": "src/quotakit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/quotakit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00115
slugdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_slug lets defaults overwrite explicit overrides. It lives in src/slugdesk/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slugdesk layout. [tool_use kind=search na...
58,073
2,903
66
73
false
synth:6913754167816954146@1
synth
[{"aliases": ["src/slugdesk/core.py", "core.py"], "kind": "file", "value": "src/slugdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/slugdesk/limits.py", "limits.py"], "kind": "file", "value": "src/slugdesk/limits.py"}, {"aliases": ["s...
[{"answers": ["src/slugdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most re...
ctx_compress-s1-00115-twin
slugdesk
ctx_compress-s1-00115
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_slug lets defaults overwrite explicit overrides. It lives in src/slugdesk/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the slugdesk layout. [tool_use kind=search na...
25,392
1,269
35
73
false
synth:6913754167816954146@1
synth
[{"aliases": ["src/slugdesk/core.py", "core.py"], "kind": "file", "value": "src/slugdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/slugdesk/limits.py", "limits.py"], "kind": "file", "value": "src/slugdesk/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": ["report.py", "src/slugdesk/report.py"]...
ctx_compress-s1-00116
calendarkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 rejects a range where lo equals hi, which is a valid single point range. It lives in src/calendarkit/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 calen...
93,124
3,072
101
110
false
synth:9018895373721257212@1.89268
synth
[{"aliases": ["src/calendarkit/audit.py", "audit.py"], "kind": "file", "value": "src/calendarkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/calendarkit/rules.py", "rules.py"], "kind": "file", "value": "src/calendarkit/rules.py"}, ...
[{"answers": ["src/calendarkit/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00116-twin
calendarkit
ctx_compress-s1-00116
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 rejects a range where lo equals hi, which is a valid single point range. It lives in src/calendarkit/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 calen...
44,698
2,234
53
110
false
synth:9018895373721257212@1.89268
synth
[{"aliases": ["src/calendarkit/audit.py", "audit.py"], "kind": "file", "value": "src/calendarkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/calendarkit/rules.py", "rules.py"], "kind": "file", "value": "src/calendarkit/rules.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": ["data/window_1.csv", "window_1.csv"], ...
ctx_compress-s1-00117
bucketkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_trace_window raises ValueError when a value itself contains an equals sign. It lives in src/bucketkit/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 bucketki...
92,968
3,072
132
143
false
synth:592414786709477661@2.08745
synth
[{"aliases": ["src/bucketkit/report.py", "report.py"], "kind": "file", "value": "src/bucketkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/counter_2.csv", "counter_2.csv"], "kind": "file", "value": "data/counter_2.csv"}, {"ali...
[{"answers": ["src/bucketkit/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00117-twin
bucketkit
ctx_compress-s1-00117
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_trace_window raises ValueError when a value itself contains an equals sign. It lives in src/bucketkit/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 bucketki...
47,957
2,397
77
143
false
synth:592414786709477661@2.08745
synth
[{"aliases": ["src/bucketkit/report.py", "report.py"], "kind": "file", "value": "src/bucketkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/counter_2.csv", "counter_2.csv"], "kind": "file", "value": "data/counter_2.csv"}, {"ali...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/span_3.csv", "span_3.csv"], "kin...
ctx_compress-s1-00118
skuflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_sku raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/skuflow/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 skuflow layo...
22,514
1,125
33
38
false
synth:4698197655861604126@1
synth
[{"aliases": ["src/skuflow/store.py", "store.py"], "kind": "file", "value": "src/skuflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/skuflow/io.py", "io.py"], "kind": "file", "value": "src/skuflow/io.py"}, {"aliases": ["data/lot_2....
[{"answers": ["src/skuflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-00118-twin
skuflow
ctx_compress-s1-00118
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_sku raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/skuflow/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 skuflow layo...
18,543
1,000
21
38
false
synth:4698197655861604126@1
synth
[{"aliases": ["src/skuflow/store.py", "store.py"], "kind": "file", "value": "src/skuflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/skuflow/io.py", "io.py"], "kind": "file", "value": "src/skuflow/io.py"}, {"aliases": ["data/lot_2....
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/skuflow/format.py"],...
ctx_compress-s1-00119
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] transfer_reorder skips the first page; page 1 returns what should be page 2. It lives in src/stockforge/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 stockforge l...
50,335
2,516
74
93
false
synth:8055274734161840506@1.4462
synth
[{"aliases": ["src/stockforge/compat.py", "compat.py"], "kind": "file", "value": "src/stockforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/stockforge/report.py", "report.py"], "kind": "file", "value": "src/stockforge/report.p...
[{"answers": ["src/stockforge/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00119-twin
stockforge
ctx_compress-s1-00119
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 skips the first page; page 1 returns what should be page 2. It lives in src/stockforge/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 stockforge l...
33,977
1,698
47
93
false
synth:8055274734161840506@1.4462
synth
[{"aliases": ["src/stockforge/compat.py", "compat.py"], "kind": "file", "value": "src/stockforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/stockforge/report.py", "report.py"], "kind": "file", "value": "src/stockforge/report.p...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-00120
chunkforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_snippet_once misses keys that differ only in case. It lives in src/chunkforge/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 chunkforge layout. [tool_use ki...
50,255
2,512
92
101
false
synth:7857669948834518893@1
synth
[{"aliases": ["src/chunkforge/store.py", "store.py"], "kind": "file", "value": "src/chunkforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/chunkforge/core.py", "core.py"], "kind": "file", "value": "src/chunkforge/core.py"}, {"alias...
[{"answers": ["src/chunkforge/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-s1-00120-twin
chunkforge
ctx_compress-s1-00120
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_snippet_once misses keys that differ only in case. It lives in src/chunkforge/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 chunkforge layout. [tool_use ki...
28,511
1,425
52
101
false
synth:7857669948834518893@1
synth
[{"aliases": ["src/chunkforge/store.py", "store.py"], "kind": "file", "value": "src/chunkforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/chunkforge/core.py", "core.py"], "kind": "file", "value": "src/chunkforge/core.py"}, {"alias...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/chunkforge/compat.py...
ctx_compress-s1-00121
lexemeforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_glyph_all raises ValueError when a value itself contains an equals sign. It lives in src/lexemeforge/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 lexemefo...
40,552
2,027
59
74
false
synth:2913327653119521735@1
synth
[{"aliases": ["src/lexemeforge/store.py", "store.py"], "kind": "file", "value": "src/lexemeforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/lexemeforge/core.py", "core.py"], "kind": "file", "value": "src/lexemeforge/core.py"}, {"a...
[{"answers": ["src/lexemeforge/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-s1-00121-twin
lexemeforge
ctx_compress-s1-00121
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_glyph_all raises ValueError when a value itself contains an equals sign. It lives in src/lexemeforge/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 lexemefo...
22,314
1,115
37
74
false
synth:2913327653119521735@1
synth
[{"aliases": ["src/lexemeforge/store.py", "store.py"], "kind": "file", "value": "src/lexemeforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/lexemeforge/core.py", "core.py"], "kind": "file", "value": "src/lexemeforge/core.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/lexemeforge/store.py", "store.py"...
ctx_compress-s1-00122
windowforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster double counts carried amounts when the period closes on a weekend. It lives in src/windowforge/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 window...
39,017
1,950
71
88
false
synth:8924226882202128167@1
synth
[{"aliases": ["src/windowforge/format.py", "format.py"], "kind": "file", "value": "src/windowforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/roster_1.csv", "roster_1.csv"], "kind": "file", "value": "data/roster_1.csv"}, {"al...
[{"answers": ["src/windowforge/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-s1-00122-twin
windowforge
ctx_compress-s1-00122
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster double counts carried amounts when the period closes on a weekend. It lives in src/windowforge/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 window...
23,507
1,175
45
88
false
synth:8924226882202128167@1
synth
[{"aliases": ["src/windowforge/format.py", "format.py"], "kind": "file", "value": "src/windowforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/roster_1.csv", "roster_1.csv"], "kind": "file", "value": "data/roster_1.csv"}, {"al...
[{"answers": ["src/windowforge/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-s1-00123
waypointkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/waypointkit/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 waypointkit...
102,555
3,072
125
136
false
synth:6502004244896902823@2.06553
synth
[{"aliases": ["src/waypointkit/audit.py", "audit.py"], "kind": "file", "value": "src/waypointkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/waypointkit/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00123-twin
waypointkit
ctx_compress-s1-00123
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/waypointkit/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 waypointkit...
56,905
2,845
67
136
false
synth:6502004244896902823@2.06553
synth
[{"aliases": ["src/waypointkit/audit.py", "audit.py"], "kind": "file", "value": "src/waypointkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases"...
[{"answers": ["src/waypointkit/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00124
rosterkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_roster_once raises WindowBoundError for bytes input in strict mode instead of decoding it. It lives in src/rosterkit/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapp...
81,796
3,072
106
121
false
synth:2522670436518056220@1.82868
synth
[{"aliases": ["src/rosterkit/format.py", "format.py"], "kind": "file", "value": "src/rosterkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/rosterkit/report.py", "report.py"], "kind": "file", "value": "src/rosterkit/report.py"},...
[{"answers": ["src/rosterkit/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 mo...
ctx_compress-s1-00124-twin
rosterkit
ctx_compress-s1-00124
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_roster_once raises WindowBoundError for bytes input in strict mode instead of decoding it. It lives in src/rosterkit/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapp...
47,055
2,352
68
121
false
synth:2522670436518056220@1.82868
synth
[{"aliases": ["src/rosterkit/format.py", "format.py"], "kind": "file", "value": "src/rosterkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/rosterkit/report.py", "report.py"], "kind": "file", "value": "src/rosterkit/report.py"},...
[{"answers": ["src/rosterkit/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00125
slotflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_slot misses keys that differ only in case. It lives in src/slotflow/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 slotflow layout. [tool_use kind=search name=B...
22,039
1,101
52
63
false
synth:4965054223854472673@1
synth
[{"aliases": ["src/slotflow/core.py", "core.py"], "kind": "file", "value": "src/slotflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/quota_1.csv", "quota_1.csv"], "kind": "file", "value": "data/quota_1.csv"}, {"aliases": ["src/slotflo...
[{"answers": ["src/slotflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00125-twin
slotflow
ctx_compress-s1-00125
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_slot misses keys that differ only in case. It lives in src/slotflow/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 slotflow layout. [tool_use kind=search name=B...
15,563
1,000
34
63
false
synth:4965054223854472673@1
synth
[{"aliases": ["src/slotflow/core.py", "core.py"], "kind": "file", "value": "src/slotflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/quota_1.csv", "quota_1.csv"], "kind": "file", "value": "data/quota_1.csv"}, {"aliases": ["src/slotflo...
[{"answers": ["src/slotflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00126
grantforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_grant_batch raises KeyError for a missing key instead of returning the default. It lives in src/grantforge/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 grantforge...
75,997
3,072
99
124
false
synth:1359939758459098799@1.66315
synth
[{"aliases": ["src/grantforge/io.py", "io.py"], "kind": "file", "value": "src/grantforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/grantforge/core.py", "core.py"], "kind": "file", "value": "src/grantforge/core.py"}, {"aliases": ["src/grantfo...
[{"answers": ["src/grantforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-00126-twin
grantforge
ctx_compress-s1-00126
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_grant_batch raises KeyError for a missing key instead of returning the default. It lives in src/grantforge/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 grantforge...
46,559
2,327
69
124
false
synth:1359939758459098799@1.66315
synth
[{"aliases": ["src/grantforge/io.py", "io.py"], "kind": "file", "value": "src/grantforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/grantforge/core.py", "core.py"], "kind": "file", "value": "src/grantforge/core.py"}, {"aliases": ["src/grantfo...
[{"answers": ["src/grantforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-00127
tileledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_tile_window rejects a range where lo equals hi, which is a valid single point range. It lives in src/tileledger/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 ...
48,790
2,439
90
101
false
synth:1693147251000937658@1.37017
synth
[{"aliases": ["src/tileledger/format.py", "format.py"], "kind": "file", "value": "src/tileledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/tileledger/rules.py", "rules.py"], "kind": "file", "value": "src/tileledger/rules.py"}...
[{"answers": ["src/tileledger/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-s1-00127-twin
tileledger
ctx_compress-s1-00127
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_tile_window rejects a range where lo equals hi, which is a valid single point range. It lives in src/tileledger/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 ...
31,744
1,587
60
101
false
synth:1693147251000937658@1.37017
synth
[{"aliases": ["src/tileledger/format.py", "format.py"], "kind": "file", "value": "src/tileledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/tileledger/rules.py", "rules.py"], "kind": "file", "value": "src/tileledger/rules.py"}...
[{"answers": ["src/tileledger/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-s1-00128
counterforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterforge/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...
91,054
3,072
126
135
false
synth:5106006269660034505@1.67708
synth
[{"aliases": ["src/counterforge/report.py", "report.py"], "kind": "file", "value": "src/counterforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/counterforge/schema.py", "schema.py"], "kind": "file", "value": "src/counterforge/...
[{"answers": ["src/counterforge/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-s1-00128-twin
counterforge
ctx_compress-s1-00128
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterforge/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...
53,194
2,659
78
135
false
synth:5106006269660034505@1.67708
synth
[{"aliases": ["src/counterforge/report.py", "report.py"], "kind": "file", "value": "src/counterforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/counterforge/schema.py", "schema.py"], "kind": "file", "value": "src/counterforge/...
[{"answers": ["src/counterforge/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-s1-00129
traceflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/traceflow/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 ...
21,488
1,074
46
51
false
synth:478325241481794244@1
synth
[{"aliases": ["src/traceflow/store.py", "store.py"], "kind": "file", "value": "src/traceflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/traceflow/audit.py", "audit.py"], "kind": "file", "value": "src/traceflow/audit.py"}, {"aliase...
[{"answers": ["src/traceflow/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-s1-00129-twin
traceflow
ctx_compress-s1-00129
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/traceflow/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 ...
7,872
1,000
26
51
false
synth:478325241481794244@1
synth
[{"aliases": ["src/traceflow/store.py", "store.py"], "kind": "file", "value": "src/traceflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/traceflow/audit.py", "audit.py"], "kind": "file", "value": "src/traceflow/audit.py"}, {"aliase...
[{"answers": ["src/traceflow/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-s1-00130
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 double counts carried amounts when the period closes on a weekend. It lives in src/revisiondesk/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 revision...
66,383
3,072
106
131
false
synth:8325304708000801914@1.81678
synth
[{"aliases": ["src/revisiondesk/limits.py", "limits.py"], "kind": "file", "value": "src/revisiondesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/revisiondesk/report.py", "report.py"], "kind": "file", "value": "src/revisiondesk/...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-00130-twin
revisiondesk
ctx_compress-s1-00130
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 double counts carried amounts when the period closes on a weekend. It lives in src/revisiondesk/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 revision...
39,233
1,961
59
131
false
synth:8325304708000801914@1.81678
synth
[{"aliases": ["src/revisiondesk/limits.py", "limits.py"], "kind": "file", "value": "src/revisiondesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/revisiondesk/report.py", "report.py"], "kind": "file", "value": "src/revisiondesk/...
[{"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/revisiondesk/audit.py...
ctx_compress-s1-00131
snippetforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cue_strict leaves a trailing dash when the title ends in punctuation. It lives in src/snippetforge/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 snippetfor...
97,256
3,072
109
136
false
synth:1951593016378407790@2.20052
synth
[{"aliases": ["src/snippetforge/compat.py", "compat.py"], "kind": "file", "value": "src/snippetforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/snippetforge/io.py", "io.py"], "kind": "file", "value": "src/snippetforge/io.py"},...
[{"answers": ["src/snippetforge/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-s1-00131-twin
snippetforge
ctx_compress-s1-00131
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cue_strict leaves a trailing dash when the title ends in punctuation. It lives in src/snippetforge/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 snippetfor...
56,696
2,834
67
136
false
synth:1951593016378407790@2.20052
synth
[{"aliases": ["src/snippetforge/compat.py", "compat.py"], "kind": "file", "value": "src/snippetforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/snippetforge/io.py", "io.py"], "kind": "file", "value": "src/snippetforge/io.py"},...
[{"answers": ["src/snippetforge/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-00132
lexemeflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_lexeme lets defaults overwrite explicit overrides. It lives in src/lexemeflow/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 lexemeflow layout. [tool_use kind=searc...
45,882
2,294
85
102
false
synth:1680643663951267835@1.02188
synth
[{"aliases": ["src/lexemeflow/cli.py", "cli.py"], "kind": "file", "value": "src/lexemeflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/lexemeflow/store.py", "store.py"], "kind": "file", "value": "src/lexemeflow/store.py"}, {"aliases": ["do...
[{"answers": ["src/lexemeflow/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 re...
ctx_compress-s1-00132-twin
lexemeflow
ctx_compress-s1-00132
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_lexeme lets defaults overwrite explicit overrides. It lives in src/lexemeflow/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 lexemeflow layout. [tool_use kind=searc...
25,616
1,280
51
102
false
synth:1680643663951267835@1.02188
synth
[{"aliases": ["src/lexemeflow/cli.py", "cli.py"], "kind": "file", "value": "src/lexemeflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/lexemeflow/store.py", "store.py"], "kind": "file", "value": "src/lexemeflow/store.py"}, {"aliases": ["do...
[{"answers": ["src/lexemeflow/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 re...
ctx_compress-s1-00133
beaconflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon double counts carried amounts when the period closes on a weekend. It lives in src/beaconflow/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 beaconflo...
54,149
2,707
85
100
false
synth:5198561270805928626@1.06218
synth
[{"aliases": ["src/beaconflow/store.py", "store.py"], "kind": "file", "value": "src/beaconflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/beaconflow/tables.py", "tables.py"], "kind": "file", "value": "src/beaconflow/tables.py"}, {...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_io.py", "tests/test_io.py"], "ki...
ctx_compress-s1-00133-twin
beaconflow
ctx_compress-s1-00133
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon double counts carried amounts when the period closes on a weekend. It lives in src/beaconflow/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 beaconflo...
42,724
2,136
53
100
false
synth:5198561270805928626@1.06218
synth
[{"aliases": ["src/beaconflow/store.py", "store.py"], "kind": "file", "value": "src/beaconflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/beaconflow/tables.py", "tables.py"], "kind": "file", "value": "src/beaconflow/tables.py"}, {...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/beaconflow/limits.py...
ctx_compress-s1-00134
revisionworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_margin_batch misses keys that differ only in case. It lives in src/revisionworks/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 revisionworks layout. [tool_...
42,335
2,116
90
101
false
synth:2548676225978334073@1.06303
synth
[{"aliases": ["src/revisionworks/limits.py", "limits.py"], "kind": "file", "value": "src/revisionworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/revisionworks/schema.py", "schema.py"], "kind": "file", "value": "src/revisionwo...
[{"answers": ["src/revisionworks/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent'...
ctx_compress-s1-00134-twin
revisionworks
ctx_compress-s1-00134
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_margin_batch misses keys that differ only in case. It lives in src/revisionworks/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 revisionworks layout. [tool_...
26,372
1,318
56
101
false
synth:2548676225978334073@1.06303
synth
[{"aliases": ["src/revisionworks/limits.py", "limits.py"], "kind": "file", "value": "src/revisionworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/revisionworks/schema.py", "schema.py"], "kind": "file", "value": "src/revisionwo...
[{"answers": ["src/revisionworks/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'...
ctx_compress-s1-00135
legdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_zone raises KeyError for a missing key instead of returning the default. It lives in src/legdesk/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 legdesk layout. [to...
44,851
2,242
75
84
false
synth:2565058381929920735@1
synth
[{"aliases": ["src/legdesk/io.py", "io.py"], "kind": "file", "value": "src/legdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/bearing_4.csv", "bearing_4.csv"], "kind": "file", "value": "data/bearing_4.csv"}, {"aliases": ["docs/changelog.md",...
[{"answers": ["src/legdesk/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-s1-00135-twin
legdesk
ctx_compress-s1-00135
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_zone raises KeyError for a missing key instead of returning the default. It lives in src/legdesk/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 legdesk layout. [to...
23,066
1,153
43
84
false
synth:2565058381929920735@1
synth
[{"aliases": ["src/legdesk/io.py", "io.py"], "kind": "file", "value": "src/legdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/bearing_4.csv", "bearing_4.csv"], "kind": "file", "value": "data/bearing_4.csv"}, {"aliases": ["docs/changelog.md",...
[{"answers": ["src/legdesk/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-s1-00136
bucketworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_series double counts carried amounts when the period closes on a weekend. It lives in src/bucketworks/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 bucketworks l...
52,499
2,624
99
112
false
synth:4384552796463489942@1.25805
synth
[{"aliases": ["src/bucketworks/core.py", "core.py"], "kind": "file", "value": "src/bucketworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/bucketworks/compat.py", "compat.py"], "kind": "file", "value": "src/bucketworks/compat.py"}, {"a...
[{"answers": ["src/bucketworks/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-s1-00136-twin
bucketworks
ctx_compress-s1-00136
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_series double counts carried amounts when the period closes on a weekend. It lives in src/bucketworks/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 bucketworks l...
30,757
1,537
61
112
false
synth:4384552796463489942@1.25805
synth
[{"aliases": ["src/bucketworks/core.py", "core.py"], "kind": "file", "value": "src/bucketworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/bucketworks/compat.py", "compat.py"], "kind": "file", "value": "src/bucketworks/compat.py"}, {"a...
[{"answers": ["src/bucketworks/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-s1-00137
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_balance keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/creditledger/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 creditle...
105,352
3,072
113
130
false
synth:2207407638032535482@2.08133
synth
[{"aliases": ["src/creditledger/format.py", "format.py"], "kind": "file", "value": "src/creditledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/creditledger/limits.py", "limits.py"], "kind": "file", "value": "src/creditledger/...
[{"answers": ["src/creditledger/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-s1-00137-twin
creditledger
ctx_compress-s1-00137
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_balance keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/creditledger/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 creditle...
56,319
2,815
61
130
false
synth:2207407638032535482@2.08133
synth
[{"aliases": ["src/creditledger/format.py", "format.py"], "kind": "file", "value": "src/creditledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/creditledger/limits.py", "limits.py"], "kind": "file", "value": "src/creditledger/...
[{"answers": ["src/creditledger/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-s1-00138
policyforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_principal keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/policyforge/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 policyforge lay...
29,925
1,496
74
89
false
synth:4305813583187532465@1
synth
[{"aliases": ["src/policyforge/io.py", "io.py"], "kind": "file", "value": "src/policyforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/policyforge/rules.py", "rules.py"], "kind": "file", "value": "src/policyforge/rules.py"}, {"aliases": ["src/...
[{"answers": ["src/policyforge/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-s1-00138-twin
policyforge
ctx_compress-s1-00138
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_principal keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/policyforge/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 policyforge lay...
15,810
1,000
44
89
false
synth:4305813583187532465@1
synth
[{"aliases": ["src/policyforge/io.py", "io.py"], "kind": "file", "value": "src/policyforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/policyforge/rules.py", "rules.py"], "kind": "file", "value": "src/policyforge/rules.py"}, {"aliases": ["src/...
[{"answers": ["src/policyforge/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-s1-00139
bookingkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_roster_safe drops override values of 0, False and empty string. It lives in src/bookingkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bookingkit layout. ...
107,511
3,072
132
145
false
synth:6830301773366635688@1.89165
synth
[{"aliases": ["src/bookingkit/rules.py", "rules.py"], "kind": "file", "value": "src/bookingkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/roster_2.csv", "roster_2.csv"], "kind": "file", "value": "data/roster_2.csv"}, {"aliases": ...
[{"answers": ["src/bookingkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00139-twin
bookingkit
ctx_compress-s1-00139
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_roster_safe drops override values of 0, False and empty string. It lives in src/bookingkit/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bookingkit layout. ...
60,999
3,049
80
145
false
synth:6830301773366635688@1.89165
synth
[{"aliases": ["src/bookingkit/rules.py", "rules.py"], "kind": "file", "value": "src/bookingkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/roster_2.csv", "roster_2.csv"], "kind": "file", "value": "data/roster_2.csv"}, {"aliases": ...
[{"answers": ["src/bookingkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00140
sampleworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_trace grows linearly instead of doubling. It lives in src/sampleworks/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 sampleworks layout. [tool_use kind=se...
81,309
3,072
104
127
false
synth:8901665534945662527@1.55783
synth
[{"aliases": ["src/sampleworks/store.py", "store.py"], "kind": "file", "value": "src/sampleworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sample_1.csv", "sample_1.csv"], "kind": "file", "value": "data/sample_1.csv"}, {"aliases"...
[{"answers": ["src/sampleworks/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-s1-00140-twin
sampleworks
ctx_compress-s1-00140
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_trace grows linearly instead of doubling. It lives in src/sampleworks/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 sampleworks layout. [tool_use kind=se...
60,275
3,013
76
127
false
synth:8901665534945662527@1.55783
synth
[{"aliases": ["src/sampleworks/store.py", "store.py"], "kind": "file", "value": "src/sampleworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sample_1.csv", "sample_1.csv"], "kind": "file", "value": "data/sample_1.csv"}, {"aliases"...
[{"answers": ["src/sampleworks/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-s1-00141
grantworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_token_all returns one item too few on every page. It lives in src/grantworks/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 grantworks layout. [tool_use kind=se...
30,599
1,529
56
61
false
synth:986124904102617586@1
synth
[{"aliases": ["src/grantworks/core.py", "core.py"], "kind": "file", "value": "src/grantworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/...
[{"answers": ["src/grantworks/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-s1-00141-twin
grantworks
ctx_compress-s1-00141
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_token_all returns one item too few on every page. It lives in src/grantworks/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 grantworks layout. [tool_use kind=se...
15,003
1,000
28
61
false
synth:986124904102617586@1
synth
[{"aliases": ["src/grantworks/core.py", "core.py"], "kind": "file", "value": "src/grantworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/...
[{"answers": ["src/grantworks/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-s1-00142
tracekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_gauge raises TraceStateError for bytes input in strict mode instead of decoding it. It lives in src/tracekit/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 tr...
82,733
3,072
113
132
false
synth:9055873932989249386@1.53005
synth
[{"aliases": ["src/tracekit/compat.py", "compat.py"], "kind": "file", "value": "src/tracekit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/tracekit/report.py", "report.py"], "kind": "file", "value": "src/tracekit/report.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/series_5.csv", "series_5.csv"], ...
ctx_compress-s1-00142-twin
tracekit
ctx_compress-s1-00142
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_gauge raises TraceStateError for bytes input in strict mode instead of decoding it. It lives in src/tracekit/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 tr...
47,325
2,366
61
132
false
synth:9055873932989249386@1.53005
synth
[{"aliases": ["src/tracekit/compat.py", "compat.py"], "kind": "file", "value": "src/tracekit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/tracekit/report.py", "report.py"], "kind": "file", "value": "src/tracekit/report.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/tracekit/format.py"]...
ctx_compress-s1-00143
quotadesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_window averages one value too many once the window is full. It lives in src/quotadesk/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 quotadesk layout. [tool_use...
65,538
3,072
102
115
false
synth:8249980245815072890@1.44785
synth
[{"aliases": ["src/quotadesk/core.py", "core.py"], "kind": "file", "value": "src/quotadesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/quotadesk/store.py", "store.py"], "kind": "file", "value": "src/quotadesk/store.py"}, {"aliases": ["...
[{"answers": ["src/quotadesk/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-s1-00143-twin
quotadesk
ctx_compress-s1-00143
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_window averages one value too many once the window is full. It lives in src/quotadesk/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 quotadesk layout. [tool_use...
34,906
1,745
56
115
false
synth:8249980245815072890@1.44785
synth
[{"aliases": ["src/quotadesk/core.py", "core.py"], "kind": "file", "value": "src/quotadesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/quotadesk/store.py", "store.py"], "kind": "file", "value": "src/quotadesk/store.py"}, {"aliases": ["...
[{"answers": ["src/quotadesk/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-s1-00144
payoutledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_receipt_batch ignores the cap, so late attempts wait for minutes. It lives in src/payoutledger/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 payoutledger...
91,095
3,072
102
113
false
synth:7619886202062186856@1.7165
synth
[{"aliases": ["src/payoutledger/report.py", "report.py"], "kind": "file", "value": "src/payoutledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/ledger_8.csv", "ledger_8.csv"], "kind": "file", "value": "data/ledger_8.csv"}, {"...
[{"answers": ["src/payoutledger/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-s1-00144-twin
payoutledger
ctx_compress-s1-00144
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_receipt_batch ignores the cap, so late attempts wait for minutes. It lives in src/payoutledger/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 payoutledger...
43,286
2,164
56
113
false
synth:7619886202062186856@1.7165
synth
[{"aliases": ["src/payoutledger/report.py", "report.py"], "kind": "file", "value": "src/payoutledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/ledger_8.csv", "ledger_8.csv"], "kind": "file", "value": "data/ledger_8.csv"}, {"...
[{"answers": ["src/payoutledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-00145
grantledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_token_strict grows linearly instead of doubling. It lives in src/grantledger/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 grantledger layout. [tool_use ki...
91,243
3,072
120
147
false
synth:7188294249433775461@1.91762
synth
[{"aliases": ["src/grantledger/report.py", "report.py"], "kind": "file", "value": "src/grantledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/policy_3.csv", "policy_3.csv"], "kind": "file", "value": "data/policy_3.csv"}, {"al...
[{"answers": ["src/grantledger/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-s1-00145-twin
grantledger
ctx_compress-s1-00145
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_token_strict grows linearly instead of doubling. It lives in src/grantledger/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 grantledger layout. [tool_use ki...
52,926
2,646
78
147
false
synth:7188294249433775461@1.91762
synth
[{"aliases": ["src/grantledger/report.py", "report.py"], "kind": "file", "value": "src/grantledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/policy_3.csv", "policy_3.csv"], "kind": "file", "value": "data/policy_3.csv"}, {"al...
[{"answers": ["src/grantledger/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-s1-00146
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] wrap_slug raises KeyError for a missing key instead of returning the default. It lives in src/cueworks/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 cueworks layout...
103,189
3,072
100
125
false
synth:6408644518746000151@2.13238
synth
[{"aliases": ["src/cueworks/rules.py", "rules.py"], "kind": "file", "value": "src/cueworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/chunk_10.csv", "chunk_10.csv"], "kind": "file", "value": "data/chunk_10.csv"}, {"aliases": ["da...
[{"answers": ["src/cueworks/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-00146-twin
cueworks
ctx_compress-s1-00146
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_slug raises KeyError for a missing key instead of returning the default. It lives in src/cueworks/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 cueworks layout...
54,792
2,739
70
125
false
synth:6408644518746000151@2.13238
synth
[{"aliases": ["src/cueworks/rules.py", "rules.py"], "kind": "file", "value": "src/cueworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/chunk_10.csv", "chunk_10.csv"], "kind": "file", "value": "data/chunk_10.csv"}, {"aliases": ["da...
[{"answers": ["src/cueworks/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-00147
gaugeledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bucket rejects a range where lo equals hi, which is a valid single point range. It lives in src/gaugeledger/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 ga...
87,934
3,072
115
144
false
synth:2706281960009324340@1.93138
synth
[{"aliases": ["src/gaugeledger/compat.py", "compat.py"], "kind": "file", "value": "src/gaugeledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/sample_9.csv", "sample_9.csv"], "kind": "file", "value": "data/sample_9.csv"}, {"al...
[{"answers": ["src/gaugeledger/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-s1-00147-twin
gaugeledger
ctx_compress-s1-00147
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bucket rejects a range where lo equals hi, which is a valid single point range. It lives in src/gaugeledger/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 ga...
65,237
3,072
83
144
false
synth:2706281960009324340@1.93138
synth
[{"aliases": ["src/gaugeledger/compat.py", "compat.py"], "kind": "file", "value": "src/gaugeledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/sample_9.csv", "sample_9.csv"], "kind": "file", "value": "data/sample_9.csv"}, {"al...
[{"answers": ["src/gaugeledger/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-s1-00148
slotdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster leaves a trailing dash when the title ends in punctuation. It lives in src/slotdesk/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 slotdesk layout. [t...
74,131
3,072
108
117
false
synth:8096848927533166110@1.28577
synth
[{"aliases": ["src/slotdesk/rules.py", "rules.py"], "kind": "file", "value": "src/slotdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slotdesk/tables.py", "tables.py"], "kind": "file", "value": "src/slotdesk/tables.py"}, {"aliases...
[{"answers": ["src/slotdesk/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-00148-twin
slotdesk
ctx_compress-s1-00148
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster leaves a trailing dash when the title ends in punctuation. It lives in src/slotdesk/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 slotdesk layout. [t...
37,937
1,896
60
117
false
synth:8096848927533166110@1.28577
synth
[{"aliases": ["src/slotdesk/rules.py", "rules.py"], "kind": "file", "value": "src/slotdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slotdesk/tables.py", "tables.py"], "kind": "file", "value": "src/slotdesk/tables.py"}, {"aliases...
[{"answers": ["src/slotdesk/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00149
skukit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reserve_bin drops override values of 0, False and empty string. It lives in src/skukit/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 skukit layout. [tool_use kind=s...
87,597
3,072
108
115
false
synth:6202819864371779614@1.51743
synth
[{"aliases": ["src/skukit/audit.py", "audit.py"], "kind": "file", "value": "src/skukit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/sku_1.csv", "sku_1.csv"], "kind": "file", "value": "data/sku_1.csv"}, {"aliases": ["docs/changelog....
[{"answers": ["src/skukit/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 re...
ctx_compress-s1-00149-twin
skukit
ctx_compress-s1-00149
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reserve_bin drops override values of 0, False and empty string. It lives in src/skukit/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 skukit layout. [tool_use kind=s...
40,819
2,040
54
115
false
synth:6202819864371779614@1.51743
synth
[{"aliases": ["src/skukit/audit.py", "audit.py"], "kind": "file", "value": "src/skukit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/sku_1.csv", "sku_1.csv"], "kind": "file", "value": "data/sku_1.csv"}, {"aliases": ["docs/changelog....
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_core.py", "tests/test_core.py"],...
ctx_compress-s1-00150
marginledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_heading_strict drops override values of 0, False and empty string. It lives in src/marginledger/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 marginledger layout...
53,865
2,693
68
81
false
synth:8042419497335135324@1
synth
[{"aliases": ["src/marginledger/core.py", "core.py"], "kind": "file", "value": "src/marginledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/section_5.csv", "section_5.csv"], "kind": "file", "value": "data/section_5.csv"}, {"aliases":...
[{"answers": ["src/marginledger/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 mos...