Datasets:
microtensor-io
/
agent-sessions

Dataset card Data Studio Files
xet
 Community
item_id
stringlengths
21
26
family
stringclasses
383 values
twin_of
stringlengths
0
21
system
stringlengths
1.51k
1.51k
user
stringlengths
18.6k
288k
input_tokens
int64
6.38k
123k
budget_tokens
int64
1k
3.07k
cut
int64
19
151
steps
int64
32
166
resolved
bool
1 class
image
stringlengths
25
33
source
stringclasses
1 value
facts
stringlengths
585
1.33k
probes
stringlengths
1.29k
1.44k
ctx_compress-s1-00604
principalkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim_once leaves a trailing dash when the title ends in punctuation. It lives in src/principalkit/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 principalk...
83,582
3,072
108
127
false
synth:3938750059244509364@1.80367
synth
[{"aliases": ["src/principalkit/report.py", "report.py"], "kind": "file", "value": "src/principalkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/nonce_5.csv", "nonce_5.csv"], "kind": "file", "value": "data/nonce_5.csv"}, {"ali...
[{"answers": ["src/principalkit/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-s1-00604-twin
principalkit
ctx_compress-s1-00604
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim_once leaves a trailing dash when the title ends in punctuation. It lives in src/principalkit/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 principalk...
47,416
2,370
74
127
false
synth:3938750059244509364@1.80367
synth
[{"aliases": ["src/principalkit/report.py", "report.py"], "kind": "file", "value": "src/principalkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/nonce_5.csv", "nonce_5.csv"], "kind": "file", "value": "data/nonce_5.csv"}, {"ali...
[{"answers": ["src/principalkit/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-s1-00605
tiledesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_corridor_strict misses keys that differ only in case. It lives in src/tiledesk/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 tiledesk layout. [tool_use ki...
54,394
2,719
99
116
false
synth:6530038337176974848@1.25602
synth
[{"aliases": ["src/tiledesk/core.py", "core.py"], "kind": "file", "value": "src/tiledesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/bearing_3.csv", "bearing_3.csv"], "kind": "file", "value": "data/bearing_3.csv"}, {"aliases": ["src/t...
[{"answers": ["src/tiledesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00605-twin
tiledesk
ctx_compress-s1-00605
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_corridor_strict misses keys that differ only in case. It lives in src/tiledesk/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 tiledesk layout. [tool_use ki...
31,274
1,563
61
116
false
synth:6530038337176974848@1.25602
synth
[{"aliases": ["src/tiledesk/core.py", "core.py"], "kind": "file", "value": "src/tiledesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/bearing_3.csv", "bearing_3.csv"], "kind": "file", "value": "data/bearing_3.csv"}, {"aliases": ["src/t...
[{"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/tiledesk/core.py"], "k...
ctx_compress-s1-00606
waypointforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_beacon lets defaults overwrite explicit overrides. It lives in src/waypointforge/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 waypointforge layout. [tool...
88,125
3,072
104
115
false
synth:8630937781935516041@1.72313
synth
[{"aliases": ["src/waypointforge/compat.py", "compat.py"], "kind": "file", "value": "src/waypointforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/corridor_2.csv", "corridor_2.csv"], "kind": "file", "value": "data/corridor_2.c...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/waypointforge/compat...
ctx_compress-s1-00606-twin
waypointforge
ctx_compress-s1-00606
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_beacon lets defaults overwrite explicit overrides. It lives in src/waypointforge/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 waypointforge layout. [tool...
47,470
2,373
64
115
false
synth:8630937781935516041@1.72313
synth
[{"aliases": ["src/waypointforge/compat.py", "compat.py"], "kind": "file", "value": "src/waypointforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/corridor_2.csv", "corridor_2.csv"], "kind": "file", "value": "data/corridor_2.c...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/waypointforge/audit.p...
ctx_compress-s1-00607
scopeworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_nonce_window double counts carried amounts when the period closes on a weekend. It lives in src/scopeworks/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 sc...
48,761
2,438
93
100
false
synth:4565868063986264341@1.48542
synth
[{"aliases": ["src/scopeworks/format.py", "format.py"], "kind": "file", "value": "src/scopeworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/scopeworks/rules.py", "rules.py"], "kind": "file", "value": "src/scopeworks/rules.py"}...
[{"answers": ["src/scopeworks/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-s1-00607-twin
scopeworks
ctx_compress-s1-00607
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_nonce_window double counts carried amounts when the period closes on a weekend. It lives in src/scopeworks/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 sc...
24,159
1,207
51
100
false
synth:4565868063986264341@1.48542
synth
[{"aliases": ["src/scopeworks/format.py", "format.py"], "kind": "file", "value": "src/scopeworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/scopeworks/rules.py", "rules.py"], "kind": "file", "value": "src/scopeworks/rules.py"}...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/scopeworks/tables.py", "tables.py...
ctx_compress-s1-00608
corridordesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/corridordesk/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 corridordesk layout. [tool_use kind=search n...
98,436
3,072
97
110
false
synth:8325413707671051741@2.22987
synth
[{"aliases": ["src/corridordesk/core.py", "core.py"], "kind": "file", "value": "src/corridordesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/waypoint_2.csv", "waypoint_2.csv"], "kind": "file", "value": "data/waypoint_2.csv"}, {"aliase...
[{"answers": ["src/corridordesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00608-twin
corridordesk
ctx_compress-s1-00608
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/corridordesk/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 corridordesk layout. [tool_use kind=search n...
58,761
2,938
55
110
false
synth:8325413707671051741@2.22987
synth
[{"aliases": ["src/corridordesk/core.py", "core.py"], "kind": "file", "value": "src/corridordesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/waypoint_2.csv", "waypoint_2.csv"], "kind": "file", "value": "data/waypoint_2.csv"}, {"aliase...
[{"answers": ["src/corridordesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00609
shipmentforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_reorder grows linearly instead of doubling. It lives in src/shipmentforge/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 shipmentforge layout. [tool_use k...
24,096
1,204
38
45
false
synth:6228296037336157148@1
synth
[{"aliases": ["src/shipmentforge/report.py", "report.py"], "kind": "file", "value": "src/shipmentforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/shipment_4.csv", "shipment_4.csv"], "kind": "file", "value": "data/shipment_4.c...
[{"answers": ["src/shipmentforge/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'...
ctx_compress-s1-00609-twin
shipmentforge
ctx_compress-s1-00609
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/shipmentforge/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 shipmentforge layout. [tool_use k...
10,156
1,000
27
45
false
synth:6228296037336157148@1
synth
[{"aliases": ["src/shipmentforge/report.py", "report.py"], "kind": "file", "value": "src/shipmentforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/shipment_4.csv", "shipment_4.csv"], "kind": "file", "value": "data/shipment_4.c...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/shipmentforge/limits...
ctx_compress-s1-00610
creditdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_receipt ignores the cap, so late attempts wait for minutes. It lives in src/creditdesk/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 creditdesk layout. [...
39,826
1,991
86
103
false
synth:3934405281586133663@1.1452
synth
[{"aliases": ["src/creditdesk/report.py", "report.py"], "kind": "file", "value": "src/creditdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/creditdesk/limits.py", "limits.py"], "kind": "file", "value": "src/creditdesk/limits.p...
[{"answers": ["src/creditdesk/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00610-twin
creditdesk
ctx_compress-s1-00610
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_receipt ignores the cap, so late attempts wait for minutes. It lives in src/creditdesk/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 creditdesk layout. [...
26,850
1,342
48
103
false
synth:3934405281586133663@1.1452
synth
[{"aliases": ["src/creditdesk/report.py", "report.py"], "kind": "file", "value": "src/creditdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/creditdesk/limits.py", "limits.py"], "kind": "file", "value": "src/creditdesk/limits.p...
[{"answers": ["src/creditdesk/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00611
marginforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_section_strict raises MarginDriftError for bytes input in strict mode instead of decoding it. It lives in src/marginforge/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 b...
32,006
1,600
71
80
false
synth:3722813415066933849@1
synth
[{"aliases": ["src/marginforge/format.py", "format.py"], "kind": "file", "value": "src/marginforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/marginforge/limits.py", "limits.py"], "kind": "file", "value": "src/marginforge/limi...
[{"answers": ["src/marginforge/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-s1-00611-twin
marginforge
ctx_compress-s1-00611
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_section_strict raises MarginDriftError for bytes input in strict mode instead of decoding it. It lives in src/marginforge/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 b...
19,523
1,000
41
80
false
synth:3722813415066933849@1
synth
[{"aliases": ["src/marginforge/format.py", "format.py"], "kind": "file", "value": "src/marginforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/marginforge/limits.py", "limits.py"], "kind": "file", "value": "src/marginforge/limi...
[{"answers": ["src/marginforge/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-s1-00612
claimforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_nonce raises NonceBoundError for bytes input in strict mode instead of decoding it. It lives in src/claimforge/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 th...
88,409
3,072
116
127
false
synth:4406084661413823273@1.64233
synth
[{"aliases": ["src/claimforge/format.py", "format.py"], "kind": "file", "value": "src/claimforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/claimforge/io.py", "io.py"], "kind": "file", "value": "src/claimforge/io.py"}, {"alias...
[{"answers": ["src/claimforge/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-s1-00612-twin
claimforge
ctx_compress-s1-00612
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_nonce raises NonceBoundError for bytes input in strict mode instead of decoding it. It lives in src/claimforge/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 th...
47,576
2,378
62
127
false
synth:4406084661413823273@1.64233
synth
[{"aliases": ["src/claimforge/format.py", "format.py"], "kind": "file", "value": "src/claimforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/claimforge/io.py", "io.py"], "kind": "file", "value": "src/claimforge/io.py"}, {"alias...
[{"answers": ["src/claimforge/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-s1-00613
lexemeforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_stem skips the first page; page 1 returns what should be page 2. It lives in src/lexemeforge/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 lexemeforge l...
94,106
3,072
109
130
false
synth:2410504097162427657@1.83268
synth
[{"aliases": ["src/lexemeforge/limits.py", "limits.py"], "kind": "file", "value": "src/lexemeforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/lexemeforge/format.py", "format.py"], "kind": "file", "value": "src/lexemeforge/form...
[{"answers": ["src/lexemeforge/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-s1-00613-twin
lexemeforge
ctx_compress-s1-00613
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_stem skips the first page; page 1 returns what should be page 2. It lives in src/lexemeforge/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 lexemeforge l...
50,935
2,546
69
130
false
synth:2410504097162427657@1.83268
synth
[{"aliases": ["src/lexemeforge/limits.py", "limits.py"], "kind": "file", "value": "src/lexemeforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/lexemeforge/format.py", "format.py"], "kind": "file", "value": "src/lexemeforge/form...
[{"answers": ["src/lexemeforge/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-s1-00614
captionledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] dedupe_slug_page averages one value too many once the window is full. It lives in src/captionledger/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 captionledger layo...
99,768
3,072
104
113
false
synth:8265660567294413440@1.84885
synth
[{"aliases": ["src/captionledger/audit.py", "audit.py"], "kind": "file", "value": "src/captionledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/captionledger/schema.py", "schema.py"], "kind": "file", "value": "src/captionledger/sc...
[{"answers": ["src/captionledger/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 ...
ctx_compress-s1-00614-twin
captionledger
ctx_compress-s1-00614
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] dedupe_slug_page averages one value too many once the window is full. It lives in src/captionledger/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 captionledger layo...
59,377
2,968
66
113
false
synth:8265660567294413440@1.84885
synth
[{"aliases": ["src/captionledger/audit.py", "audit.py"], "kind": "file", "value": "src/captionledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/captionledger/schema.py", "schema.py"], "kind": "file", "value": "src/captionledger/sc...
[{"answers": ["src/captionledger/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 ...
ctx_compress-s1-00615
serieskit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe double counts carried amounts when the period closes on a weekend. It lives in src/serieskit/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 serieskit layout....
80,744
3,072
107
124
false
synth:4019004338489753702@1.7286
synth
[{"aliases": ["src/serieskit/core.py", "core.py"], "kind": "file", "value": "src/serieskit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/bucket_5.csv", "bucket_5.csv"], "kind": "file", "value": "data/bucket_5.csv"}, {"aliases": ["src/se...
[{"answers": ["src/serieskit/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-s1-00615-twin
serieskit
ctx_compress-s1-00615
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe double counts carried amounts when the period closes on a weekend. It lives in src/serieskit/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 serieskit layout....
36,498
1,824
59
124
false
synth:4019004338489753702@1.7286
synth
[{"aliases": ["src/serieskit/core.py", "core.py"], "kind": "file", "value": "src/serieskit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/bucket_5.csv", "bucket_5.csv"], "kind": "file", "value": "data/bucket_5.csv"}, {"aliases": ["src/se...
[{"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": ["counter_9.csv", "data/counter_9.csv"]...
ctx_compress-s1-00616
legdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_leg double counts carried amounts when the period closes on a weekend. It lives in src/legdesk/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 legdesk layo...
79,468
3,072
94
113
false
synth:1579850673154478305@1.85945
synth
[{"aliases": ["src/legdesk/format.py", "format.py"], "kind": "file", "value": "src/legdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/waypoint_5.csv", "waypoint_5.csv"], "kind": "file", "value": "data/waypoint_5.csv"}, {"alia...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/legdesk/limits.py"],...
ctx_compress-s1-00616-twin
legdesk
ctx_compress-s1-00616
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_leg double counts carried amounts when the period closes on a weekend. It lives in src/legdesk/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 legdesk layo...
37,945
1,897
55
113
false
synth:1579850673154478305@1.85945
synth
[{"aliases": ["src/legdesk/format.py", "format.py"], "kind": "file", "value": "src/legdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/waypoint_5.csv", "waypoint_5.csv"], "kind": "file", "value": "data/waypoint_5.csv"}, {"alia...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["corridor_6.csv", "data/corridor_6.csv...
ctx_compress-s1-00617
quotaworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/quotaworks/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 quotaworks layout. [tool_use kind=sea...
75,475
3,072
78
85
false
synth:8688061379774516908@1.3181
synth
[{"aliases": ["src/quotaworks/compat.py", "compat.py"], "kind": "file", "value": "src/quotaworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/quotaworks/io.py", "io.py"], "kind": "file", "value": "src/quotaworks/io.py"}, {"alias...
[{"answers": ["src/quotaworks/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-s1-00617-twin
quotaworks
ctx_compress-s1-00617
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 grows linearly instead of doubling. It lives in src/quotaworks/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 quotaworks layout. [tool_use kind=sea...
37,807
1,890
44
85
false
synth:8688061379774516908@1.3181
synth
[{"aliases": ["src/quotaworks/compat.py", "compat.py"], "kind": "file", "value": "src/quotaworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/quotaworks/io.py", "io.py"], "kind": "file", "value": "src/quotaworks/io.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": ["format.py", "src/quotaworks/format.py...
ctx_compress-s1-00618
legforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon leaves a trailing dash when the title ends in punctuation. It lives in src/legforge/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 legforge layout. [to...
35,934
1,796
52
63
false
synth:1089626759523681876@1
synth
[{"aliases": ["src/legforge/compat.py", "compat.py"], "kind": "file", "value": "src/legforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/legforge/limits.py", "limits.py"], "kind": "file", "value": "src/legforge/limits.py"}, {"a...
[{"answers": ["src/legforge/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 mos...
ctx_compress-s1-00618-twin
legforge
ctx_compress-s1-00618
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon leaves a trailing dash when the title ends in punctuation. It lives in src/legforge/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 legforge layout. [to...
16,627
1,000
30
63
false
synth:1089626759523681876@1
synth
[{"aliases": ["src/legforge/compat.py", "compat.py"], "kind": "file", "value": "src/legforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/legforge/limits.py", "limits.py"], "kind": "file", "value": "src/legforge/limits.py"}, {"a...
[{"answers": ["src/legforge/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 mos...
ctx_compress-s1-00619
reorderdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_pallet raises WarehouseShapeError for bytes input in strict mode instead of decoding it. It lives in src/reorderdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping ...
103,135
3,072
116
127
false
synth:3041464222269180198@1.57862
synth
[{"aliases": ["src/reorderdesk/store.py", "store.py"], "kind": "file", "value": "src/reorderdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sku_11.csv", "sku_11.csv"], "kind": "file", "value": "data/sku_11.csv"}, {"aliases": ["da...
[{"answers": ["src/reorderdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00619-twin
reorderdesk
ctx_compress-s1-00619
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_pallet raises WarehouseShapeError for bytes input in strict mode instead of decoding it. It lives in src/reorderdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping ...
42,234
2,111
58
127
false
synth:3041464222269180198@1.57862
synth
[{"aliases": ["src/reorderdesk/store.py", "store.py"], "kind": "file", "value": "src/reorderdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sku_11.csv", "sku_11.csv"], "kind": "file", "value": "data/sku_11.csv"}, {"aliases": ["da...
[{"answers": ["src/reorderdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00620
headingdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_revision leaves a trailing dash when the title ends in punctuation. It lives in src/headingdesk/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 headingdesk la...
18,568
1,000
47
50
false
synth:2291376594152224876@1
synth
[{"aliases": ["src/headingdesk/rules.py", "rules.py"], "kind": "file", "value": "src/headingdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/margin_0.csv", "margin_0.csv"], "kind": "file", "value": "data/margin_0.csv"}, {"aliases"...
[{"answers": ["tests/test_rules.py", "test_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 mo...
ctx_compress-s1-00621
balancekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_refund rejects a range where lo equals hi, which is a valid single point range. It lives in src/balancekit/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 ba...
35,583
1,779
51
64
false
synth:4417513050783524842@1
synth
[{"aliases": ["src/balancekit/compat.py", "compat.py"], "kind": "file", "value": "src/balancekit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/refund_0.csv", "refund_0.csv"], "kind": "file", "value": "data/refund_0.csv"}, {"alia...
[{"answers": ["src/balancekit/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-s1-00621-twin
balancekit
ctx_compress-s1-00621
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_refund rejects a range where lo equals hi, which is a valid single point range. It lives in src/balancekit/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 ba...
17,965
1,000
31
64
false
synth:4417513050783524842@1
synth
[{"aliases": ["src/balancekit/compat.py", "compat.py"], "kind": "file", "value": "src/balancekit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/refund_0.csv", "refund_0.csv"], "kind": "file", "value": "data/refund_0.csv"}, {"alia...
[{"answers": ["src/balancekit/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-s1-00622
windowflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_shift drops override values of 0, False and empty string. It lives in src/windowflow/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 windowflow layout. [too...
41,172
2,058
76
85
false
synth:2500768240705312338@1
synth
[{"aliases": ["src/windowflow/report.py", "report.py"], "kind": "file", "value": "src/windowflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/cadence_2.csv", "cadence_2.csv"], "kind": "file", "value": "data/cadence_2.csv"}, {"a...
[{"answers": ["src/windowflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-00622-twin
windowflow
ctx_compress-s1-00622
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_shift drops override values of 0, False and empty string. It lives in src/windowflow/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 windowflow layout. [too...
19,823
1,000
42
85
false
synth:2500768240705312338@1
synth
[{"aliases": ["src/windowflow/report.py", "report.py"], "kind": "file", "value": "src/windowflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/cadence_2.csv", "cadence_2.csv"], "kind": "file", "value": "data/cadence_2.csv"}, {"a...
[{"answers": ["src/windowflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-00623
lexemeflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_caption_window keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/lexemeflow/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 le...
90,888
3,072
114
125
false
synth:6585120801610667426@1.58252
synth
[{"aliases": ["src/lexemeflow/store.py", "store.py"], "kind": "file", "value": "src/lexemeflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/lexemeflow/tables.py", "tables.py"], "kind": "file", "value": "src/lexemeflow/tables.py"}, {...
[{"answers": ["src/lexemeflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00623-twin
lexemeflow
ctx_compress-s1-00623
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_caption_window keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/lexemeflow/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 le...
46,446
2,322
66
125
false
synth:6585120801610667426@1.58252
synth
[{"aliases": ["src/lexemeflow/store.py", "store.py"], "kind": "file", "value": "src/lexemeflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/lexemeflow/tables.py", "tables.py"], "kind": "file", "value": "src/lexemeflow/tables.py"}, {...
[{"answers": ["src/lexemeflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00624
creditworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_ledger rejects a range where lo equals hi, which is a valid single point range. It lives in src/creditworks/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 c...
24,020
1,201
42
45
false
synth:2317291511554236273@1
synth
[{"aliases": ["src/creditworks/format.py", "format.py"], "kind": "file", "value": "src/creditworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/creditworks/store.py", "store.py"], "kind": "file", "value": "src/creditworks/store....
[{"answers": ["tests/test_format.py", "test_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-s1-00624-twin
creditworks
ctx_compress-s1-00624
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_ledger rejects a range where lo equals hi, which is a valid single point range. It lives in src/creditworks/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 c...
14,916
1,000
24
45
false
synth:2317291511554236273@1
synth
[{"aliases": ["src/creditworks/format.py", "format.py"], "kind": "file", "value": "src/creditworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/creditworks/store.py", "store.py"], "kind": "file", "value": "src/creditworks/store....
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_limits.py", "tests/test_limits.p...
ctx_compress-s1-00625
calendarworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_calendar_window raises RosterStateError for bytes input in strict mode instead of decoding it. It lives in src/calendarworks/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 mapp...
43,362
2,168
88
111
false
synth:728280953970556853@1.08567
synth
[{"aliases": ["src/calendarworks/io.py", "io.py"], "kind": "file", "value": "src/calendarworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/calendarworks/tables.py", "tables.py"], "kind": "file", "value": "src/calendarworks/tables.py"}, {"alias...
[{"answers": ["src/calendarworks/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 r...
ctx_compress-s1-00625-twin
calendarworks
ctx_compress-s1-00625
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_calendar_window raises RosterStateError for bytes input in strict mode instead of decoding it. It lives in src/calendarworks/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 mapp...
25,107
1,255
52
111
false
synth:728280953970556853@1.08567
synth
[{"aliases": ["src/calendarworks/io.py", "io.py"], "kind": "file", "value": "src/calendarworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/calendarworks/tables.py", "tables.py"], "kind": "file", "value": "src/calendarworks/tables.py"}, {"alias...
[{"answers": ["src/calendarworks/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 r...
ctx_compress-s1-00626
beaconflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_bearing raises ValueError when a value itself contains an equals sign. It lives in src/beaconflow/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 beaconflow l...
19,614
1,000
31
36
false
synth:8856653975797936120@1
synth
[{"aliases": ["src/beaconflow/rules.py", "rules.py"], "kind": "file", "value": "src/beaconflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/beaconflow/format.py", "format.py"], "kind": "file", "value": "src/beaconflow/format.py"}, {...
[{"answers": ["src/beaconflow/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00627
invoicedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_credit_strict rejects a range where lo equals hi, which is a valid single point range. It lives in src/invoicedesk/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 ...
79,139
3,072
99
118
false
synth:2465971219291684711@1.9527
synth
[{"aliases": ["src/invoicedesk/core.py", "core.py"], "kind": "file", "value": "src/invoicedesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/invoicedesk/cli.py", "cli.py"], "kind": "file", "value": "src/invoicedesk/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": ["limits.py", "src/invoicedesk/limits.p...
ctx_compress-s1-00627-twin
invoicedesk
ctx_compress-s1-00627
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_credit_strict rejects a range where lo equals hi, which is a valid single point range. It lives in src/invoicedesk/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 ...
47,238
2,361
57
118
false
synth:2465971219291684711@1.9527
synth
[{"aliases": ["src/invoicedesk/core.py", "core.py"], "kind": "file", "value": "src/invoicedesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/invoicedesk/cli.py", "cli.py"], "kind": "file", "value": "src/invoicedesk/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": ["test_limits.py", "tests/test_limits.p...
ctx_compress-s1-00628
probeledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_span_all produces doubled dashes for consecutive separators. It lives in src/probeledger/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 probeledger layout. [t...
82,151
3,072
100
119
false
synth:2977098679141793993@1.78425
synth
[{"aliases": ["src/probeledger/compat.py", "compat.py"], "kind": "file", "value": "src/probeledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/probeledger/store.py", "store.py"], "kind": "file", "value": "src/probeledger/store....
[{"answers": ["src/probeledger/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-s1-00628-twin
probeledger
ctx_compress-s1-00628
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_span_all produces doubled dashes for consecutive separators. It lives in src/probeledger/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 probeledger layout. [t...
46,330
2,316
62
119
false
synth:2977098679141793993@1.78425
synth
[{"aliases": ["src/probeledger/compat.py", "compat.py"], "kind": "file", "value": "src/probeledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/probeledger/store.py", "store.py"], "kind": "file", "value": "src/probeledger/store....
[{"answers": ["src/probeledger/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-s1-00629
reorderflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bin keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/reorderflow/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 reorderflow la...
21,713
1,085
47
54
false
synth:1961169072815563246@1
synth
[{"aliases": ["src/reorderflow/limits.py", "limits.py"], "kind": "file", "value": "src/reorderflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/reorderflow/audit.py", "audit.py"], "kind": "file", "value": "src/reorderflow/audit....
[{"answers": ["src/reorderflow/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-s1-00629-twin
reorderflow
ctx_compress-s1-00629
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bin keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/reorderflow/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 reorderflow la...
11,033
1,000
29
54
false
synth:1961169072815563246@1
synth
[{"aliases": ["src/reorderflow/limits.py", "limits.py"], "kind": "file", "value": "src/reorderflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/reorderflow/audit.py", "audit.py"], "kind": "file", "value": "src/reorderflow/audit....
[{"answers": ["src/reorderflow/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-s1-00630
glyphflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_lexeme averages one value too many once the window is full. It lives in src/glyphflow/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 glyphflow layout. [tool_u...
87,567
3,072
110
119
false
synth:5272623175032475193@2.12495
synth
[{"aliases": ["src/glyphflow/report.py", "report.py"], "kind": "file", "value": "src/glyphflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/glyphflow/schema.py", "schema.py"], "kind": "file", "value": "src/glyphflow/schema.py"},...
[{"answers": ["src/glyphflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00630-twin
glyphflow
ctx_compress-s1-00630
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_lexeme averages one value too many once the window is full. It lives in src/glyphflow/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 glyphflow layout. [tool_u...
51,200
2,560
68
119
false
synth:5272623175032475193@2.12495
synth
[{"aliases": ["src/glyphflow/report.py", "report.py"], "kind": "file", "value": "src/glyphflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/glyphflow/schema.py", "schema.py"], "kind": "file", "value": "src/glyphflow/schema.py"},...
[{"answers": ["src/glyphflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00631
shipmentdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_sku raises BinRangeError for bytes input in strict mode instead of decoding it. It lives in src/shipmentdesk/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 shipm...
52,799
2,639
67
84
false
synth:2601847778007750603@1
synth
[{"aliases": ["src/shipmentdesk/core.py", "core.py"], "kind": "file", "value": "src/shipmentdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/shipmentdesk/tables.py", "tables.py"], "kind": "file", "value": "src/shipmentdesk/tables.py"},...
[{"answers": ["src/shipmentdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00631-twin
shipmentdesk
ctx_compress-s1-00631
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_sku raises BinRangeError for bytes input in strict mode instead of decoding it. It lives in src/shipmentdesk/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 shipm...
24,989
1,249
39
84
false
synth:2601847778007750603@1
synth
[{"aliases": ["src/shipmentdesk/core.py", "core.py"], "kind": "file", "value": "src/shipmentdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/shipmentdesk/tables.py", "tables.py"], "kind": "file", "value": "src/shipmentdesk/tables.py"},...
[{"answers": ["src/shipmentdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00632
seriesworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_sample_strict leaves a trailing dash when the title ends in punctuation. It lives in src/seriesworks/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 series...
70,340
3,072
92
103
false
synth:5920737968419482906@2.03385
synth
[{"aliases": ["src/seriesworks/audit.py", "audit.py"], "kind": "file", "value": "src/seriesworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/sample_0.csv", "sample_0.csv"], "kind": "file", "value": "data/sample_0.csv"}, {"aliases"...
[{"answers": ["src/seriesworks/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00632-twin
seriesworks
ctx_compress-s1-00632
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_sample_strict leaves a trailing dash when the title ends in punctuation. It lives in src/seriesworks/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 series...
44,321
2,216
51
103
false
synth:5920737968419482906@2.03385
synth
[{"aliases": ["src/seriesworks/audit.py", "audit.py"], "kind": "file", "value": "src/seriesworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/sample_0.csv", "sample_0.csv"], "kind": "file", "value": "data/sample_0.csv"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/seriesworks/audit.py"...
ctx_compress-s1-00633
principalflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_scope lets defaults overwrite explicit overrides. It lives in src/principalflow/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 principalflow layout. [tool_use kind=...
42,813
2,140
70
85
false
synth:8381137093665670203@1.30307
synth
[{"aliases": ["src/principalflow/io.py", "io.py"], "kind": "file", "value": "src/principalflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/principalflow/rules.py", "rules.py"], "kind": "file", "value": "src/principalflow/rules.py"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["core.py", "src/principalflow/core.py"...
ctx_compress-s1-00633-twin
principalflow
ctx_compress-s1-00633
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_scope lets defaults overwrite explicit overrides. It lives in src/principalflow/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 principalflow layout. [tool_use kind=...
22,370
1,118
39
85
false
synth:8381137093665670203@1.30307
synth
[{"aliases": ["src/principalflow/io.py", "io.py"], "kind": "file", "value": "src/principalflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/principalflow/rules.py", "rules.py"], "kind": "file", "value": "src/principalflow/rules.py"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_core.py", "tests/test_core.py"],...
ctx_compress-s1-00634
payoutworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_credit lets defaults overwrite explicit overrides. It lives in src/payoutworks/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 payoutworks layout. [tool_use kin...
79,514
3,072
99
116
false
synth:4819988588269645064@2.17013
synth
[{"aliases": ["src/payoutworks/limits.py", "limits.py"], "kind": "file", "value": "src/payoutworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/payout_0.csv", "payout_0.csv"], "kind": "file", "value": "data/payout_0.csv"}, {"al...
[{"answers": ["src/payoutworks/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-s1-00634-twin
payoutworks
ctx_compress-s1-00634
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_credit lets defaults overwrite explicit overrides. It lives in src/payoutworks/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 payoutworks layout. [tool_use kin...
44,517
2,225
57
116
false
synth:4819988588269645064@2.17013
synth
[{"aliases": ["src/payoutworks/limits.py", "limits.py"], "kind": "file", "value": "src/payoutworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/payout_0.csv", "payout_0.csv"], "kind": "file", "value": "data/payout_0.csv"}, {"al...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-00635
cadencedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_window_safe raises BookingBoundError for bytes input in strict mode instead of decoding it. It lives in src/cadencedesk/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 ma...
34,825
1,741
75
96
false
synth:2024467416457056925@1.3827
synth
[{"aliases": ["src/cadencedesk/audit.py", "audit.py"], "kind": "file", "value": "src/cadencedesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/cadencedesk/schema.py", "schema.py"], "kind": "file", "value": "src/cadencedesk/schema.py"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["core.py", "src/cadencedesk/core.py"],...
ctx_compress-s1-00635-twin
cadencedesk
ctx_compress-s1-00635
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_window_safe raises BookingBoundError for bytes input in strict mode instead of decoding it. It lives in src/cadencedesk/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 ma...
25,342
1,267
49
96
false
synth:2024467416457056925@1.3827
synth
[{"aliases": ["src/cadencedesk/audit.py", "audit.py"], "kind": "file", "value": "src/cadencedesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/cadencedesk/schema.py", "schema.py"], "kind": "file", "value": "src/cadencedesk/schema.py"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["schema.py", "src/cadencedesk/schema.p...
ctx_compress-s1-00636
revisionforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_revision misses keys that differ only in case. It lives in src/revisionforge/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 revisionforge layout. [tool_use ...
87,465
3,072
106
131
false
synth:4255077795256445049@1.84745
synth
[{"aliases": ["src/revisionforge/store.py", "store.py"], "kind": "file", "value": "src/revisionforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/margin_0.csv", "margin_0.csv"], "kind": "file", "value": "data/margin_0.csv"}, {"alia...
[{"answers": ["src/revisionforge/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-s1-00636-twin
revisionforge
ctx_compress-s1-00636
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_revision misses keys that differ only in case. It lives in src/revisionforge/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 revisionforge layout. [tool_use ...
47,066
2,353
62
131
false
synth:4255077795256445049@1.84745
synth
[{"aliases": ["src/revisionforge/store.py", "store.py"], "kind": "file", "value": "src/revisionforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/margin_0.csv", "margin_0.csv"], "kind": "file", "value": "data/margin_0.csv"}, {"alia...
[{"answers": ["src/revisionforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-00637
waypointflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_waypoint double counts carried amounts when the period closes on a weekend. It lives in src/waypointflow/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 waypoint...
71,733
3,072
95
118
false
synth:1164390288009651002@1.8898
synth
[{"aliases": ["src/waypointflow/store.py", "store.py"], "kind": "file", "value": "src/waypointflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/waypointflow/core.py", "core.py"], "kind": "file", "value": "src/waypointflow/core.py"},...
[{"answers": ["src/waypointflow/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-s1-00637-twin
waypointflow
ctx_compress-s1-00637
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_waypoint double counts carried amounts when the period closes on a weekend. It lives in src/waypointflow/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 waypoint...
44,612
2,230
63
118
false
synth:1164390288009651002@1.8898
synth
[{"aliases": ["src/waypointflow/store.py", "store.py"], "kind": "file", "value": "src/waypointflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/waypointflow/core.py", "core.py"], "kind": "file", "value": "src/waypointflow/core.py"},...
[{"answers": ["src/waypointflow/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-s1-00638
indexworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_index_page produces doubled dashes for consecutive separators. It lives in src/indexworks/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 indexworks layout. [t...
36,188
1,809
60
71
false
synth:3917271177467157989@1
synth
[{"aliases": ["src/indexworks/report.py", "report.py"], "kind": "file", "value": "src/indexworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/indexworks/limits.py", "limits.py"], "kind": "file", "value": "src/indexworks/limits.p...
[{"answers": ["src/indexworks/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-00638-twin
indexworks
ctx_compress-s1-00638
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_index_page produces doubled dashes for consecutive separators. It lives in src/indexworks/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 indexworks layout. [t...
19,969
1,000
34
71
false
synth:3917271177467157989@1
synth
[{"aliases": ["src/indexworks/report.py", "report.py"], "kind": "file", "value": "src/indexworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/indexworks/limits.py", "limits.py"], "kind": "file", "value": "src/indexworks/limits.p...
[{"answers": ["src/indexworks/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-00639
policyworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_principal double counts carried amounts when the period closes on a weekend. It lives in src/policyworks/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 policywo...
48,272
2,413
74
89
false
synth:579162397011775997@1.412
synth
[{"aliases": ["src/policyworks/core.py", "core.py"], "kind": "file", "value": "src/policyworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/policyworks/io.py", "io.py"], "kind": "file", "value": "src/policyworks/io.py"}, {"aliases": ["d...
[{"answers": ["src/policyworks/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00639-twin
policyworks
ctx_compress-s1-00639
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_principal double counts carried amounts when the period closes on a weekend. It lives in src/policyworks/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 policywo...
32,066
1,603
49
89
false
synth:579162397011775997@1.412
synth
[{"aliases": ["src/policyworks/core.py", "core.py"], "kind": "file", "value": "src/policyworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/policyworks/io.py", "io.py"], "kind": "file", "value": "src/policyworks/io.py"}, {"aliases": ["d...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-00640
bucketforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_page grows linearly instead of doubling. It lives in src/bucketforge/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 bucketforge layout. [tool_use kind...
76,737
3,072
91
116
false
synth:7287543781282495709@2.22052
synth
[{"aliases": ["src/bucketforge/report.py", "report.py"], "kind": "file", "value": "src/bucketforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/bucketforge/io.py", "io.py"], "kind": "file", "value": "src/bucketforge/io.py"}, {"a...
[{"answers": ["src/bucketforge/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-s1-00640-twin
bucketforge
ctx_compress-s1-00640
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_page grows linearly instead of doubling. It lives in src/bucketforge/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 bucketforge layout. [tool_use kind...
49,066
2,453
63
116
false
synth:7287543781282495709@2.22052
synth
[{"aliases": ["src/bucketforge/report.py", "report.py"], "kind": "file", "value": "src/bucketforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/bucketforge/io.py", "io.py"], "kind": "file", "value": "src/bucketforge/io.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/trace_3.csv", "trace_3.csv"], "k...
ctx_compress-s1-00641
beaconledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon returns the upper bound for every value instead of clamping into the range. It lives in src/beaconledger/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 t...
54,181
2,709
89
112
false
synth:8726023941951416950@1.86993
synth
[{"aliases": ["src/beaconledger/audit.py", "audit.py"], "kind": "file", "value": "src/beaconledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/fix_5.csv", "fix_5.csv"], "kind": "file", "value": "data/fix_5.csv"}, {"aliases": ["src...
[{"answers": ["src/beaconledger/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-s1-00641-twin
beaconledger
ctx_compress-s1-00641
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_beacon returns the upper bound for every value instead of clamping into the range. It lives in src/beaconledger/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 t...
36,600
1,830
53
112
false
synth:8726023941951416950@1.86993
synth
[{"aliases": ["src/beaconledger/audit.py", "audit.py"], "kind": "file", "value": "src/beaconledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/fix_5.csv", "fix_5.csv"], "kind": "file", "value": "data/fix_5.csv"}, {"aliases": ["src...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_compat.py", "tests/test_compat.p...
ctx_compress-s1-00642
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_waypoint leaves a trailing dash when the title ends in punctuation. It lives in src/bearingflow/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 bearingflow ...
45,099
2,254
70
87
false
synth:166371721381650906@1
synth
[{"aliases": ["src/bearingflow/report.py", "report.py"], "kind": "file", "value": "src/bearingflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/fix_1.csv", "fix_1.csv"], "kind": "file", "value": "data/fix_1.csv"}, {"aliases": [...
[{"answers": ["src/bearingflow/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-s1-00642-twin
bearingflow
ctx_compress-s1-00642
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_waypoint leaves a trailing dash when the title ends in punctuation. It lives in src/bearingflow/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 bearingflow ...
28,171
1,408
46
87
false
synth:166371721381650906@1
synth
[{"aliases": ["src/bearingflow/report.py", "report.py"], "kind": "file", "value": "src/bearingflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/fix_1.csv", "fix_1.csv"], "kind": "file", "value": "data/fix_1.csv"}, {"aliases": [...
[{"answers": ["src/bearingflow/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-s1-00643
bookingkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_roster skips the first page; page 1 returns what should be page 2. It lives in src/bookingkit/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 bookingkit layo...
59,742
2,987
112
129
false
synth:5133140366388010208@1.41092
synth
[{"aliases": ["src/bookingkit/limits.py", "limits.py"], "kind": "file", "value": "src/bookingkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/bookingkit/rules.py", "rules.py"], "kind": "file", "value": "src/bookingkit/rules.py"}...
[{"answers": ["src/bookingkit/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-s1-00643-twin
bookingkit
ctx_compress-s1-00643
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_roster skips the first page; page 1 returns what should be page 2. It lives in src/bookingkit/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 bookingkit layo...
29,468
1,473
66
129
false
synth:5133140366388010208@1.41092
synth
[{"aliases": ["src/bookingkit/limits.py", "limits.py"], "kind": "file", "value": "src/bookingkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/bookingkit/rules.py", "rules.py"], "kind": "file", "value": "src/bookingkit/rules.py"}...
[{"answers": ["src/bookingkit/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-s1-00644
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] count_warehouse_once double counts carried amounts when the period closes on a weekend. It lives in src/shipmentkit/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 sh...
62,740
3,072
90
111
false
synth:5970656183968085884@1.37062
synth
[{"aliases": ["src/shipmentkit/store.py", "store.py"], "kind": "file", "value": "src/shipmentkit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/shipmentkit/compat.py", "compat.py"], "kind": "file", "value": "src/shipmentkit/compat.py"...
[{"answers": ["src/shipmentkit/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00644-twin
shipmentkit
ctx_compress-s1-00644
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse_once double counts carried amounts when the period closes on a weekend. It lives in src/shipmentkit/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 sh...
35,728
1,786
62
111
false
synth:5970656183968085884@1.37062
synth
[{"aliases": ["src/shipmentkit/store.py", "store.py"], "kind": "file", "value": "src/shipmentkit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/shipmentkit/compat.py", "compat.py"], "kind": "file", "value": "src/shipmentkit/compat.py"...
[{"answers": ["src/shipmentkit/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00645
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_leg raises ValueError when a value itself contains an equals sign. It lives in src/bearingflow/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 bearingflow l...
27,413
1,370
45
52
false
synth:3511784158643257146@1
synth
[{"aliases": ["src/bearingflow/format.py", "format.py"], "kind": "file", "value": "src/bearingflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/beacon_2.csv", "beacon_2.csv"], "kind": "file", "value": "data/beacon_2.csv"}, {"al...
[{"answers": ["src/bearingflow/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-s1-00645-twin
bearingflow
ctx_compress-s1-00645
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_leg raises ValueError when a value itself contains an equals sign. It lives in src/bearingflow/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 bearingflow l...
15,160
1,000
25
52
false
synth:3511784158643257146@1
synth
[{"aliases": ["src/bearingflow/format.py", "format.py"], "kind": "file", "value": "src/bearingflow/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/beacon_2.csv", "beacon_2.csv"], "kind": "file", "value": "data/beacon_2.csv"}, {"al...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/bearingflow/format.p...
ctx_compress-s1-00646
marginledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_footnote drops override values of 0, False and empty string. It lives in src/marginledger/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 marginledger layout. [to...
89,537
3,072
112
123
false
synth:4972092716549724796@1.70843
synth
[{"aliases": ["src/marginledger/cli.py", "cli.py"], "kind": "file", "value": "src/marginledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/marginledger/store.py", "store.py"], "kind": "file", "value": "src/marginledger/store.py"}, {"aliase...
[{"answers": ["src/marginledger/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-s1-00646-twin
marginledger
ctx_compress-s1-00646
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_footnote drops override values of 0, False and empty string. It lives in src/marginledger/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 marginledger layout. [to...
46,223
2,311
66
123
false
synth:4972092716549724796@1.70843
synth
[{"aliases": ["src/marginledger/cli.py", "cli.py"], "kind": "file", "value": "src/marginledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/marginledger/store.py", "store.py"], "kind": "file", "value": "src/marginledger/store.py"}, {"aliase...
[{"answers": ["src/marginledger/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00647
slotworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] split_booking lets defaults overwrite explicit overrides. It lives in src/slotworks/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 slotworks layout. [tool_use kind...
65,469
3,072
91
114
false
synth:5148653259097962885@2.2747
synth
[{"aliases": ["src/slotworks/report.py", "report.py"], "kind": "file", "value": "src/slotworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/slotworks/schema.py", "schema.py"], "kind": "file", "value": "src/slotworks/schema.py"},...
[{"answers": ["src/slotworks/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00647-twin
slotworks
ctx_compress-s1-00647
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] split_booking lets defaults overwrite explicit overrides. It lives in src/slotworks/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 slotworks layout. [tool_use kind...
43,810
2,190
63
114
false
synth:5148653259097962885@2.2747
synth
[{"aliases": ["src/slotworks/report.py", "report.py"], "kind": "file", "value": "src/slotworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/slotworks/schema.py", "schema.py"], "kind": "file", "value": "src/slotworks/schema.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_limits.py", "tests/test_limits.p...
ctx_compress-s1-00648
tariffworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_tariff_strict double counts carried amounts when the period closes on a weekend. It lives in src/tariffworks/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 ta...
64,121
3,072
94
115
false
synth:8400658898012035076@1.65173
synth
[{"aliases": ["src/tariffworks/report.py", "report.py"], "kind": "file", "value": "src/tariffworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/refund_1.csv", "refund_1.csv"], "kind": "file", "value": "data/refund_1.csv"}, {"al...
[{"answers": ["src/tariffworks/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-s1-00648-twin
tariffworks
ctx_compress-s1-00648
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_tariff_strict double counts carried amounts when the period closes on a weekend. It lives in src/tariffworks/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 ta...
37,406
1,870
60
115
false
synth:8400658898012035076@1.65173
synth
[{"aliases": ["src/tariffworks/report.py", "report.py"], "kind": "file", "value": "src/tariffworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/refund_1.csv", "refund_1.csv"], "kind": "file", "value": "data/refund_1.csv"}, {"al...
[{"answers": ["src/tariffworks/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-s1-00649
zoneworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_tile raises ValueError when a value itself contains an equals sign. It lives in src/zoneworks/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 zoneworks la...
84,532
3,072
94
103
false
synth:5059868441191009312@1.96418
synth
[{"aliases": ["src/zoneworks/rules.py", "rules.py"], "kind": "file", "value": "src/zoneworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/zoneworks/schema.py", "schema.py"], "kind": "file", "value": "src/zoneworks/schema.py"}, {"ali...
[{"answers": ["src/zoneworks/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-s1-00649-twin
zoneworks
ctx_compress-s1-00649
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_tile raises ValueError when a value itself contains an equals sign. It lives in src/zoneworks/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 zoneworks la...
46,977
2,348
47
103
false
synth:5059868441191009312@1.96418
synth
[{"aliases": ["src/zoneworks/rules.py", "rules.py"], "kind": "file", "value": "src/zoneworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/zoneworks/schema.py", "schema.py"], "kind": "file", "value": "src/zoneworks/schema.py"}, {"ali...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["rules.py", "src/zoneworks/rules.py"],...
ctx_compress-s1-00650
tokenflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] mint_grant_strict returns one item too few on every page. It lives in src/tokenflow/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 tokenflow layout. [tool_use kind=s...
19,963
1,000
50
55
false
synth:1355902226676026002@1
synth
[{"aliases": ["src/tokenflow/rules.py", "rules.py"], "kind": "file", "value": "src/tokenflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/tokenflow/limits.py", "limits.py"], "kind": "file", "value": "src/tokenflow/limits.py"}, {"ali...
[{"answers": ["src/tokenflow/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-00650-twin
tokenflow
ctx_compress-s1-00650
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_strict returns one item too few on every page. It lives in src/tokenflow/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 tokenflow layout. [tool_use kind=s...
14,868
1,000
32
55
false
synth:1355902226676026002@1
synth
[{"aliases": ["src/tokenflow/rules.py", "rules.py"], "kind": "file", "value": "src/tokenflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/tokenflow/limits.py", "limits.py"], "kind": "file", "value": "src/tokenflow/limits.py"}, {"ali...
[{"answers": ["src/tokenflow/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-00651
rosterkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_booking leaves a trailing dash when the title ends in punctuation. It lives in src/rosterkit/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 rosterkit layout. [too...
26,020
1,301
64
71
false
synth:5521093201928455876@1
synth
[{"aliases": ["src/rosterkit/cli.py", "cli.py"], "kind": "file", "value": "src/rosterkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/rosterkit/schema.py", "schema.py"], "kind": "file", "value": "src/rosterkit/schema.py"}, {"aliases": ["src...
[{"answers": ["src/rosterkit/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rec...
ctx_compress-s1-00651-twin
rosterkit
ctx_compress-s1-00651
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_booking leaves a trailing dash when the title ends in punctuation. It lives in src/rosterkit/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 rosterkit layout. [too...
14,752
1,000
38
71
false
synth:5521093201928455876@1
synth
[{"aliases": ["src/rosterkit/cli.py", "cli.py"], "kind": "file", "value": "src/rosterkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/rosterkit/schema.py", "schema.py"], "kind": "file", "value": "src/rosterkit/schema.py"}, {"aliases": ["src...
[{"answers": ["src/rosterkit/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rec...
ctx_compress-s1-00652
traceflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_probe_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/traceflow/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 tracef...
39,690
1,984
78
93
false
synth:9116539271272256971@1
synth
[{"aliases": ["src/traceflow/audit.py", "audit.py"], "kind": "file", "value": "src/traceflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/traceflow/tables.py", "tables.py"], "kind": "file", "value": "src/traceflow/tables.py"}, {"ali...
[{"answers": ["src/traceflow/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-s1-00652-twin
traceflow
ctx_compress-s1-00652
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe_all raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/traceflow/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 tracef...
20,278
1,013
44
93
false
synth:9116539271272256971@1
synth
[{"aliases": ["src/traceflow/audit.py", "audit.py"], "kind": "file", "value": "src/traceflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/traceflow/tables.py", "tables.py"], "kind": "file", "value": "src/traceflow/tables.py"}, {"ali...
[{"answers": ["src/traceflow/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-s1-00653
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] roll_payout returns the upper bound for every value instead of clamping into the range. It lives in src/receiptdesk/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 re...
92,958
3,072
112
131
false
synth:8607645706260390739@1.89197
synth
[{"aliases": ["src/receiptdesk/rules.py", "rules.py"], "kind": "file", "value": "src/receiptdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/receiptdesk/report.py", "report.py"], "kind": "file", "value": "src/receiptdesk/report.py"...
[{"answers": ["src/receiptdesk/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-s1-00653-twin
receiptdesk
ctx_compress-s1-00653
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_payout returns the upper bound for every value instead of clamping into the range. It lives in src/receiptdesk/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 re...
56,094
2,804
76
131
false
synth:8607645706260390739@1.89197
synth
[{"aliases": ["src/receiptdesk/rules.py", "rules.py"], "kind": "file", "value": "src/receiptdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/receiptdesk/report.py", "report.py"], "kind": "file", "value": "src/receiptdesk/report.py"...
[{"answers": ["src/receiptdesk/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-s1-00654
balanceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_ledger_safe misses keys that differ only in case. It lives in src/balanceforge/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 balanceforge layout. [tool_use...
35,277
1,763
69
84
false
synth:1438805945102053188@1
synth
[{"aliases": ["src/balanceforge/format.py", "format.py"], "kind": "file", "value": "src/balanceforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"...
[{"answers": ["src/balanceforge/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-s1-00654-twin
balanceforge
ctx_compress-s1-00654
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_ledger_safe misses keys that differ only in case. It lives in src/balanceforge/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 balanceforge layout. [tool_use...
17,768
1,000
39
84
false
synth:1438805945102053188@1
synth
[{"aliases": ["src/balanceforge/format.py", "format.py"], "kind": "file", "value": "src/balanceforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"...
[{"answers": ["src/balanceforge/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...