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-s1-00705
draftworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_margin_window skips the first page; page 1 returns what should be page 2. It lives in src/draftworks/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 draftwor...
112,469
3,072
132
153
false
synth:4249654783418792648@2.14398
synth
[{"aliases": ["src/draftworks/limits.py", "limits.py"], "kind": "file", "value": "src/draftworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"alia...
[{"answers": ["src/draftworks/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-s1-00705-twin
draftworks
ctx_compress-s1-00705
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_margin_window skips the first page; page 1 returns what should be page 2. It lives in src/draftworks/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 draftwor...
62,781
3,072
82
153
false
synth:4249654783418792648@2.14398
synth
[{"aliases": ["src/draftworks/limits.py", "limits.py"], "kind": "file", "value": "src/draftworks/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"alia...
[{"answers": ["src/draftworks/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-s1-00706
quotadesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_calendar_safe lets defaults overwrite explicit overrides. It lives in src/quotadesk/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 quotadesk layout. [tool_...
32,719
1,635
59
70
false
synth:7619598349630408647@1
synth
[{"aliases": ["src/quotadesk/report.py", "report.py"], "kind": "file", "value": "src/quotadesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/quotadesk/core.py", "core.py"], "kind": "file", "value": "src/quotadesk/core.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": ["report.py", "src/quotadesk/report.py"...
ctx_compress-s1-00706-twin
quotadesk
ctx_compress-s1-00706
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_calendar_safe lets defaults overwrite explicit overrides. It lives in src/quotadesk/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 quotadesk layout. [tool_...
13,704
1,000
37
70
false
synth:7619598349630408647@1
synth
[{"aliases": ["src/quotadesk/report.py", "report.py"], "kind": "file", "value": "src/quotadesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/quotadesk/core.py", "core.py"], "kind": "file", "value": "src/quotadesk/core.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": ["schema.py", "src/quotadesk/schema.py"...
ctx_compress-s1-00707
sampleworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] smooth_counter_strict skips the first page; page 1 returns what should be page 2. It lives in src/sampleworks/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the samplewo...
43,857
2,192
64
75
false
synth:813593184060813597@1
synth
[{"aliases": ["src/sampleworks/store.py", "store.py"], "kind": "file", "value": "src/sampleworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/sampleworks/io.py", "io.py"], "kind": "file", "value": "src/sampleworks/io.py"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/sampleworks/store.py", "store.py"...
ctx_compress-s1-00707-twin
sampleworks
ctx_compress-s1-00707
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] smooth_counter_strict skips the first page; page 1 returns what should be page 2. It lives in src/sampleworks/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the samplewo...
24,543
1,227
40
75
false
synth:813593184060813597@1
synth
[{"aliases": ["src/sampleworks/store.py", "store.py"], "kind": "file", "value": "src/sampleworks/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/sampleworks/io.py", "io.py"], "kind": "file", "value": "src/sampleworks/io.py"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["bucket_0.csv", "data/bucket_0.csv"], ...
ctx_compress-s1-00708
cadenceledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_cadence grows linearly instead of doubling. It lives in src/cadenceledger/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 cadenceledger layout. [tool_use kind=...
61,099
3,054
96
115
false
synth:3187256531609678870@1.3465
synth
[{"aliases": ["src/cadenceledger/rules.py", "rules.py"], "kind": "file", "value": "src/cadenceledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/cadenceledger/audit.py", "audit.py"], "kind": "file", "value": "src/cadenceledger/audi...
[{"answers": ["src/cadenceledger/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-00708-twin
cadenceledger
ctx_compress-s1-00708
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_cadence grows linearly instead of doubling. It lives in src/cadenceledger/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 cadenceledger layout. [tool_use kind=...
36,704
1,835
57
115
false
synth:3187256531609678870@1.3465
synth
[{"aliases": ["src/cadenceledger/rules.py", "rules.py"], "kind": "file", "value": "src/cadenceledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/cadenceledger/audit.py", "audit.py"], "kind": "file", "value": "src/cadenceledger/audi...
[{"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/cadenceledger/rules.p...
ctx_compress-s1-00709
beaconkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 skips the first page; page 1 returns what should be page 2. It lives in src/beaconkit/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 beaconkit layout. [too...
56,022
2,801
84
107
false
synth:6885148102757817207@1.9548
synth
[{"aliases": ["src/beaconkit/io.py", "io.py"], "kind": "file", "value": "src/beaconkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/beaconkit/report.py", "report.py"], "kind": "file", "value": "src/beaconkit/report.py"}, {"aliases": ["data/leg_...
[{"answers": ["src/beaconkit/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-00709-twin
beaconkit
ctx_compress-s1-00709
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 skips the first page; page 1 returns what should be page 2. It lives in src/beaconkit/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 beaconkit layout. [too...
39,349
1,967
60
107
false
synth:6885148102757817207@1.9548
synth
[{"aliases": ["src/beaconkit/io.py", "io.py"], "kind": "file", "value": "src/beaconkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/beaconkit/report.py", "report.py"], "kind": "file", "value": "src/beaconkit/report.py"}, {"aliases": ["data/leg_...
[{"answers": ["src/beaconkit/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-00710
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] reconcile_receipt averages one value too many once the window is full. It lives in src/creditworks/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 creditworks layout....
59,819
2,990
86
95
false
synth:7551095619278189064@1
synth
[{"aliases": ["src/creditworks/audit.py", "audit.py"], "kind": "file", "value": "src/creditworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/creditworks/limits.py", "limits.py"], "kind": "file", "value": "src/creditworks/limits.py"...
[{"answers": ["src/creditworks/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00710-twin
creditworks
ctx_compress-s1-00710
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_receipt averages one value too many once the window is full. It lives in src/creditworks/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 creditworks layout....
26,300
1,315
44
95
false
synth:7551095619278189064@1
synth
[{"aliases": ["src/creditworks/audit.py", "audit.py"], "kind": "file", "value": "src/creditworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/creditworks/limits.py", "limits.py"], "kind": "file", "value": "src/creditworks/limits.py"...
[{"answers": ["src/creditworks/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-s1-00711
quotadesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_booking_window ignores the cap, so late attempts wait for minutes. It lives in src/quotadesk/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 quotadesk layou...
44,523
2,226
79
90
false
synth:8098482102736423497@1
synth
[{"aliases": ["src/quotadesk/report.py", "report.py"], "kind": "file", "value": "src/quotadesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/window_1.csv", "window_1.csv"], "kind": "file", "value": "data/window_1.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["report.py", "src/quotadesk/report.py"...
ctx_compress-s1-00711-twin
quotadesk
ctx_compress-s1-00711
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_booking_window ignores the cap, so late attempts wait for minutes. It lives in src/quotadesk/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 quotadesk layou...
22,557
1,127
43
90
false
synth:8098482102736423497@1
synth
[{"aliases": ["src/quotadesk/report.py", "report.py"], "kind": "file", "value": "src/quotadesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/window_1.csv", "window_1.csv"], "kind": "file", "value": "data/window_1.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/window_1.csv", "window_1.csv"], ...
ctx_compress-s1-00712
shiftdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window rejects a range where lo equals hi, which is a valid single point range. It lives in src/shiftdesk/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 shi...
24,068
1,203
56
61
false
synth:8762254870748463856@1
synth
[{"aliases": ["src/shiftdesk/io.py", "io.py"], "kind": "file", "value": "src/shiftdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/shiftdesk/compat.py", "compat.py"], "kind": "file", "value": "src/shiftdesk/compat.py"}, {"aliases": ["src/shift...
[{"answers": ["src/shiftdesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-00712-twin
shiftdesk
ctx_compress-s1-00712
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window rejects a range where lo equals hi, which is a valid single point range. It lives in src/shiftdesk/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 shi...
14,855
1,000
36
61
false
synth:8762254870748463856@1
synth
[{"aliases": ["src/shiftdesk/io.py", "io.py"], "kind": "file", "value": "src/shiftdesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/shiftdesk/compat.py", "compat.py"], "kind": "file", "value": "src/shiftdesk/compat.py"}, {"aliases": ["src/shift...
[{"answers": ["src/shiftdesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-00713
slugforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_chunk raises StemStateError for bytes input in strict mode instead of decoding it. It lives in src/slugforge/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 slugforg...
91,768
3,072
115
134
false
synth:8701190038997720848@1.6924
synth
[{"aliases": ["src/slugforge/io.py", "io.py"], "kind": "file", "value": "src/slugforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/caption_8.csv", "caption_8.csv"], "kind": "file", "value": "data/caption_8.csv"}, {"aliases": ["src/slugforge/r...
[{"answers": ["src/slugforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-00713-twin
slugforge
ctx_compress-s1-00713
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_chunk raises StemStateError for bytes input in strict mode instead of decoding it. It lives in src/slugforge/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 slugforg...
42,211
2,110
63
134
false
synth:8701190038997720848@1.6924
synth
[{"aliases": ["src/slugforge/io.py", "io.py"], "kind": "file", "value": "src/slugforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/caption_8.csv", "caption_8.csv"], "kind": "file", "value": "data/caption_8.csv"}, {"aliases": ["src/slugforge/r...
[{"answers": ["src/slugforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-00714
creditflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_ledger double counts carried amounts when the period closes on a weekend. It lives in src/creditflow/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the creditflow l...
57,257
2,862
80
91
false
synth:3701880448443897263@1
synth
[{"aliases": ["src/creditflow/core.py", "core.py"], "kind": "file", "value": "src/creditflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/balance_2.csv", "balance_2.csv"], "kind": "file", "value": "data/balance_2.csv"}, {"aliases": ["d...
[{"answers": ["src/creditflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00714-twin
creditflow
ctx_compress-s1-00714
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_ledger double counts carried amounts when the period closes on a weekend. It lives in src/creditflow/core.py; `python -m pytest tests/test_core.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the creditflow l...
29,067
1,453
42
91
false
synth:3701880448443897263@1
synth
[{"aliases": ["src/creditflow/core.py", "core.py"], "kind": "file", "value": "src/creditflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/balance_2.csv", "balance_2.csv"], "kind": "file", "value": "data/balance_2.csv"}, {"aliases": ["d...
[{"answers": ["src/creditflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00715
slugkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_chunk_all double counts carried amounts when the period closes on a weekend. It lives in src/slugkit/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 slugkit layo...
70,097
3,072
96
105
false
synth:7058921339674867132@1.50432
synth
[{"aliases": ["src/slugkit/audit.py", "audit.py"], "kind": "file", "value": "src/slugkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/slugkit/tables.py", "tables.py"], "kind": "file", "value": "src/slugkit/tables.py"}, {"aliases": [...
[{"answers": ["src/slugkit/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-00715-twin
slugkit
ctx_compress-s1-00715
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_chunk_all double counts carried amounts when the period closes on a weekend. It lives in src/slugkit/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 slugkit layo...
32,370
1,618
54
105
false
synth:7058921339674867132@1.50432
synth
[{"aliases": ["src/slugkit/audit.py", "audit.py"], "kind": "file", "value": "src/slugkit/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/slugkit/tables.py", "tables.py"], "kind": "file", "value": "src/slugkit/tables.py"}, {"aliases": [...
[{"answers": ["src/slugkit/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-00716
cueforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_cue leaves a trailing dash when the title ends in punctuation. It lives in src/cueforge/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 cueforge layout. [t...
94,996
3,072
102
113
false
synth:3214171097248932237@1.61823
synth
[{"aliases": ["src/cueforge/report.py", "report.py"], "kind": "file", "value": "src/cueforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/cue_12.csv", "cue_12.csv"], "kind": "file", "value": "data/cue_12.csv"}, {"aliases": ["sr...
[{"answers": ["src/cueforge/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 mos...
ctx_compress-s1-00716-twin
cueforge
ctx_compress-s1-00716
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_cue leaves a trailing dash when the title ends in punctuation. It lives in src/cueforge/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 cueforge layout. [t...
43,330
2,166
56
113
false
synth:3214171097248932237@1.61823
synth
[{"aliases": ["src/cueforge/report.py", "report.py"], "kind": "file", "value": "src/cueforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/cue_12.csv", "cue_12.csv"], "kind": "file", "value": "data/cue_12.csv"}, {"aliases": ["sr...
[{"answers": ["src/cueforge/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 mos...
ctx_compress-s1-00717
draftledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_index raises IndexDriftError for bytes input in strict mode instead of decoding it. It lives in src/draftledger/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...
34,293
1,714
60
67
false
synth:3639256744827512465@1
synth
[{"aliases": ["src/draftledger/limits.py", "limits.py"], "kind": "file", "value": "src/draftledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/clause_2.csv", "clause_2.csv"], "kind": "file", "value": "data/clause_2.csv"}, {"al...
[{"answers": ["src/draftledger/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-s1-00717-twin
draftledger
ctx_compress-s1-00717
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_index raises IndexDriftError for bytes input in strict mode instead of decoding it. It lives in src/draftledger/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...
15,530
1,000
34
67
false
synth:3639256744827512465@1
synth
[{"aliases": ["src/draftledger/limits.py", "limits.py"], "kind": "file", "value": "src/draftledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/clause_2.csv", "clause_2.csv"], "kind": "file", "value": "data/clause_2.csv"}, {"al...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/footnote_3.csv", "footnote_3.csv...
ctx_compress-s1-00718
sessionforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim grows linearly instead of doubling. It lives in src/sessionforge/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 sessionforge layout. [tool_use kind=search nam...
61,884
3,072
87
108
false
synth:7662836585841585331@2.07025
synth
[{"aliases": ["src/sessionforge/io.py", "io.py"], "kind": "file", "value": "src/sessionforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/token_0.csv", "token_0.csv"], "kind": "file", "value": "data/token_0.csv"}, {"aliases": ["data/session_5....
[{"answers": ["src/sessionforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most re...
ctx_compress-s1-00718-twin
sessionforge
ctx_compress-s1-00718
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim grows linearly instead of doubling. It lives in src/sessionforge/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 sessionforge layout. [tool_use kind=search nam...
42,196
2,109
55
108
false
synth:7662836585841585331@2.07025
synth
[{"aliases": ["src/sessionforge/io.py", "io.py"], "kind": "file", "value": "src/sessionforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/token_0.csv", "token_0.csv"], "kind": "file", "value": "data/token_0.csv"}, {"aliases": ["data/session_5....
[{"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/sessionforge/core.py"]...
ctx_compress-s1-00719
fixledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bearing averages one value too many once the window is full. It lives in src/fixledger/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 fixledger layout. [to...
78,245
3,072
106
123
false
synth:7751275704686605018@2.26658
synth
[{"aliases": ["src/fixledger/compat.py", "compat.py"], "kind": "file", "value": "src/fixledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/fixledger/limits.py", "limits.py"], "kind": "file", "value": "src/fixledger/limits.py"},...
[{"answers": ["src/fixledger/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00719-twin
fixledger
ctx_compress-s1-00719
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bearing averages one value too many once the window is full. It lives in src/fixledger/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 fixledger layout. [to...
54,062
2,703
67
123
false
synth:7751275704686605018@2.26658
synth
[{"aliases": ["src/fixledger/compat.py", "compat.py"], "kind": "file", "value": "src/fixledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/fixledger/limits.py", "limits.py"], "kind": "file", "value": "src/fixledger/limits.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/fixledger/compat.py"...
ctx_compress-s1-00720
cueledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_chunk skips the first page; page 1 returns what should be page 2. It lives in src/cueledger/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 cueledger layout. [...
38,006
1,900
89
106
false
synth:1507605139636546059@1
synth
[{"aliases": ["src/cueledger/store.py", "store.py"], "kind": "file", "value": "src/cueledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/cueledger/format.py", "format.py"], "kind": "file", "value": "src/cueledger/format.py"}, {"ali...
[{"answers": ["src/cueledger/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00720-twin
cueledger
ctx_compress-s1-00720
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_chunk skips the first page; page 1 returns what should be page 2. It lives in src/cueledger/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 cueledger layout. [...
27,417
1,370
63
106
false
synth:1507605139636546059@1
synth
[{"aliases": ["src/cueledger/store.py", "store.py"], "kind": "file", "value": "src/cueledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/cueledger/format.py", "format.py"], "kind": "file", "value": "src/cueledger/format.py"}, {"ali...
[{"answers": ["src/cueledger/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00721
sessionkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_nonce rejects a range where lo equals hi, which is a valid single point range. It lives in src/sessionkit/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 sessi...
53,249
2,662
95
102
false
synth:7192405900402396880@1.00547
synth
[{"aliases": ["src/sessionkit/rules.py", "rules.py"], "kind": "file", "value": "src/sessionkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ...
[{"answers": ["src/sessionkit/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-s1-00721-twin
sessionkit
ctx_compress-s1-00721
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_nonce rejects a range where lo equals hi, which is a valid single point range. It lives in src/sessionkit/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 sessi...
30,584
1,529
57
102
false
synth:7192405900402396880@1.00547
synth
[{"aliases": ["src/sessionkit/rules.py", "rules.py"], "kind": "file", "value": "src/sessionkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ...
[{"answers": ["src/sessionkit/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-s1-00722
probeflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] smooth_gauge raises KeyError for a missing key instead of returning the default. It lives in src/probeflow/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 probeflow lay...
26,301
1,315
54
59
false
synth:239283708012312364@1
synth
[{"aliases": ["src/probeflow/core.py", "core.py"], "kind": "file", "value": "src/probeflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/probeflow/format.py", "format.py"], "kind": "file", "value": "src/probeflow/format.py"}, {"aliases":...
[{"answers": ["src/probeflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-00722-twin
probeflow
ctx_compress-s1-00722
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] smooth_gauge raises KeyError for a missing key instead of returning the default. It lives in src/probeflow/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 probeflow lay...
16,009
1,000
30
59
false
synth:239283708012312364@1
synth
[{"aliases": ["src/probeflow/core.py", "core.py"], "kind": "file", "value": "src/probeflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/probeflow/format.py", "format.py"], "kind": "file", "value": "src/probeflow/format.py"}, {"aliases":...
[{"answers": ["src/probeflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-00723
grantforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_policy returns the upper bound for every value instead of clamping into the range. It lives in src/grantforge/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 grant...
118,103
3,072
124
143
false
synth:4513562064120720914@2.01457
synth
[{"aliases": ["src/grantforge/cli.py", "cli.py"], "kind": "file", "value": "src/grantforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/claim_8.csv", "claim_8.csv"], "kind": "file", "value": "data/claim_8.csv"}, {"aliases": ["data/scope_0....
[{"answers": ["src/grantforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00723-twin
grantforge
ctx_compress-s1-00723
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_policy returns the upper bound for every value instead of clamping into the range. It lives in src/grantforge/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 grant...
72,524
3,072
80
143
false
synth:4513562064120720914@2.01457
synth
[{"aliases": ["src/grantforge/cli.py", "cli.py"], "kind": "file", "value": "src/grantforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/claim_8.csv", "claim_8.csv"], "kind": "file", "value": "data/claim_8.csv"}, {"aliases": ["data/scope_0....
[{"answers": ["src/grantforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00724
payoutforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_balance raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/payoutforge/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 p...
77,746
3,072
100
111
false
synth:1564093940861409877@1.55105
synth
[{"aliases": ["src/payoutforge/report.py", "report.py"], "kind": "file", "value": "src/payoutforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/payout_2.csv", "payout_2.csv"], "kind": "file", "value": "data/payout_2.csv"}, {"al...
[{"answers": ["src/payoutforge/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-s1-00724-twin
payoutforge
ctx_compress-s1-00724
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_balance raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/payoutforge/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 p...
48,606
2,430
66
111
false
synth:1564093940861409877@1.55105
synth
[{"aliases": ["src/payoutforge/report.py", "report.py"], "kind": "file", "value": "src/payoutforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/payout_2.csv", "payout_2.csv"], "kind": "file", "value": "data/payout_2.csv"}, {"al...
[{"answers": ["src/payoutforge/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-s1-00725
calendardesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster leaves a trailing dash when the title ends in punctuation. It lives in src/calendardesk/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 calendardesk ...
76,100
3,072
106
131
false
synth:4443827476056330572@2.16087
synth
[{"aliases": ["src/calendardesk/report.py", "report.py"], "kind": "file", "value": "src/calendardesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/calendardesk/format.py", "format.py"], "kind": "file", "value": "src/calendardesk/...
[{"answers": ["src/calendardesk/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-s1-00725-twin
calendardesk
ctx_compress-s1-00725
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_roster leaves a trailing dash when the title ends in punctuation. It lives in src/calendardesk/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 calendardesk ...
58,680
2,934
78
131
false
synth:4443827476056330572@2.16087
synth
[{"aliases": ["src/calendardesk/report.py", "report.py"], "kind": "file", "value": "src/calendardesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/calendardesk/format.py", "format.py"], "kind": "file", "value": "src/calendardesk/...
[{"answers": ["src/calendardesk/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-s1-00726
legkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_zone raises KeyError for a missing key instead of returning the default. It lives in src/legkit/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 legkit layout....
19,900
1,000
53
58
false
synth:7376221738519149284@1
synth
[{"aliases": ["src/legkit/store.py", "store.py"], "kind": "file", "value": "src/legkit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/legkit/tables.py", "tables.py"], "kind": "file", "value": "src/legkit/tables.py"}, {"aliases": ["dat...
[{"answers": ["src/legkit/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most re...
ctx_compress-s1-00726-twin
legkit
ctx_compress-s1-00726
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_zone raises KeyError for a missing key instead of returning the default. It lives in src/legkit/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 legkit layout....
13,144
1,000
31
58
false
synth:7376221738519149284@1
synth
[{"aliases": ["src/legkit/store.py", "store.py"], "kind": "file", "value": "src/legkit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/legkit/tables.py", "tables.py"], "kind": "file", "value": "src/legkit/tables.py"}, {"aliases": ["dat...
[{"answers": ["src/legkit/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00727
receiptforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_receipt_window raises KeyError for a missing key instead of returning the default. It lives in src/receiptforge/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 receiptf...
59,298
2,964
87
104
false
synth:2791054778109604221@1.61325
synth
[{"aliases": ["src/receiptforge/io.py", "io.py"], "kind": "file", "value": "src/receiptforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/payout_3.csv", "payout_3.csv"], "kind": "file", "value": "data/payout_3.csv"}, {"aliases": ["data/invoice...
[{"answers": ["src/receiptforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most re...
ctx_compress-s1-00727-twin
receiptforge
ctx_compress-s1-00727
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_receipt_window raises KeyError for a missing key instead of returning the default. It lives in src/receiptforge/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 receiptf...
29,420
1,471
51
104
false
synth:2791054778109604221@1.61325
synth
[{"aliases": ["src/receiptforge/io.py", "io.py"], "kind": "file", "value": "src/receiptforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/payout_3.csv", "payout_3.csv"], "kind": "file", "value": "data/payout_3.csv"}, {"aliases": ["data/invoice...
[{"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/payout_3.csv", "payout_3.csv"], ...
ctx_compress-s1-00728
bookingledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cadence keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/bookingledger/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 bookingle...
51,905
2,595
80
101
false
synth:5689599046354078643@1.1398
synth
[{"aliases": ["src/bookingledger/store.py", "store.py"], "kind": "file", "value": "src/bookingledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/bookingledger/report.py", "report.py"], "kind": "file", "value": "src/bookingledger/re...
[{"answers": ["src/bookingledger/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-s1-00728-twin
bookingledger
ctx_compress-s1-00728
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cadence keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/bookingledger/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 bookingle...
33,715
1,685
58
101
false
synth:5689599046354078643@1.1398
synth
[{"aliases": ["src/bookingledger/store.py", "store.py"], "kind": "file", "value": "src/bookingledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/bookingledger/report.py", "report.py"], "kind": "file", "value": "src/bookingledger/re...
[{"answers": ["src/bookingledger/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-s1-00729
balanceworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_receipt skips the first page; page 1 returns what should be page 2. It lives in src/balanceworks/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the balancewor...
93,718
3,072
105
118
false
synth:7749084062325204102@1.77572
synth
[{"aliases": ["src/balanceworks/report.py", "report.py"], "kind": "file", "value": "src/balanceworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/tariff_3.csv", "tariff_3.csv"], "kind": "file", "value": "data/tariff_3.csv"}, {"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["core.py", "src/balanceworks/core.py"]...
ctx_compress-s1-00729-twin
balanceworks
ctx_compress-s1-00729
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 skips the first page; page 1 returns what should be page 2. It lives in src/balanceworks/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the balancewor...
51,719
2,585
57
118
false
synth:7749084062325204102@1.77572
synth
[{"aliases": ["src/balanceworks/report.py", "report.py"], "kind": "file", "value": "src/balanceworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/tariff_3.csv", "tariff_3.csv"], "kind": "file", "value": "data/tariff_3.csv"}, {"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_core.py", "tests/test_core.py"],...
ctx_compress-s1-00730
claimkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim_window leaves a trailing dash when the title ends in punctuation. It lives in src/claimkit/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 claimkit layout. [...
99,574
3,072
111
140
false
synth:4080993949201343565@2.1097
synth
[{"aliases": ["src/claimkit/cli.py", "cli.py"], "kind": "file", "value": "src/claimkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/claimkit/audit.py", "audit.py"], "kind": "file", "value": "src/claimkit/audit.py"}, {"aliases": ["data/scope...
[{"answers": ["src/claimkit/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-00730-twin
claimkit
ctx_compress-s1-00730
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim_window leaves a trailing dash when the title ends in punctuation. It lives in src/claimkit/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 claimkit layout. [...
67,316
3,072
83
140
false
synth:4080993949201343565@2.1097
synth
[{"aliases": ["src/claimkit/cli.py", "cli.py"], "kind": "file", "value": "src/claimkit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/claimkit/audit.py", "audit.py"], "kind": "file", "value": "src/claimkit/audit.py"}, {"aliases": ["data/scope...
[{"answers": ["src/claimkit/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-00731
counterforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_counter lets defaults overwrite explicit overrides. It lives in src/counterforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the counterforge layout. [tool...
50,286
2,514
92
115
false
synth:328057460376695113@1
synth
[{"aliases": ["src/counterforge/store.py", "store.py"], "kind": "file", "value": "src/counterforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sample_1.csv", "sample_1.csv"], "kind": "file", "value": "data/sample_1.csv"}, {"aliase...
[{"answers": ["src/counterforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00731-twin
counterforge
ctx_compress-s1-00731
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_counter lets defaults overwrite explicit overrides. It lives in src/counterforge/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the counterforge layout. [tool...
26,830
1,341
56
115
false
synth:328057460376695113@1
synth
[{"aliases": ["src/counterforge/store.py", "store.py"], "kind": "file", "value": "src/counterforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sample_1.csv", "sample_1.csv"], "kind": "file", "value": "data/sample_1.csv"}, {"aliase...
[{"answers": ["src/counterforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00732
invoiceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_credit_page raises PayoutDriftError for bytes input in strict mode instead of decoding it. It lives in src/invoiceforge/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 m...
90,380
3,072
106
117
false
synth:8120247651218722065@2.10732
synth
[{"aliases": ["src/invoiceforge/compat.py", "compat.py"], "kind": "file", "value": "src/invoiceforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/ledger_7.csv", "ledger_7.csv"], "kind": "file", "value": "data/ledger_7.csv"}, {"...
[{"answers": ["src/invoiceforge/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-s1-00732-twin
invoiceforge
ctx_compress-s1-00732
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_credit_page raises PayoutDriftError for bytes input in strict mode instead of decoding it. It lives in src/invoiceforge/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 m...
54,227
2,711
65
117
false
synth:8120247651218722065@2.10732
synth
[{"aliases": ["src/invoiceforge/compat.py", "compat.py"], "kind": "file", "value": "src/invoiceforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/ledger_7.csv", "ledger_7.csv"], "kind": "file", "value": "data/ledger_7.csv"}, {"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/invoice_5.csv", "invoice_5.csv"]...
ctx_compress-s1-00733
samplekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_series_all misses keys that differ only in case. It lives in src/samplekit/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 samplekit layout. [tool_use kind=se...
39,194
1,959
84
93
false
synth:1057827438626942860@1
synth
[{"aliases": ["src/samplekit/report.py", "report.py"], "kind": "file", "value": "src/samplekit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/samplekit/compat.py", "compat.py"], "kind": "file", "value": "src/samplekit/compat.py"},...
[{"answers": ["src/samplekit/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 mo...
ctx_compress-s1-00733-twin
samplekit
ctx_compress-s1-00733
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_series_all misses keys that differ only in case. It lives in src/samplekit/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 samplekit layout. [tool_use kind=se...
26,489
1,324
52
93
false
synth:1057827438626942860@1
synth
[{"aliases": ["src/samplekit/report.py", "report.py"], "kind": "file", "value": "src/samplekit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/samplekit/compat.py", "compat.py"], "kind": "file", "value": "src/samplekit/compat.py"},...
[{"answers": ["src/samplekit/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 mo...
ctx_compress-s1-00734
scopedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_scope misses keys that differ only in case. It lives in src/scopedesk/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 scopedesk layout. [tool_use kind=search n...
83,907
3,072
85
98
false
synth:4775917370478522392@1.64305
synth
[{"aliases": ["src/scopedesk/audit.py", "audit.py"], "kind": "file", "value": "src/scopedesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/scopedesk/io.py", "io.py"], "kind": "file", "value": "src/scopedesk/io.py"}, {"aliases": ["dat...
[{"answers": ["src/scopedesk/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-s1-00734-twin
scopedesk
ctx_compress-s1-00734
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_scope misses keys that differ only in case. It lives in src/scopedesk/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 scopedesk layout. [tool_use kind=search n...
45,121
2,256
47
98
false
synth:4775917370478522392@1.64305
synth
[{"aliases": ["src/scopedesk/audit.py", "audit.py"], "kind": "file", "value": "src/scopedesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/scopedesk/io.py", "io.py"], "kind": "file", "value": "src/scopedesk/io.py"}, {"aliases": ["dat...
[{"answers": ["src/scopedesk/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-00735
tokendesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_nonce ignores the cap, so late attempts wait for minutes. It lives in src/tokendesk/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 tokendesk layout. [tool_u...
68,238
3,072
88
107
false
synth:7162232637574386906@1.21935
synth
[{"aliases": ["src/tokendesk/report.py", "report.py"], "kind": "file", "value": "src/tokendesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/tokendesk/io.py", "io.py"], "kind": "file", "value": "src/tokendesk/io.py"}, {"aliases":...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/tokendesk/store.py", "store.py"],...
ctx_compress-s1-00735-twin
tokendesk
ctx_compress-s1-00735
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_nonce ignores the cap, so late attempts wait for minutes. It lives in src/tokendesk/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 tokendesk layout. [tool_u...
37,818
1,890
52
107
false
synth:7162232637574386906@1.21935
synth
[{"aliases": ["src/tokendesk/report.py", "report.py"], "kind": "file", "value": "src/tokendesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/tokendesk/io.py", "io.py"], "kind": "file", "value": "src/tokendesk/io.py"}, {"aliases":...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_compat.py", "tests/test_compat.p...
ctx_compress-s1-00736
traceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_gauge double counts carried amounts when the period closes on a weekend. It lives in src/traceforge/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 traceforge layout. [...
81,728
3,072
103
120
false
synth:8844864356841534736@1.8036
synth
[{"aliases": ["src/traceforge/io.py", "io.py"], "kind": "file", "value": "src/traceforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/traceforge/audit.py", "audit.py"], "kind": "file", "value": "src/traceforge/audit.py"}, {"aliases": ["data/pro...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/gauge_0.csv", "gauge_0.csv"], "k...
ctx_compress-s1-00736-twin
traceforge
ctx_compress-s1-00736
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_gauge double counts carried amounts when the period closes on a weekend. It lives in src/traceforge/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 traceforge layout. [...
42,219
2,110
55
120
false
synth:8844864356841534736@1.8036
synth
[{"aliases": ["src/traceforge/io.py", "io.py"], "kind": "file", "value": "src/traceforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/traceforge/audit.py", "audit.py"], "kind": "file", "value": "src/traceforge/audit.py"}, {"aliases": ["data/pro...
[{"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/traceforge/compat.py...
ctx_compress-s1-00737
cueworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_caption ignores the cap, so late attempts wait for minutes. It lives in src/cueworks/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 cueworks layout. [tool...
51,048
2,552
82
95
false
synth:8980714636643058829@1
synth
[{"aliases": ["src/cueworks/compat.py", "compat.py"], "kind": "file", "value": "src/cueworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/cueworks/tables.py", "tables.py"], "kind": "file", "value": "src/cueworks/tables.py"}, {"a...
[{"answers": ["src/cueworks/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00737-twin
cueworks
ctx_compress-s1-00737
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_caption ignores the cap, so late attempts wait for minutes. It lives in src/cueworks/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 cueworks layout. [tool...
30,712
1,535
50
95
false
synth:8980714636643058829@1
synth
[{"aliases": ["src/cueworks/compat.py", "compat.py"], "kind": "file", "value": "src/cueworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/cueworks/tables.py", "tables.py"], "kind": "file", "value": "src/cueworks/tables.py"}, {"a...
[{"answers": ["src/cueworks/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-s1-00738
gaugekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_counter grows linearly instead of doubling. It lives in src/gaugekit/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 gaugekit layout. [tool_use kind=search nam...
115,759
3,072
126
135
false
synth:8990580749168266709@1.92257
synth
[{"aliases": ["src/gaugekit/cli.py", "cli.py"], "kind": "file", "value": "src/gaugekit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/gaugekit/schema.py", "schema.py"], "kind": "file", "value": "src/gaugekit/schema.py"}, {"aliases": ["data/bu...
[{"answers": ["src/gaugekit/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rece...
ctx_compress-s1-00738-twin
gaugekit
ctx_compress-s1-00738
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_counter grows linearly instead of doubling. It lives in src/gaugekit/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 gaugekit layout. [tool_use kind=search nam...
56,787
2,839
68
135
false
synth:8990580749168266709@1.92257
synth
[{"aliases": ["src/gaugekit/cli.py", "cli.py"], "kind": "file", "value": "src/gaugekit/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/gaugekit/schema.py", "schema.py"], "kind": "file", "value": "src/gaugekit/schema.py"}, {"aliases": ["data/bu...
[{"answers": ["src/gaugekit/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-00739
spandesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_sample averages one value too many once the window is full. It lives in src/spandesk/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 spandesk layout. [tool_use kind=se...
46,669
2,333
70
79
false
synth:1202154584458415978@1
synth
[{"aliases": ["src/spandesk/io.py", "io.py"], "kind": "file", "value": "src/spandesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/spandesk/schema.py", "schema.py"], "kind": "file", "value": "src/spandesk/schema.py"}, {"aliases": ["src/spandesk/...
[{"answers": ["src/spandesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent...
ctx_compress-s1-00739-twin
spandesk
ctx_compress-s1-00739
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_sample averages one value too many once the window is full. It lives in src/spandesk/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 spandesk layout. [tool_use kind=se...
27,475
1,373
46
79
false
synth:1202154584458415978@1
synth
[{"aliases": ["src/spandesk/io.py", "io.py"], "kind": "file", "value": "src/spandesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/spandesk/schema.py", "schema.py"], "kind": "file", "value": "src/spandesk/schema.py"}, {"aliases": ["src/spandesk/...
[{"answers": ["src/spandesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recent...
ctx_compress-s1-00740
stockdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_sku double counts carried amounts when the period closes on a weekend. It lives in src/stockdesk/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 stockdesk layout....
64,010
3,072
100
111
false
synth:5637436983337430262@1.20395
synth
[{"aliases": ["src/stockdesk/store.py", "store.py"], "kind": "file", "value": "src/stockdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sku_2.csv", "sku_2.csv"], "kind": "file", "value": "data/sku_2.csv"}, {"aliases": ["src/stock...
[{"answers": ["src/stockdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-00740-twin
stockdesk
ctx_compress-s1-00740
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_sku double counts carried amounts when the period closes on a weekend. It lives in src/stockdesk/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 stockdesk layout....
27,395
1,369
50
111
false
synth:5637436983337430262@1.20395
synth
[{"aliases": ["src/stockdesk/store.py", "store.py"], "kind": "file", "value": "src/stockdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/sku_2.csv", "sku_2.csv"], "kind": "file", "value": "data/sku_2.csv"}, {"aliases": ["src/stock...
[{"answers": ["src/stockdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00741
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] reconcile_tariff double counts carried amounts when the period closes on a weekend. It lives in src/creditdesk/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 creditdes...
54,222
2,711
102
109
false
synth:1156732651204088018@1.46033
synth
[{"aliases": ["src/creditdesk/core.py", "core.py"], "kind": "file", "value": "src/creditdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["data...
[{"answers": ["src/creditdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00741-twin
creditdesk
ctx_compress-s1-00741
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_tariff double counts carried amounts when the period closes on a weekend. It lives in src/creditdesk/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 creditdes...
41,418
2,070
64
109
false
synth:1156732651204088018@1.46033
synth
[{"aliases": ["src/creditdesk/core.py", "core.py"], "kind": "file", "value": "src/creditdesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["data...
[{"answers": ["src/creditdesk/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00742
stemworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_caption misses keys that differ only in case. It lives in src/stemworks/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 stemworks layout. [tool_use kind=se...
83,828
3,072
96
115
false
synth:4827522671938818064@2.13512
synth
[{"aliases": ["src/stemworks/format.py", "format.py"], "kind": "file", "value": "src/stemworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/cue_0.csv", "cue_0.csv"], "kind": "file", "value": "data/cue_0.csv"}, {"aliases": ["dat...
[{"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/stemworks/core.py"], "...
ctx_compress-s1-00742-twin
stemworks
ctx_compress-s1-00742
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] tokenise_caption misses keys that differ only in case. It lives in src/stemworks/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 stemworks layout. [tool_use kind=se...
63,241
3,072
66
115
false
synth:4827522671938818064@2.13512
synth
[{"aliases": ["src/stemworks/format.py", "format.py"], "kind": "file", "value": "src/stemworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/cue_0.csv", "cue_0.csv"], "kind": "file", "value": "data/cue_0.csv"}, {"aliases": ["dat...
[{"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/stemworks/rules.py"],...
ctx_compress-s1-00743
seriesflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/seriesflow/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 seriesflo...
87,679
3,072
120
147
false
synth:588092166281428738@2.0985
synth
[{"aliases": ["src/seriesflow/cli.py", "cli.py"], "kind": "file", "value": "src/seriesflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/sample_2.csv", "sample_2.csv"], "kind": "file", "value": "data/sample_2.csv"}, {"aliases": ["src/series...
[{"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/seriesflow/tables.py", "tables.py...
ctx_compress-s1-00743-twin
seriesflow
ctx_compress-s1-00743
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_counter rejects a range where lo equals hi, which is a valid single point range. It lives in src/seriesflow/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 seriesflo...
63,789
3,072
88
147
false
synth:588092166281428738@2.0985
synth
[{"aliases": ["src/seriesflow/cli.py", "cli.py"], "kind": "file", "value": "src/seriesflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/sample_2.csv", "sample_2.csv"], "kind": "file", "value": "data/sample_2.csv"}, {"aliases": ["src/series...
[{"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/seriesflow/limits.py...
ctx_compress-s1-00744
seriesflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_gauge_window averages one value too many once the window is full. It lives in src/seriesflow/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 seriesflow layout. [t...
18,560
1,000
41
48
false
synth:7916556967236088067@1
synth
[{"aliases": ["src/seriesflow/rules.py", "rules.py"], "kind": "file", "value": "src/seriesflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/counter_3.csv", "counter_3.csv"], "kind": "file", "value": "data/counter_3.csv"}, {"aliases...
[{"answers": ["src/seriesflow/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-s1-00744-twin
seriesflow
ctx_compress-s1-00744
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_gauge_window averages one value too many once the window is full. It lives in src/seriesflow/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 seriesflow layout. [t...
10,348
1,000
25
48
false
synth:7916556967236088067@1
synth
[{"aliases": ["src/seriesflow/rules.py", "rules.py"], "kind": "file", "value": "src/seriesflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/counter_3.csv", "counter_3.csv"], "kind": "file", "value": "data/counter_3.csv"}, {"aliases...
[{"answers": ["src/seriesflow/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-s1-00745
counterworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series raises ValueError when a value itself contains an equals sign. It lives in src/counterworks/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 counterwor...
93,134
3,072
108
129
false
synth:4730720379756976380@1.66675
synth
[{"aliases": ["src/counterworks/core.py", "core.py"], "kind": "file", "value": "src/counterworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/trace_8.csv", "trace_8.csv"], "kind": "file", "value": "data/trace_8.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": ["limits.py", "src/counterworks/limits....
ctx_compress-s1-00745-twin
counterworks
ctx_compress-s1-00745
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series raises ValueError when a value itself contains an equals sign. It lives in src/counterworks/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 counterwor...
44,412
2,220
62
129
false
synth:4730720379756976380@1.66675
synth
[{"aliases": ["src/counterworks/core.py", "core.py"], "kind": "file", "value": "src/counterworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/trace_8.csv", "trace_8.csv"], "kind": "file", "value": "data/trace_8.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": ["data/probe_0.csv", "probe_0.csv"], "k...
ctx_compress-s1-00746
reorderledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_sku_batch raises PalletShapeError for bytes input in strict mode instead of decoding it. It lives in src/reorderledger/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 ...
38,472
1,923
86
93
false
synth:4299710673896595255@1
synth
[{"aliases": ["src/reorderledger/io.py", "io.py"], "kind": "file", "value": "src/reorderledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/reorderledger/report.py", "report.py"], "kind": "file", "value": "src/reorderledger/report.py"}, {"alias...
[{"answers": ["src/reorderledger/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-00746-twin
reorderledger
ctx_compress-s1-00746
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_sku_batch raises PalletShapeError for bytes input in strict mode instead of decoding it. It lives in src/reorderledger/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 ...
21,629
1,081
52
93
false
synth:4299710673896595255@1
synth
[{"aliases": ["src/reorderledger/io.py", "io.py"], "kind": "file", "value": "src/reorderledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/reorderledger/report.py", "report.py"], "kind": "file", "value": "src/reorderledger/report.py"}, {"alias...
[{"answers": ["src/reorderledger/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most r...
ctx_compress-s1-00747
stockworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_stock returns the upper bound for every value instead of clamping into the range. It lives in src/stockworks/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 th...
63,039
3,072
84
101
false
synth:4441469778259168191@1.10692
synth
[{"aliases": ["src/stockworks/compat.py", "compat.py"], "kind": "file", "value": "src/stockworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/stockworks/tables.py", "tables.py"], "kind": "file", "value": "src/stockworks/tables.p...
[{"answers": ["src/stockworks/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-s1-00747-twin
stockworks
ctx_compress-s1-00747
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_stock returns the upper bound for every value instead of clamping into the range. It lives in src/stockworks/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 th...
33,133
1,656
50
101
false
synth:4441469778259168191@1.10692
synth
[{"aliases": ["src/stockworks/compat.py", "compat.py"], "kind": "file", "value": "src/stockworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/stockworks/tables.py", "tables.py"], "kind": "file", "value": "src/stockworks/tables.p...
[{"answers": ["src/stockworks/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-s1-00748
stockflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reserve_sku_safe returns one item too few on every page. It lives in src/stockflow/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 stockflow layout. [tool_use kind=...
40,828
2,041
66
75
false
synth:7454038719251674490@1
synth
[{"aliases": ["src/stockflow/compat.py", "compat.py"], "kind": "file", "value": "src/stockflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliase...
[{"answers": ["src/stockflow/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00748-twin
stockflow
ctx_compress-s1-00748
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reserve_sku_safe returns one item too few on every page. It lives in src/stockflow/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 stockflow layout. [tool_use kind=...
29,873
1,493
44
75
false
synth:7454038719251674490@1
synth
[{"aliases": ["src/stockflow/compat.py", "compat.py"], "kind": "file", "value": "src/stockflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliase...
[{"answers": ["src/stockflow/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00749
sectiondesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 raises FootnoteStateError for bytes input in strict mode instead of decoding it. It lives in src/sectiondesk/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 mapp...
37,020
1,851
69
88
false
synth:7155879012244389821@1
synth
[{"aliases": ["src/sectiondesk/store.py", "store.py"], "kind": "file", "value": "src/sectiondesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/sectiondesk/schema.py", "schema.py"], "kind": "file", "value": "src/sectiondesk/schema.py"...
[{"answers": ["src/sectiondesk/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-s1-00749-twin
sectiondesk
ctx_compress-s1-00749
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 raises FootnoteStateError for bytes input in strict mode instead of decoding it. It lives in src/sectiondesk/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 mapp...
22,719
1,135
49
88
false
synth:7155879012244389821@1
synth
[{"aliases": ["src/sectiondesk/store.py", "store.py"], "kind": "file", "value": "src/sectiondesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/sectiondesk/schema.py", "schema.py"], "kind": "file", "value": "src/sectiondesk/schema.py"...
[{"answers": ["src/sectiondesk/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-s1-00750
stockforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] restock_reorder double counts carried amounts when the period closes on a weekend. It lives in src/stockforge/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 stockfor...
85,103
3,072
103
124
false
synth:5127974433549925875@2.35862
synth
[{"aliases": ["src/stockforge/audit.py", "audit.py"], "kind": "file", "value": "src/stockforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/warehouse_3.csv", "warehouse_3.csv"], "kind": "file", "value": "data/warehouse_3.csv"}, {"a...
[{"answers": ["src/stockforge/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00750-twin
stockforge
ctx_compress-s1-00750
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] restock_reorder double counts carried amounts when the period closes on a weekend. It lives in src/stockforge/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 stockfor...
53,485
2,674
69
124
false
synth:5127974433549925875@2.35862
synth
[{"aliases": ["src/stockforge/audit.py", "audit.py"], "kind": "file", "value": "src/stockforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/warehouse_3.csv", "warehouse_3.csv"], "kind": "file", "value": "data/warehouse_3.csv"}, {"a...
[{"answers": ["src/stockforge/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 mos...
ctx_compress-s1-00751
draftforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_heading_window skips the first page; page 1 returns what should be page 2. It lives in src/draftforge/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 draftfo...
44,449
2,222
73
90
false
synth:3152999529738955834@1.19055
synth
[{"aliases": ["src/draftforge/rules.py", "rules.py"], "kind": "file", "value": "src/draftforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/draftforge/compat.py", "compat.py"], "kind": "file", "value": "src/draftforge/compat.py"}, {...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/section_1.csv", "section_1.csv"]...
ctx_compress-s1-00751-twin
draftforge
ctx_compress-s1-00751
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_heading_window skips the first page; page 1 returns what should be page 2. It lives in src/draftforge/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 draftfo...
37,837
1,891
53
90
false
synth:3152999529738955834@1.19055
synth
[{"aliases": ["src/draftforge/rules.py", "rules.py"], "kind": "file", "value": "src/draftforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/draftforge/compat.py", "compat.py"], "kind": "file", "value": "src/draftforge/compat.py"}, {...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/draftforge/limits.py...
ctx_compress-s1-00752
grantflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_policy grows linearly instead of doubling. It lives in src/grantflow/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 grantflow layout. [tool_use kind=search name=B...
78,214
3,072
108
137
false
synth:6024379716765499156@2.0846
synth
[{"aliases": ["src/grantflow/cli.py", "cli.py"], "kind": "file", "value": "src/grantflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/grantflow/store.py", "store.py"], "kind": "file", "value": "src/grantflow/store.py"}, {"aliases": ["data/n...
[{"answers": ["src/grantflow/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-00752-twin
grantflow
ctx_compress-s1-00752
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_policy grows linearly instead of doubling. It lives in src/grantflow/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 grantflow layout. [tool_use kind=search name=B...
53,396
2,669
78
137
false
synth:6024379716765499156@2.0846
synth
[{"aliases": ["src/grantflow/cli.py", "cli.py"], "kind": "file", "value": "src/grantflow/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/grantflow/store.py", "store.py"], "kind": "file", "value": "src/grantflow/store.py"}, {"aliases": ["data/n...
[{"answers": ["src/grantflow/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-00753
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] revoke_claim skips the first page; page 1 returns what should be page 2. It lives in src/principalkit/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 principalkit layou...
36,211
1,810
84
99
false
synth:1052687703553687952@1
synth
[{"aliases": ["src/principalkit/core.py", "core.py"], "kind": "file", "value": "src/principalkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/claim_2.csv", "claim_2.csv"], "kind": "file", "value": "data/claim_2.csv"}, {"aliases": ["src...
[{"answers": ["src/principalkit/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-s1-00753-twin
principalkit
ctx_compress-s1-00753
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim skips the first page; page 1 returns what should be page 2. It lives in src/principalkit/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 principalkit layou...
23,783
1,189
56
99
false
synth:1052687703553687952@1
synth
[{"aliases": ["src/principalkit/core.py", "core.py"], "kind": "file", "value": "src/principalkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/claim_2.csv", "claim_2.csv"], "kind": "file", "value": "data/claim_2.csv"}, {"aliases": ["src...
[{"answers": ["src/principalkit/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-s1-00754
sectionledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_index raises SectionRangeError for bytes input in strict mode instead of decoding it. It lives in src/sectionledger/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...
65,227
3,072
103
110
false
synth:740464941829183886@1.45745
synth
[{"aliases": ["src/sectionledger/store.py", "store.py"], "kind": "file", "value": "src/sectionledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/sectionledger/tables.py", "tables.py"], "kind": "file", "value": "src/sectionledger/ta...
[{"answers": ["src/sectionledger/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-s1-00754-twin
sectionledger
ctx_compress-s1-00754
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_index raises SectionRangeError for bytes input in strict mode instead of decoding it. It lives in src/sectionledger/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...
37,314
1,865
53
110
false
synth:740464941829183886@1.45745
synth
[{"aliases": ["src/sectionledger/store.py", "store.py"], "kind": "file", "value": "src/sectionledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/sectionledger/tables.py", "tables.py"], "kind": "file", "value": "src/sectionledger/ta...
[{"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/sectionledger/schema...