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-s1-01156
corridorledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_fix misses keys that differ only in case. It lives in src/corridorledger/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 corridorledger layout. [tool_use kind=sear...
100,904
3,072
119
134
false
synth:6045520494533991254@1.9215
synth
[{"aliases": ["src/corridorledger/io.py", "io.py"], "kind": "file", "value": "src/corridorledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/corridorledger/cli.py", "cli.py"], "kind": "file", "value": "src/corridorledger/cli.py"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/waypoint_13.csv", "waypoint_13.c...
ctx_compress-s1-01156-twin
corridorledger
ctx_compress-s1-01156
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_fix misses keys that differ only in case. It lives in src/corridorledger/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 corridorledger layout. [tool_use kind=sear...
58,485
2,924
67
134
false
synth:6045520494533991254@1.9215
synth
[{"aliases": ["src/corridorledger/io.py", "io.py"], "kind": "file", "value": "src/corridorledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/corridorledger/cli.py", "cli.py"], "kind": "file", "value": "src/corridorledger/cli.py"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/fix_1.csv", "fix_1.csv"], "kind"...
ctx_compress-s1-01157
nonceledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_grant averages one value too many once the window is full. It lives in src/nonceledger/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 nonceledger layout. [t...
67,427
3,072
90
115
false
synth:3374487255553821602@2.30095
synth
[{"aliases": ["src/nonceledger/compat.py", "compat.py"], "kind": "file", "value": "src/nonceledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/session_3.csv", "session_3.csv"], "kind": "file", "value": "data/session_3.csv"}, {...
[{"answers": ["src/nonceledger/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-s1-01157-twin
nonceledger
ctx_compress-s1-01157
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 averages one value too many once the window is full. It lives in src/nonceledger/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 nonceledger layout. [t...
47,946
2,397
60
115
false
synth:3374487255553821602@2.30095
synth
[{"aliases": ["src/nonceledger/compat.py", "compat.py"], "kind": "file", "value": "src/nonceledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/session_3.csv", "session_3.csv"], "kind": "file", "value": "data/session_3.csv"}, {...
[{"answers": ["src/nonceledger/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-s1-01158
sampledesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_trace_batch drops override values of 0, False and empty string. It lives in src/sampledesk/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 sampledesk layout. [to...
69,024
3,072
90
111
false
synth:793175783794675592@2.0322
synth
[{"aliases": ["src/sampledesk/store.py", "store.py"], "kind": "file", "value": "src/sampledesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/bucket_2.csv", "bucket_2.csv"], "kind": "file", "value": "data/bucket_2.csv"}, {"aliases": ...
[{"answers": ["src/sampledesk/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-s1-01158-twin
sampledesk
ctx_compress-s1-01158
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_trace_batch drops override values of 0, False and empty string. It lives in src/sampledesk/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 sampledesk layout. [to...
40,241
2,012
53
111
false
synth:793175783794675592@2.0322
synth
[{"aliases": ["src/sampledesk/store.py", "store.py"], "kind": "file", "value": "src/sampledesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/bucket_2.csv", "bucket_2.csv"], "kind": "file", "value": "data/bucket_2.csv"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_limits.py", "tests/test_limits.p...
ctx_compress-s1-01159
bearingflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_leg leaves a trailing dash when the title ends in punctuation. It lives in src/bearingflow/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 bearingflow layout....
44,417
2,220
80
101
false
synth:176466156962977671@1
synth
[{"aliases": ["src/bearingflow/rules.py", "rules.py"], "kind": "file", "value": "src/bearingflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/bearingflow/tables.py", "tables.py"], "kind": "file", "value": "src/bearingflow/tables.py"...
[{"answers": ["src/bearingflow/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01159-twin
bearingflow
ctx_compress-s1-01159
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_leg leaves a trailing dash when the title ends in punctuation. It lives in src/bearingflow/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 bearingflow layout....
32,516
1,625
60
101
false
synth:176466156962977671@1
synth
[{"aliases": ["src/bearingflow/rules.py", "rules.py"], "kind": "file", "value": "src/bearingflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/bearingflow/tables.py", "tables.py"], "kind": "file", "value": "src/bearingflow/tables.py"...
[{"answers": ["src/bearingflow/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01160
windowledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_window_page double counts carried amounts when the period closes on a weekend. It lives in src/windowledger/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 wi...
43,980
2,199
90
97
false
synth:8480173535815623778@1
synth
[{"aliases": ["src/windowledger/audit.py", "audit.py"], "kind": "file", "value": "src/windowledger/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"}, {"aliase...
[{"answers": ["src/windowledger/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 m...
ctx_compress-s1-01160-twin
windowledger
ctx_compress-s1-01160
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_window_page double counts carried amounts when the period closes on a weekend. It lives in src/windowledger/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 wi...
25,462
1,273
56
97
false
synth:8480173535815623778@1
synth
[{"aliases": ["src/windowledger/audit.py", "audit.py"], "kind": "file", "value": "src/windowledger/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"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_compat.py", "tests/test_compat.p...
ctx_compress-s1-01161
shipmentflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_stock raises KeyError for a missing key instead of returning the default. It lives in src/shipmentflow/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 shipmentflo...
39,122
1,956
92
103
false
synth:6196764705943742041@1
synth
[{"aliases": ["src/shipmentflow/audit.py", "audit.py"], "kind": "file", "value": "src/shipmentflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/shipmentflow/limits.py", "limits.py"], "kind": "file", "value": "src/shipmentflow/limits...
[{"answers": ["src/shipmentflow/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 m...
ctx_compress-s1-01161-twin
shipmentflow
ctx_compress-s1-01161
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_stock raises KeyError for a missing key instead of returning the default. It lives in src/shipmentflow/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 shipmentflo...
24,785
1,239
56
103
false
synth:6196764705943742041@1
synth
[{"aliases": ["src/shipmentflow/audit.py", "audit.py"], "kind": "file", "value": "src/shipmentflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/shipmentflow/limits.py", "limits.py"], "kind": "file", "value": "src/shipmentflow/limits...
[{"answers": ["src/shipmentflow/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 m...
ctx_compress-s1-01162
palletflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_reorder lets defaults overwrite explicit overrides. It lives in src/palletflow/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 palletflow layout. [tool_use ki...
61,183
3,059
90
109
false
synth:6531141021948643597@1.73632
synth
[{"aliases": ["src/palletflow/compat.py", "compat.py"], "kind": "file", "value": "src/palletflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/stock_0.csv", "stock_0.csv"], "kind": "file", "value": "data/stock_0.csv"}, {"aliases...
[{"answers": ["src/palletflow/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-s1-01162-twin
palletflow
ctx_compress-s1-01162
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_reorder lets defaults overwrite explicit overrides. It lives in src/palletflow/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 palletflow layout. [tool_use ki...
40,236
2,011
60
109
false
synth:6531141021948643597@1.73632
synth
[{"aliases": ["src/palletflow/compat.py", "compat.py"], "kind": "file", "value": "src/palletflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/stock_0.csv", "stock_0.csv"], "kind": "file", "value": "data/stock_0.csv"}, {"aliases...
[{"answers": ["src/palletflow/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 m...
ctx_compress-s1-01163
waypointledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 returns one item too few on every page. It lives in src/waypointledger/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 waypointledger layout. [tool_use k...
38,248
1,912
78
87
false
synth:8473840539503215385@1
synth
[{"aliases": ["src/waypointledger/store.py", "store.py"], "kind": "file", "value": "src/waypointledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/waypointledger/format.py", "format.py"], "kind": "file", "value": "src/waypointledge...
[{"answers": ["src/waypointledger/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-01163-twin
waypointledger
ctx_compress-s1-01163
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 returns one item too few on every page. It lives in src/waypointledger/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 waypointledger layout. [tool_use k...
24,293
1,214
48
87
false
synth:8473840539503215385@1
synth
[{"aliases": ["src/waypointledger/store.py", "store.py"], "kind": "file", "value": "src/waypointledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/waypointledger/format.py", "format.py"], "kind": "file", "value": "src/waypointledge...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["core.py", "src/waypointledger/core.py...
ctx_compress-s1-01164
seriesledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_trace produces doubled dashes for consecutive separators. It lives in src/seriesledger/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 seriesledger layout. [too...
67,680
3,072
94
105
false
synth:4905926631731135387@1.138
synth
[{"aliases": ["src/seriesledger/limits.py", "limits.py"], "kind": "file", "value": "src/seriesledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/seriesledger/io.py", "io.py"], "kind": "file", "value": "src/seriesledger/io.py"},...
[{"answers": ["src/seriesledger/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-s1-01164-twin
seriesledger
ctx_compress-s1-01164
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_trace produces doubled dashes for consecutive separators. It lives in src/seriesledger/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 seriesledger layout. [too...
30,193
1,509
49
105
false
synth:4905926631731135387@1.138
synth
[{"aliases": ["src/seriesledger/limits.py", "limits.py"], "kind": "file", "value": "src/seriesledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/seriesledger/io.py", "io.py"], "kind": "file", "value": "src/seriesledger/io.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/seriesledger/limits....
ctx_compress-s1-01165
fixflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_zone_once returns the upper bound for every value instead of clamping into the range. It lives in src/fixflow/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 fixflow ...
57,216
2,860
94
105
false
synth:6915343420198651980@1.29455
synth
[{"aliases": ["src/fixflow/io.py", "io.py"], "kind": "file", "value": "src/fixflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/fix_0.csv", "fix_0.csv"], "kind": "file", "value": "data/fix_0.csv"}, {"aliases": ["docs/changelog.md", "changelog....
[{"answers": ["src/fixflow/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-s1-01165-twin
fixflow
ctx_compress-s1-01165
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_zone_once returns the upper bound for every value instead of clamping into the range. It lives in src/fixflow/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 fixflow ...
32,173
1,608
58
105
false
synth:6915343420198651980@1.29455
synth
[{"aliases": ["src/fixflow/io.py", "io.py"], "kind": "file", "value": "src/fixflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/fix_0.csv", "fix_0.csv"], "kind": "file", "value": "data/fix_0.csv"}, {"aliases": ["docs/changelog.md", "changelog....
[{"answers": ["src/fixflow/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recent ...
ctx_compress-s1-01166
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] anchor_revision produces doubled dashes for consecutive separators. It lives in src/revisiondesk/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 revisiondesk layout...
63,250
3,072
106
115
false
synth:4598796658077417855@1.45423
synth
[{"aliases": ["src/revisiondesk/format.py", "format.py"], "kind": "file", "value": "src/revisiondesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/draft_2.csv", "draft_2.csv"], "kind": "file", "value": "data/draft_2.csv"}, {"ali...
[{"answers": ["src/revisiondesk/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-s1-01166-twin
revisiondesk
ctx_compress-s1-01166
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_revision produces doubled dashes for consecutive separators. It lives in src/revisiondesk/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 revisiondesk layout...
41,870
2,093
66
115
false
synth:4598796658077417855@1.45423
synth
[{"aliases": ["src/revisiondesk/format.py", "format.py"], "kind": "file", "value": "src/revisiondesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/draft_2.csv", "draft_2.csv"], "kind": "file", "value": "data/draft_2.csv"}, {"ali...
[{"answers": ["src/revisiondesk/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-s1-01167
fixforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bound_beacon_once raises KeyError for a missing key instead of returning the default. It lives in src/fixforge/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 fixforge la...
86,457
3,072
105
128
false
synth:665476339085299242@2.2331
synth
[{"aliases": ["src/fixforge/cli.py", "cli.py"], "kind": "file", "value": "src/fixforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/waypoint_9.csv", "waypoint_9.csv"], "kind": "file", "value": "data/waypoint_9.csv"}, {"aliases": ["src/fixf...
[{"answers": ["src/fixforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-01167-twin
fixforge
ctx_compress-s1-01167
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon_once raises KeyError for a missing key instead of returning the default. It lives in src/fixforge/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 fixforge la...
57,853
2,892
69
128
false
synth:665476339085299242@2.2331
synth
[{"aliases": ["src/fixforge/cli.py", "cli.py"], "kind": "file", "value": "src/fixforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/waypoint_9.csv", "waypoint_9.csv"], "kind": "file", "value": "data/waypoint_9.csv"}, {"aliases": ["src/fixf...
[{"answers": ["src/fixforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rece...
ctx_compress-s1-01168
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] rate_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterforge/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 cou...
26,386
1,319
54
67
false
synth:7788747067547444040@1
synth
[{"aliases": ["src/counterforge/store.py", "store.py"], "kind": "file", "value": "src/counterforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/span_0.csv", "span_0.csv"], "kind": "file", "value": "data/span_0.csv"}, {"aliases": ["...
[{"answers": ["src/counterforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01168-twin
counterforge
ctx_compress-s1-01168
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterforge/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 cou...
13,964
1,000
36
67
false
synth:7788747067547444040@1
synth
[{"aliases": ["src/counterforge/store.py", "store.py"], "kind": "file", "value": "src/counterforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/span_0.csv", "span_0.csv"], "kind": "file", "value": "data/span_0.csv"}, {"aliases": ["...
[{"answers": ["src/counterforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01169
nonceworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch raises NonceBoundError for bytes input in strict mode instead of decoding it. It lives in src/nonceworks/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 ma...
31,930
1,596
71
84
false
synth:3428168016766114173@1
synth
[{"aliases": ["src/nonceworks/format.py", "format.py"], "kind": "file", "value": "src/nonceworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/nonceworks/report.py", "report.py"], "kind": "file", "value": "src/nonceworks/report.p...
[{"answers": ["src/nonceworks/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-s1-01169-twin
nonceworks
ctx_compress-s1-01169
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch raises NonceBoundError for bytes input in strict mode instead of decoding it. It lives in src/nonceworks/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 ma...
20,343
1,017
45
84
false
synth:3428168016766114173@1
synth
[{"aliases": ["src/nonceworks/format.py", "format.py"], "kind": "file", "value": "src/nonceworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/nonceworks/report.py", "report.py"], "kind": "file", "value": "src/nonceworks/report.p...
[{"answers": ["src/nonceworks/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-s1-01170
receiptforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_tariff produces doubled dashes for consecutive separators. It lives in src/receiptforge/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 receiptforge layout. [tool_use...
41,019
2,050
71
86
false
synth:6165457939766128466@1
synth
[{"aliases": ["src/receiptforge/cli.py", "cli.py"], "kind": "file", "value": "src/receiptforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/ledger_0.csv", "ledger_0.csv"], "kind": "file", "value": "data/ledger_0.csv"}, {"aliases": ["data/b...
[{"answers": ["src/receiptforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-01170-twin
receiptforge
ctx_compress-s1-01170
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_tariff produces doubled dashes for consecutive separators. It lives in src/receiptforge/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 receiptforge layout. [tool_use...
28,536
1,426
49
86
false
synth:6165457939766128466@1
synth
[{"aliases": ["src/receiptforge/cli.py", "cli.py"], "kind": "file", "value": "src/receiptforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/ledger_0.csv", "ledger_0.csv"], "kind": "file", "value": "data/ledger_0.csv"}, {"aliases": ["data/b...
[{"answers": ["src/receiptforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-01171
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] cluster_tile rejects a range where lo equals hi, which is a valid single point range. It lives in src/tileledger/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 tileledge...
58,161
2,908
105
112
false
synth:1367229308144847399@1.10252
synth
[{"aliases": ["src/tileledger/cli.py", "cli.py"], "kind": "file", "value": "src/tileledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/tile_4.csv", "tile_4.csv"], "kind": "file", "value": "data/tile_4.csv"}, {"aliases": ["data/tile_0.csv"...
[{"answers": ["src/tileledger/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-s1-01171-twin
tileledger
ctx_compress-s1-01171
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_tile rejects a range where lo equals hi, which is a valid single point range. It lives in src/tileledger/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 tileledge...
31,539
1,576
59
112
false
synth:1367229308144847399@1.10252
synth
[{"aliases": ["src/tileledger/cli.py", "cli.py"], "kind": "file", "value": "src/tileledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/tile_4.csv", "tile_4.csv"], "kind": "file", "value": "data/tile_4.csv"}, {"aliases": ["data/tile_0.csv"...
[{"answers": ["src/tileledger/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-s1-01172
headingforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_revision_once double counts carried amounts when the period closes on a weekend. It lives in src/headingforge/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 h...
96,364
3,072
101
126
false
synth:6433243924407711358@1.96582
synth
[{"aliases": ["src/headingforge/rules.py", "rules.py"], "kind": "file", "value": "src/headingforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/revision_11.csv", "revision_11.csv"], "kind": "file", "value": "data/revision_11.csv"},...
[{"answers": ["src/headingforge/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 m...
ctx_compress-s1-01172-twin
headingforge
ctx_compress-s1-01172
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_revision_once double counts carried amounts when the period closes on a weekend. It lives in src/headingforge/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 h...
68,527
3,072
71
126
false
synth:6433243924407711358@1.96582
synth
[{"aliases": ["src/headingforge/rules.py", "rules.py"], "kind": "file", "value": "src/headingforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/revision_11.csv", "revision_11.csv"], "kind": "file", "value": "data/revision_11.csv"},...
[{"answers": ["src/headingforge/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 m...
ctx_compress-s1-01173
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] pick_reorder raises PalletBoundError for bytes input in strict mode instead of decoding it. It lives in src/skuflow/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 ...
58,146
2,907
60
71
false
synth:2399814531848704937@1
synth
[{"aliases": ["src/skuflow/compat.py", "compat.py"], "kind": "file", "value": "src/skuflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/reorder_5.csv", "reorder_5.csv"], "kind": "file", "value": "data/reorder_5.csv"}, {"aliases...
[{"answers": ["src/skuflow/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01173-twin
skuflow
ctx_compress-s1-01173
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_reorder raises PalletBoundError for bytes input in strict mode instead of decoding it. It lives in src/skuflow/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 ...
26,648
1,332
35
71
false
synth:2399814531848704937@1
synth
[{"aliases": ["src/skuflow/compat.py", "compat.py"], "kind": "file", "value": "src/skuflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/reorder_5.csv", "reorder_5.csv"], "kind": "file", "value": "data/reorder_5.csv"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/skuflow/compat.py"],...
ctx_compress-s1-01174
calendarflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_slot produces doubled dashes for consecutive separators. It lives in src/calendarflow/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 calendarflow layout. [...
66,122
3,072
96
117
false
synth:1475816442533188426@1.70705
synth
[{"aliases": ["src/calendarflow/compat.py", "compat.py"], "kind": "file", "value": "src/calendarflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/window_5.csv", "window_5.csv"], "kind": "file", "value": "data/window_5.csv"}, {"...
[{"answers": ["src/calendarflow/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-s1-01174-twin
calendarflow
ctx_compress-s1-01174
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_slot produces doubled dashes for consecutive separators. It lives in src/calendarflow/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 calendarflow layout. [...
37,088
1,854
58
117
false
synth:1475816442533188426@1.70705
synth
[{"aliases": ["src/calendarflow/compat.py", "compat.py"], "kind": "file", "value": "src/calendarflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/window_5.csv", "window_5.csv"], "kind": "file", "value": "data/window_5.csv"}, {"...
[{"answers": ["src/calendarflow/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-s1-01175
sectionflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_section_page returns the upper bound for every value instead of clamping into the range. It lives in src/sectionflow/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...
40,194
2,009
70
81
false
synth:1171566528066015286@1
synth
[{"aliases": ["src/sectionflow/format.py", "format.py"], "kind": "file", "value": "src/sectionflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/sectionflow/report.py", "report.py"], "kind": "file", "value": "src/sectionflow/repo...
[{"answers": ["src/sectionflow/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-s1-01175-twin
sectionflow
ctx_compress-s1-01175
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_section_page returns the upper bound for every value instead of clamping into the range. It lives in src/sectionflow/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...
17,688
1,000
38
81
false
synth:1171566528066015286@1
synth
[{"aliases": ["src/sectionflow/format.py", "format.py"], "kind": "file", "value": "src/sectionflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/sectionflow/report.py", "report.py"], "kind": "file", "value": "src/sectionflow/repo...
[{"answers": ["src/sectionflow/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-s1-01176
snippetdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] normalise_glyph_batch raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/snippetdesk/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 ...
100,435
3,072
122
143
false
synth:970021087299201844@2.36097
synth
[{"aliases": ["src/snippetdesk/audit.py", "audit.py"], "kind": "file", "value": "src/snippetdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/chunk_1.csv", "chunk_1.csv"], "kind": "file", "value": "data/chunk_1.csv"}, {"aliases": [...
[{"answers": ["src/snippetdesk/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-s1-01176-twin
snippetdesk
ctx_compress-s1-01176
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] normalise_glyph_batch raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/snippetdesk/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 ...
72,617
3,072
84
143
false
synth:970021087299201844@2.36097
synth
[{"aliases": ["src/snippetdesk/audit.py", "audit.py"], "kind": "file", "value": "src/snippetdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/chunk_1.csv", "chunk_1.csv"], "kind": "file", "value": "data/chunk_1.csv"}, {"aliases": [...
[{"answers": ["src/snippetdesk/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-s1-01177
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] truncate_caption double counts carried amounts when the period closes on a weekend. It lives in src/cueworks/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 cueworks ...
71,715
3,072
100
113
false
synth:7920113150741292670@2.36397
synth
[{"aliases": ["src/cueworks/store.py", "store.py"], "kind": "file", "value": "src/cueworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/cueworks/cli.py", "cli.py"], "kind": "file", "value": "src/cueworks/cli.py"}, {"aliases": ["data...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/glyph_2.csv", "glyph_2.csv"], "k...
ctx_compress-s1-01177-twin
cueworks
ctx_compress-s1-01177
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_caption double counts carried amounts when the period closes on a weekend. It lives in src/cueworks/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 cueworks ...
38,240
1,912
55
113
false
synth:7920113150741292670@2.36397
synth
[{"aliases": ["src/cueworks/store.py", "store.py"], "kind": "file", "value": "src/cueworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/cueworks/cli.py", "cli.py"], "kind": "file", "value": "src/cueworks/cli.py"}, {"aliases": ["data...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["cli.py", "src/cueworks/cli.py"], "kin...
ctx_compress-s1-01178
slugworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_snippet_window double counts carried amounts when the period closes on a weekend. It lives in src/slugworks/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 slu...
35,349
1,767
71
80
false
synth:3363517245482993500@1
synth
[{"aliases": ["src/slugworks/compat.py", "compat.py"], "kind": "file", "value": "src/slugworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/cue_1.csv", "cue_1.csv"], "kind": "file", "value": "data/cue_1.csv"}, {"aliases": ["src...
[{"answers": ["src/slugworks/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01178-twin
slugworks
ctx_compress-s1-01178
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_snippet_window double counts carried amounts when the period closes on a weekend. It lives in src/slugworks/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 slu...
15,518
1,000
43
80
false
synth:3363517245482993500@1
synth
[{"aliases": ["src/slugworks/compat.py", "compat.py"], "kind": "file", "value": "src/slugworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/cue_1.csv", "cue_1.csv"], "kind": "file", "value": "data/cue_1.csv"}, {"aliases": ["src...
[{"answers": ["src/slugworks/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01179
payoutdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_refund grows linearly instead of doubling. It lives in src/payoutdesk/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 payoutdesk layout. [tool_use kind=search name=...
55,596
2,779
94
101
false
synth:3707398473572972158@1.49263
synth
[{"aliases": ["src/payoutdesk/core.py", "core.py"], "kind": "file", "value": "src/payoutdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/payoutdesk/schema.py", "schema.py"], "kind": "file", "value": "src/payoutdesk/schema.py"}, {"alias...
[{"answers": ["src/payoutdesk/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-s1-01179-twin
payoutdesk
ctx_compress-s1-01179
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_refund grows linearly instead of doubling. It lives in src/payoutdesk/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 payoutdesk layout. [tool_use kind=search name=...
32,571
1,628
47
101
false
synth:3707398473572972158@1.49263
synth
[{"aliases": ["src/payoutdesk/core.py", "core.py"], "kind": "file", "value": "src/payoutdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/payoutdesk/schema.py", "schema.py"], "kind": "file", "value": "src/payoutdesk/schema.py"}, {"alias...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/invoice_0.csv", "invoice_0.csv"]...
ctx_compress-s1-01180
shipmentkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] restock_shipment returns the upper bound for every value instead of clamping into the range. It lives in src/shipmentkit/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...
21,711
1,085
43
52
false
synth:5798838782149120383@1
synth
[{"aliases": ["src/shipmentkit/compat.py", "compat.py"], "kind": "file", "value": "src/shipmentkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/shipmentkit/report.py", "report.py"], "kind": "file", "value": "src/shipmentkit/repo...
[{"answers": ["src/shipmentkit/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-s1-01180-twin
shipmentkit
ctx_compress-s1-01180
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] restock_shipment returns the upper bound for every value instead of clamping into the range. It lives in src/shipmentkit/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...
9,740
1,000
25
52
false
synth:5798838782149120383@1
synth
[{"aliases": ["src/shipmentkit/compat.py", "compat.py"], "kind": "file", "value": "src/shipmentkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/shipmentkit/report.py", "report.py"], "kind": "file", "value": "src/shipmentkit/repo...
[{"answers": ["src/shipmentkit/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-s1-01181
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] snap_zone raises LegDriftError for bytes input in strict mode instead of decoding it. It lives in src/tileledger/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 til...
25,178
1,258
52
57
false
synth:8473241239816300414@1
synth
[{"aliases": ["src/tileledger/limits.py", "limits.py"], "kind": "file", "value": "src/tileledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/tileledger/report.py", "report.py"], "kind": "file", "value": "src/tileledger/report.p...
[{"answers": ["src/tileledger/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01181-twin
tileledger
ctx_compress-s1-01181
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] snap_zone raises LegDriftError for bytes input in strict mode instead of decoding it. It lives in src/tileledger/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 til...
15,499
1,000
30
57
false
synth:8473241239816300414@1
synth
[{"aliases": ["src/tileledger/limits.py", "limits.py"], "kind": "file", "value": "src/tileledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/tileledger/report.py", "report.py"], "kind": "file", "value": "src/tileledger/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": ["io.py", "src/tileledger/io.py"], "kin...
ctx_compress-s1-01182
snippetledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_slug produces doubled dashes for consecutive separators. It lives in src/snippetledger/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 snippetledger layout. [t...
37,258
1,862
66
73
false
synth:6983368770850269044@1
synth
[{"aliases": ["src/snippetledger/core.py", "core.py"], "kind": "file", "value": "src/snippetledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/snippetledger/io.py", "io.py"], "kind": "file", "value": "src/snippetledger/io.py"}, {"alias...
[{"answers": ["src/snippetledger/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 mo...
ctx_compress-s1-01182-twin
snippetledger
ctx_compress-s1-01182
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_slug produces doubled dashes for consecutive separators. It lives in src/snippetledger/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 snippetledger layout. [t...
20,635
1,031
36
73
false
synth:6983368770850269044@1
synth
[{"aliases": ["src/snippetledger/core.py", "core.py"], "kind": "file", "value": "src/snippetledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/snippetledger/io.py", "io.py"], "kind": "file", "value": "src/snippetledger/io.py"}, {"alias...
[{"answers": ["src/snippetledger/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 mo...
ctx_compress-s1-01183
clausedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_draft_once raises HeadingRangeError for bytes input in strict mode instead of decoding it. It lives in src/clausedesk/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 mappi...
97,731
3,072
111
126
false
synth:6631114825078226373@2.02893
synth
[{"aliases": ["src/clausedesk/rules.py", "rules.py"], "kind": "file", "value": "src/clausedesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/clausedesk/tables.py", "tables.py"], "kind": "file", "value": "src/clausedesk/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": ["compat.py", "src/clausedesk/compat.py...
ctx_compress-s1-01183-twin
clausedesk
ctx_compress-s1-01183
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_draft_once raises HeadingRangeError for bytes input in strict mode instead of decoding it. It lives in src/clausedesk/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 mappi...
64,326
3,072
71
126
false
synth:6631114825078226373@2.02893
synth
[{"aliases": ["src/clausedesk/rules.py", "rules.py"], "kind": "file", "value": "src/clausedesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/clausedesk/tables.py", "tables.py"], "kind": "file", "value": "src/clausedesk/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": ["cli.py", "src/clausedesk/cli.py"], "k...
ctx_compress-s1-01184
draftdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_revision_strict returns one item too few on every page. It lives in src/draftdesk/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 draftdesk layout. [tool_use ...
64,296
3,072
92
115
false
synth:754896125517037232@1.48335
synth
[{"aliases": ["src/draftdesk/audit.py", "audit.py"], "kind": "file", "value": "src/draftdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/draftdesk/rules.py", "rules.py"], "kind": "file", "value": "src/draftdesk/rules.py"}, {"aliase...
[{"answers": ["src/draftdesk/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01184-twin
draftdesk
ctx_compress-s1-01184
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_revision_strict returns one item too few on every page. It lives in src/draftdesk/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 draftdesk layout. [tool_use ...
36,320
1,816
64
115
false
synth:754896125517037232@1.48335
synth
[{"aliases": ["src/draftdesk/audit.py", "audit.py"], "kind": "file", "value": "src/draftdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/draftdesk/rules.py", "rules.py"], "kind": "file", "value": "src/draftdesk/rules.py"}, {"aliase...
[{"answers": ["src/draftdesk/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-01185
headingworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_index_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/headingworks/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping t...
16,579
1,000
48
55
false
synth:3695713398173008604@1
synth
[{"aliases": ["src/headingworks/report.py", "report.py"], "kind": "file", "value": "src/headingworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/headingworks/core.py", "core.py"], "kind": "file", "value": "src/headingworks/core...
[{"answers": ["src/headingworks/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-s1-01185-twin
headingworks
ctx_compress-s1-01185
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_index_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/headingworks/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping t...
7,936
1,000
27
55
false
synth:3695713398173008604@1
synth
[{"aliases": ["src/headingworks/report.py", "report.py"], "kind": "file", "value": "src/headingworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/headingworks/core.py", "core.py"], "kind": "file", "value": "src/headingworks/core...
[{"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/headingworks/report....
ctx_compress-s1-01186
reorderledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_strict ignores the cap, so late attempts wait for minutes. It lives in src/reorderledger/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 reorderled...
85,165
3,072
115
140
false
synth:3299521934826462635@2.0941
synth
[{"aliases": ["src/reorderledger/report.py", "report.py"], "kind": "file", "value": "src/reorderledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/bin_10.csv", "bin_10.csv"], "kind": "file", "value": "data/bin_10.csv"}, {"alia...
[{"answers": ["src/reorderledger/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'...
ctx_compress-s1-01186-twin
reorderledger
ctx_compress-s1-01186
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_strict ignores the cap, so late attempts wait for minutes. It lives in src/reorderledger/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 reorderled...
56,635
2,831
73
140
false
synth:3299521934826462635@2.0941
synth
[{"aliases": ["src/reorderledger/report.py", "report.py"], "kind": "file", "value": "src/reorderledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/bin_10.csv", "bin_10.csv"], "kind": "file", "value": "data/bin_10.csv"}, {"alia...
[{"answers": ["src/reorderledger/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'...
ctx_compress-s1-01187
bearingworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] snap_corridor_strict produces doubled dashes for consecutive separators. It lives in src/bearingworks/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 bearingworks lay...
33,843
1,692
60
73
false
synth:2269679665671192410@1
synth
[{"aliases": ["src/bearingworks/audit.py", "audit.py"], "kind": "file", "value": "src/bearingworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/bearingworks/format.py", "format.py"], "kind": "file", "value": "src/bearingworks/format...
[{"answers": ["src/bearingworks/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01187-twin
bearingworks
ctx_compress-s1-01187
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] snap_corridor_strict produces doubled dashes for consecutive separators. It lives in src/bearingworks/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 bearingworks lay...
16,037
1,000
36
73
false
synth:2269679665671192410@1
synth
[{"aliases": ["src/bearingworks/audit.py", "audit.py"], "kind": "file", "value": "src/bearingworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/bearingworks/format.py", "format.py"], "kind": "file", "value": "src/bearingworks/format...
[{"answers": ["src/bearingworks/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 m...
ctx_compress-s1-01188
receiptdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_receipt returns the upper bound for every value instead of clamping into the range. It lives in src/receiptdesk/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 receiptd...
102,296
3,072
117
126
false
synth:7198261570547427136@2.12715
synth
[{"aliases": ["src/receiptdesk/io.py", "io.py"], "kind": "file", "value": "src/receiptdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/receiptdesk/schema.py", "schema.py"], "kind": "file", "value": "src/receiptdesk/schema.py"}, {"aliases": ["s...
[{"answers": ["src/receiptdesk/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-s1-01188-twin
receiptdesk
ctx_compress-s1-01188
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_receipt returns the upper bound for every value instead of clamping into the range. It lives in src/receiptdesk/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 receiptd...
57,512
2,875
73
126
false
synth:7198261570547427136@2.12715
synth
[{"aliases": ["src/receiptdesk/io.py", "io.py"], "kind": "file", "value": "src/receiptdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/receiptdesk/schema.py", "schema.py"], "kind": "file", "value": "src/receiptdesk/schema.py"}, {"aliases": ["s...
[{"answers": ["src/receiptdesk/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-s1-01189
rosterledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] split_slot returns the upper bound for every value instead of clamping into the range. It lives in src/rosterledger/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the ...
92,717
3,072
118
141
false
synth:6320701203861542245@2.0897
synth
[{"aliases": ["src/rosterledger/report.py", "report.py"], "kind": "file", "value": "src/rosterledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/rosterledger/audit.py", "audit.py"], "kind": "file", "value": "src/rosterledger/au...
[{"answers": ["src/rosterledger/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-s1-01189-twin
rosterledger
ctx_compress-s1-01189
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] split_slot returns the upper bound for every value instead of clamping into the range. It lives in src/rosterledger/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 ...
49,850
2,492
70
141
false
synth:6320701203861542245@2.0897
synth
[{"aliases": ["src/rosterledger/report.py", "report.py"], "kind": "file", "value": "src/rosterledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/rosterledger/audit.py", "audit.py"], "kind": "file", "value": "src/rosterledger/au...
[{"answers": ["src/rosterledger/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-s1-01190
claimflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] mint_grant_once grows linearly instead of doubling. It lives in src/claimflow/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 claimflow layout. [tool_use kind=search na...
99,276
3,072
135
152
false
synth:9204188634513937388@1.88892
synth
[{"aliases": ["src/claimflow/core.py", "core.py"], "kind": "file", "value": "src/claimflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/claimflow/rules.py", "rules.py"], "kind": "file", "value": "src/claimflow/rules.py"}, {"aliases": ["...
[{"answers": ["src/claimflow/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-s1-01190-twin
claimflow
ctx_compress-s1-01190
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] mint_grant_once grows linearly instead of doubling. It lives in src/claimflow/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 claimflow layout. [tool_use kind=search na...
60,698
3,034
91
152
false
synth:9204188634513937388@1.88892
synth
[{"aliases": ["src/claimflow/core.py", "core.py"], "kind": "file", "value": "src/claimflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/claimflow/rules.py", "rules.py"], "kind": "file", "value": "src/claimflow/rules.py"}, {"aliases": ["...
[{"answers": ["src/claimflow/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-s1-01191
ledgerflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_invoice_safe double counts carried amounts when the period closes on a weekend. It lives in src/ledgerflow/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 le...
46,221
2,311
82
95
false
synth:7051933649390451983@1
synth
[{"aliases": ["src/ledgerflow/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/ledgerflow/tables.py", "tables.py"], "kind": "file", "value": "src/ledgerflow/tables.p...
[{"answers": ["src/ledgerflow/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-s1-01191-twin
ledgerflow
ctx_compress-s1-01191
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_invoice_safe double counts carried amounts when the period closes on a weekend. It lives in src/ledgerflow/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 le...
23,363
1,168
50
95
false
synth:7051933649390451983@1
synth
[{"aliases": ["src/ledgerflow/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/ledgerflow/tables.py", "tables.py"], "kind": "file", "value": "src/ledgerflow/tables.p...
[{"answers": ["src/ledgerflow/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01192
draftdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_heading lets defaults overwrite explicit overrides. It lives in src/draftdesk/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 draftdesk layout. [tool_use kind=search...
111,424
3,072
112
125
false
synth:5108174814173866601@1.80865
synth
[{"aliases": ["src/draftdesk/io.py", "io.py"], "kind": "file", "value": "src/draftdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/draft_4.csv", "draft_4.csv"], "kind": "file", "value": "data/draft_4.csv"}, {"aliases": ["data/margin_2.csv", "...
[{"answers": ["src/draftdesk/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-s1-01192-twin
draftdesk
ctx_compress-s1-01192
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_heading lets defaults overwrite explicit overrides. It lives in src/draftdesk/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 draftdesk layout. [tool_use kind=search...
66,696
3,072
74
125
false
synth:5108174814173866601@1.80865
synth
[{"aliases": ["src/draftdesk/io.py", "io.py"], "kind": "file", "value": "src/draftdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/draft_4.csv", "draft_4.csv"], "kind": "file", "value": "data/draft_4.csv"}, {"aliases": ["data/margin_2.csv", "...
[{"answers": ["src/draftdesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-01193
beaconforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_tile misses keys that differ only in case. It lives in src/beaconforge/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 beaconforge layout. [tool_use kind=s...
69,395
3,072
104
111
false
synth:7298574307624580441@1.2524
synth
[{"aliases": ["src/beaconforge/limits.py", "limits.py"], "kind": "file", "value": "src/beaconforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/beaconforge/audit.py", "audit.py"], "kind": "file", "value": "src/beaconforge/audit....
[{"answers": ["src/beaconforge/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-s1-01193-twin
beaconforge
ctx_compress-s1-01193
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_tile misses keys that differ only in case. It lives in src/beaconforge/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 beaconforge layout. [tool_use kind=s...
38,226
1,911
58
111
false
synth:7298574307624580441@1.2524
synth
[{"aliases": ["src/beaconforge/limits.py", "limits.py"], "kind": "file", "value": "src/beaconforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/beaconforge/audit.py", "audit.py"], "kind": "file", "value": "src/beaconforge/audit....
[{"answers": ["src/beaconforge/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-s1-01194
clausekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_section_batch keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/clausekit/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 clausekit...
46,504
2,325
79
96
false
synth:647107971209308624@1.13807
synth
[{"aliases": ["src/clausekit/cli.py", "cli.py"], "kind": "file", "value": "src/clausekit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/clausekit/limits.py", "limits.py"], "kind": "file", "value": "src/clausekit/limits.py"}, {"aliases": ["src...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/clausekit/compat.py"...
ctx_compress-s1-01194-twin
clausekit
ctx_compress-s1-01194
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_section_batch keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/clausekit/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 clausekit...
31,062
1,553
53
96
false
synth:647107971209308624@1.13807
synth
[{"aliases": ["src/clausekit/cli.py", "cli.py"], "kind": "file", "value": "src/clausekit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/clausekit/limits.py", "limits.py"], "kind": "file", "value": "src/clausekit/limits.py"}, {"aliases": ["src...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/clausekit/tables.py", "tables.py"...
ctx_compress-s1-01195
traceledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_sample misses keys that differ only in case. It lives in src/traceledger/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 traceledger layout. [tool_use kind=se...
25,374
1,268
46
53
false
synth:1277216997547752624@1
synth
[{"aliases": ["src/traceledger/compat.py", "compat.py"], "kind": "file", "value": "src/traceledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/traceledger/io.py", "io.py"], "kind": "file", "value": "src/traceledger/io.py"}, {"a...
[{"answers": ["src/traceledger/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-s1-01195-twin
traceledger
ctx_compress-s1-01195
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_sample misses keys that differ only in case. It lives in src/traceledger/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 traceledger layout. [tool_use kind=se...
12,173
1,000
26
53
false
synth:1277216997547752624@1
synth
[{"aliases": ["src/traceledger/compat.py", "compat.py"], "kind": "file", "value": "src/traceledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/traceledger/io.py", "io.py"], "kind": "file", "value": "src/traceledger/io.py"}, {"a...
[{"answers": ["src/traceledger/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-s1-01196
seriesdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_span raises KeyError for a missing key instead of returning the default. It lives in src/seriesdesk/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 seriesdesk layout...
70,074
3,072
104
123
false
synth:6259468984040453723@2.09485
synth
[{"aliases": ["src/seriesdesk/io.py", "io.py"], "kind": "file", "value": "src/seriesdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/seriesdesk/core.py", "core.py"], "kind": "file", "value": "src/seriesdesk/core.py"}, {"aliases": ["src/seriesd...
[{"answers": ["src/seriesdesk/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-s1-01196-twin
seriesdesk
ctx_compress-s1-01196
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_span raises KeyError for a missing key instead of returning the default. It lives in src/seriesdesk/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 seriesdesk layout...
42,369
2,118
62
123
false
synth:6259468984040453723@2.09485
synth
[{"aliases": ["src/seriesdesk/io.py", "io.py"], "kind": "file", "value": "src/seriesdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/seriesdesk/core.py", "core.py"], "kind": "file", "value": "src/seriesdesk/core.py"}, {"aliases": ["src/seriesd...
[{"answers": ["src/seriesdesk/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 rece...
ctx_compress-s1-01197
receiptledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_refund ignores the cap, so late attempts wait for minutes. It lives in src/receiptledger/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 receiptledger layout. ...
73,025
3,072
102
129
false
synth:7986545230261687858@2.1426
synth
[{"aliases": ["src/receiptledger/store.py", "store.py"], "kind": "file", "value": "src/receiptledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/refund_1.csv", "refund_1.csv"], "kind": "file", "value": "data/refund_1.csv"}, {"alia...
[{"answers": ["src/receiptledger/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01197-twin
receiptledger
ctx_compress-s1-01197
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_refund ignores the cap, so late attempts wait for minutes. It lives in src/receiptledger/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 receiptledger layout. ...
49,657
2,482
70
129
false
synth:7986545230261687858@2.1426
synth
[{"aliases": ["src/receiptledger/store.py", "store.py"], "kind": "file", "value": "src/receiptledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/refund_1.csv", "refund_1.csv"], "kind": "file", "value": "data/refund_1.csv"}, {"alia...
[{"answers": ["src/receiptledger/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01198
creditkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_payout ignores the cap, so late attempts wait for minutes. It lives in src/creditkit/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 creditkit layout. [tool_use ...
23,440
1,172
40
49
false
synth:2389458382335178064@1
synth
[{"aliases": ["src/creditkit/core.py", "core.py"], "kind": "file", "value": "src/creditkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/creditkit/report.py", "report.py"], "kind": "file", "value": "src/creditkit/report.py"}, {"aliases":...
[{"answers": ["src/creditkit/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-s1-01198-twin
creditkit
ctx_compress-s1-01198
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_payout ignores the cap, so late attempts wait for minutes. It lives in src/creditkit/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 creditkit layout. [tool_use ...
9,529
1,000
25
49
false
synth:2389458382335178064@1
synth
[{"aliases": ["src/creditkit/core.py", "core.py"], "kind": "file", "value": "src/creditkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/creditkit/report.py", "report.py"], "kind": "file", "value": "src/creditkit/report.py"}, {"aliases":...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/creditkit/tables.py", "tables.py"...
ctx_compress-s1-01199
tokenledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] mint_session_all produces doubled dashes for consecutive separators. It lives in src/tokenledger/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 tokenledger layout. [...
56,576
2,828
92
101
false
synth:7484422071667526045@1
synth
[{"aliases": ["src/tokenledger/audit.py", "audit.py"], "kind": "file", "value": "src/tokenledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/tokenledger/rules.py", "rules.py"], "kind": "file", "value": "src/tokenledger/rules.py"}, ...
[{"answers": ["src/tokenledger/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-s1-01199-twin
tokenledger
ctx_compress-s1-01199
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] mint_session_all produces doubled dashes for consecutive separators. It lives in src/tokenledger/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 tokenledger layout. [...
25,442
1,272
50
101
false
synth:7484422071667526045@1
synth
[{"aliases": ["src/tokenledger/audit.py", "audit.py"], "kind": "file", "value": "src/tokenledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/tokenledger/rules.py", "rules.py"], "kind": "file", "value": "src/tokenledger/rules.py"}, ...
[{"answers": ["src/tokenledger/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-s1-01200
probeforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/probeforge/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 probeforge layout. [tool_use kind=search ...
57,507
2,875
95
102
false
synth:6871234658798587482@1
synth
[{"aliases": ["src/probeforge/format.py", "format.py"], "kind": "file", "value": "src/probeforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/probeforge/core.py", "core.py"], "kind": "file", "value": "src/probeforge/core.py"}, {...
[{"answers": ["src/probeforge/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-s1-01200-twin
probeforge
ctx_compress-s1-01200
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/probeforge/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 probeforge layout. [tool_use kind=search ...
32,603
1,630
61
102
false
synth:6871234658798587482@1
synth
[{"aliases": ["src/probeforge/format.py", "format.py"], "kind": "file", "value": "src/probeforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/probeforge/core.py", "core.py"], "kind": "file", "value": "src/probeforge/core.py"}, {...
[{"answers": ["src/probeforge/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-s1-01201
bearingflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_tile lets defaults overwrite explicit overrides. It lives in src/bearingflow/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 bearingflow layout. [tool_use k...
58,401
2,920
60
65
false
synth:9140125026381439300@1
synth
[{"aliases": ["src/bearingflow/compat.py", "compat.py"], "kind": "file", "value": "src/bearingflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/tile_3.csv", "tile_3.csv"], "kind": "file", "value": "data/tile_3.csv"}, {"aliases"...
[{"answers": ["src/bearingflow/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-s1-01201-twin
bearingflow
ctx_compress-s1-01201
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 lets defaults overwrite explicit overrides. It lives in src/bearingflow/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 bearingflow layout. [tool_use k...
28,192
1,409
38
65
false
synth:9140125026381439300@1
synth
[{"aliases": ["src/bearingflow/compat.py", "compat.py"], "kind": "file", "value": "src/bearingflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/tile_3.csv", "tile_3.csv"], "kind": "file", "value": "data/tile_3.csv"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["schema.py", "src/bearingflow/schema.p...
ctx_compress-s1-01202
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] smooth_probe averages one value too many once the window is full. It lives in src/bucketkit/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 bucketkit layout. [tool_use ...
89,605
3,072
108
125
false
synth:4046312873389788218@2.32135
synth
[{"aliases": ["src/bucketkit/core.py", "core.py"], "kind": "file", "value": "src/bucketkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/bucketkit/format.py", "format.py"], "kind": "file", "value": "src/bucketkit/format.py"}, {"aliases":...
[{"answers": ["src/bucketkit/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-s1-01202-twin
bucketkit
ctx_compress-s1-01202
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 averages one value too many once the window is full. It lives in src/bucketkit/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 bucketkit layout. [tool_use ...
52,279
2,613
63
125
false
synth:4046312873389788218@2.32135
synth
[{"aliases": ["src/bucketkit/core.py", "core.py"], "kind": "file", "value": "src/bucketkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/bucketkit/format.py", "format.py"], "kind": "file", "value": "src/bucketkit/format.py"}, {"aliases":...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-01203
grantkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 leaves a trailing dash when the title ends in punctuation. It lives in src/grantkit/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 grantkit layout. [too...
43,663
2,183
71
90
false
synth:5757020553138014929@1
synth
[{"aliases": ["src/grantkit/rules.py", "rules.py"], "kind": "file", "value": "src/grantkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/grantkit/audit.py", "audit.py"], "kind": "file", "value": "src/grantkit/audit.py"}, {"aliases": ...
[{"answers": ["src/grantkit/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-s1-01203-twin
grantkit
ctx_compress-s1-01203
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 leaves a trailing dash when the title ends in punctuation. It lives in src/grantkit/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 grantkit layout. [too...
19,777
1,000
41
90
false
synth:5757020553138014929@1
synth
[{"aliases": ["src/grantkit/rules.py", "rules.py"], "kind": "file", "value": "src/grantkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/grantkit/audit.py", "audit.py"], "kind": "file", "value": "src/grantkit/audit.py"}, {"aliases": ...
[{"answers": ["src/grantkit/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-s1-01204
rosterledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_cadence skips the first page; page 1 returns what should be page 2. It lives in src/rosterledger/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 rosterledge...
40,476
2,023
69
74
false
synth:7293781516793373105@1
synth
[{"aliases": ["src/rosterledger/compat.py", "compat.py"], "kind": "file", "value": "src/rosterledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/rosterledger/rules.py", "rules.py"], "kind": "file", "value": "src/rosterledger/ru...
[{"answers": ["src/rosterledger/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-s1-01204-twin
rosterledger
ctx_compress-s1-01204
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_cadence skips the first page; page 1 returns what should be page 2. It lives in src/rosterledger/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 rosterledge...
17,278
1,000
35
74
false
synth:7293781516793373105@1
synth
[{"aliases": ["src/rosterledger/compat.py", "compat.py"], "kind": "file", "value": "src/rosterledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/rosterledger/rules.py", "rules.py"], "kind": "file", "value": "src/rosterledger/ru...
[{"answers": ["src/rosterledger/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-s1-01205
draftkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_index returns the upper bound for every value instead of clamping into the range. It lives in src/draftkit/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 dr...
79,546
3,072
96
119
false
synth:561884515644260910@1.5015
synth
[{"aliases": ["src/draftkit/audit.py", "audit.py"], "kind": "file", "value": "src/draftkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/footnote_0.csv", "footnote_0.csv"], "kind": "file", "value": "data/footnote_0.csv"}, {"aliases"...
[{"answers": ["src/draftkit/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...
ctx_compress-s1-01205-twin
draftkit
ctx_compress-s1-01205
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_index returns the upper bound for every value instead of clamping into the range. It lives in src/draftkit/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 dr...
44,636
2,231
68
119
false
synth:561884515644260910@1.5015
synth
[{"aliases": ["src/draftkit/audit.py", "audit.py"], "kind": "file", "value": "src/draftkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/footnote_0.csv", "footnote_0.csv"], "kind": "file", "value": "data/footnote_0.csv"}, {"aliases"...
[{"answers": ["src/draftkit/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 ...