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-s1-00452-twin
tokenkit
ctx_compress-s1-00452
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_grant double counts carried amounts when the period closes on a weekend. It lives in src/tokenkit/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 tokenkit layout. [t...
28,357
1,417
47
103
false
synth:3326873436872431527@1.4957
synth
[{"aliases": ["src/tokenkit/io.py", "io.py"], "kind": "file", "value": "src/tokenkit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/tokenkit/format.py", "format.py"], "kind": "file", "value": "src/tokenkit/format.py"}, {"aliases": ["src/tokenkit/...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/tokenkit/format.py"]...
ctx_compress-s1-00453
rosterworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_calendar returns the upper bound for every value instead of clamping into the range. It lives in src/rosterworks/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 ro...
99,206
3,072
91
106
false
synth:6034673624995286098@1.7094
synth
[{"aliases": ["src/rosterworks/cli.py", "cli.py"], "kind": "file", "value": "src/rosterworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/quota_10.csv", "quota_10.csv"], "kind": "file", "value": "data/quota_10.csv"}, {"aliases": ["data/ros...
[{"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/quota_9.csv", "quota_9.csv"], "k...
ctx_compress-s1-00453-twin
rosterworks
ctx_compress-s1-00453
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_calendar returns the upper bound for every value instead of clamping into the range. It lives in src/rosterworks/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 ro...
56,055
2,802
59
106
false
synth:6034673624995286098@1.7094
synth
[{"aliases": ["src/rosterworks/cli.py", "cli.py"], "kind": "file", "value": "src/rosterworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/quota_10.csv", "quota_10.csv"], "kind": "file", "value": "data/quota_10.csv"}, {"aliases": ["data/ros...
[{"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/slot_6.csv", "slot_6.csv"], "kin...
ctx_compress-s1-00454
tariffdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_invoice_window grows linearly instead of doubling. It lives in src/tariffdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tariffdesk layout. [tool_use kind=se...
80,477
3,072
108
133
false
synth:2662084719251992724@2.27933
synth
[{"aliases": ["src/tariffdesk/store.py", "store.py"], "kind": "file", "value": "src/tariffdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/tariffdesk/compat.py", "compat.py"], "kind": "file", "value": "src/tariffdesk/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/payout_0.csv", "payout_0.csv"], ...
ctx_compress-s1-00454-twin
tariffdesk
ctx_compress-s1-00454
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_invoice_window grows linearly instead of doubling. It lives in src/tariffdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the tariffdesk layout. [tool_use kind=se...
50,755
2,537
70
133
false
synth:2662084719251992724@2.27933
synth
[{"aliases": ["src/tariffdesk/store.py", "store.py"], "kind": "file", "value": "src/tariffdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/tariffdesk/compat.py", "compat.py"], "kind": "file", "value": "src/tariffdesk/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": ["cli.py", "src/tariffdesk/cli.py"], "k...
ctx_compress-s1-00455
glyphdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cue grows linearly instead of doubling. It lives in src/glyphdesk/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 glyphdesk layout. [tool_use kind=search name=...
18,176
1,000
48
53
false
synth:3191210565277584891@1
synth
[{"aliases": ["src/glyphdesk/report.py", "report.py"], "kind": "file", "value": "src/glyphdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/glyphdesk/schema.py", "schema.py"], "kind": "file", "value": "src/glyphdesk/schema.py"},...
[{"answers": ["src/glyphdesk/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-s1-00456
tracedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_trace produces doubled dashes for consecutive separators. It lives in src/tracedesk/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 tracedesk layout. [tool_use ki...
70,918
3,072
92
113
false
synth:6014045415980740284@1.81583
synth
[{"aliases": ["src/tracedesk/audit.py", "audit.py"], "kind": "file", "value": "src/tracedesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/tracedesk/limits.py", "limits.py"], "kind": "file", "value": "src/tracedesk/limits.py"}, {"ali...
[{"answers": ["src/tracedesk/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-s1-00456-twin
tracedesk
ctx_compress-s1-00456
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_trace produces doubled dashes for consecutive separators. It lives in src/tracedesk/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 tracedesk layout. [tool_use ki...
42,182
2,109
64
113
false
synth:6014045415980740284@1.81583
synth
[{"aliases": ["src/tracedesk/audit.py", "audit.py"], "kind": "file", "value": "src/tracedesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/tracedesk/limits.py", "limits.py"], "kind": "file", "value": "src/tracedesk/limits.py"}, {"ali...
[{"answers": ["src/tracedesk/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-s1-00457
counterflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bucket_once ignores the cap, so late attempts wait for minutes. It lives in src/counterflow/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 counterflow layout. [to...
47,991
2,399
63
76
false
synth:4635516983771242073@1
synth
[{"aliases": ["src/counterflow/core.py", "core.py"], "kind": "file", "value": "src/counterflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/series_0.csv", "series_0.csv"], "kind": "file", "value": "data/series_0.csv"}, {"aliases": ["da...
[{"answers": ["src/counterflow/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00457-twin
counterflow
ctx_compress-s1-00457
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bucket_once ignores the cap, so late attempts wait for minutes. It lives in src/counterflow/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 counterflow layout. [to...
33,629
1,681
43
76
false
synth:4635516983771242073@1
synth
[{"aliases": ["src/counterflow/core.py", "core.py"], "kind": "file", "value": "src/counterflow/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/series_0.csv", "series_0.csv"], "kind": "file", "value": "data/series_0.csv"}, {"aliases": ["da...
[{"answers": ["src/counterflow/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-s1-00458
claimforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_principal raises ScopeRangeError for bytes input in strict mode instead of decoding it. It lives in src/claimforge/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 mappin...
39,529
1,976
71
80
false
synth:2295835685450799438@1
synth
[{"aliases": ["src/claimforge/compat.py", "compat.py"], "kind": "file", "value": "src/claimforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/grant_0.csv", "grant_0.csv"], "kind": "file", "value": "data/grant_0.csv"}, {"aliases...
[{"answers": ["src/claimforge/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-s1-00458-twin
claimforge
ctx_compress-s1-00458
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_principal raises ScopeRangeError for bytes input in strict mode instead of decoding it. It lives in src/claimforge/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 mappin...
23,775
1,188
39
80
false
synth:2295835685450799438@1
synth
[{"aliases": ["src/claimforge/compat.py", "compat.py"], "kind": "file", "value": "src/claimforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/grant_0.csv", "grant_0.csv"], "kind": "file", "value": "data/grant_0.csv"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["schema.py", "src/claimforge/schema.py...
ctx_compress-s1-00459
palletkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_bin lets defaults overwrite explicit overrides. It lives in src/palletkit/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 palletkit layout. [tool_use kind=sea...
88,687
3,072
121
142
false
synth:8773445627866125129@2.26762
synth
[{"aliases": ["src/palletkit/compat.py", "compat.py"], "kind": "file", "value": "src/palletkit/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/palletkit/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-s1-00459-twin
palletkit
ctx_compress-s1-00459
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_bin lets defaults overwrite explicit overrides. It lives in src/palletkit/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 palletkit layout. [tool_use kind=sea...
59,951
2,997
75
142
false
synth:8773445627866125129@2.26762
synth
[{"aliases": ["src/palletkit/compat.py", "compat.py"], "kind": "file", "value": "src/palletkit/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/palletkit/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-s1-00460
seriesledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_gauge drops override values of 0, False and empty string. It lives in src/seriesledger/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 seriesledger layout. [to...
97,301
3,072
115
126
false
synth:9092208910453417362@1.6717
synth
[{"aliases": ["src/seriesledger/compat.py", "compat.py"], "kind": "file", "value": "src/seriesledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/seriesledger/schema.py", "schema.py"], "kind": "file", "value": "src/seriesledger/...
[{"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/seriesledger/compat....
ctx_compress-s1-00460-twin
seriesledger
ctx_compress-s1-00460
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_gauge drops override values of 0, False and empty string. It lives in src/seriesledger/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 seriesledger layout. [to...
58,996
2,949
69
126
false
synth:9092208910453417362@1.6717
synth
[{"aliases": ["src/seriesledger/compat.py", "compat.py"], "kind": "file", "value": "src/seriesledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/seriesledger/schema.py", "schema.py"], "kind": "file", "value": "src/seriesledger/...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["counter_3.csv", "data/counter_3.csv"]...
ctx_compress-s1-00461
principalforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_token drops override values of 0, False and empty string. It lives in src/principalforge/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 principalforge layou...
63,581
3,072
96
105
false
synth:6312684508687824843@1.41502
synth
[{"aliases": ["src/principalforge/limits.py", "limits.py"], "kind": "file", "value": "src/principalforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/principalforge/cli.py", "cli.py"], "kind": "file", "value": "src/principalforg...
[{"answers": ["src/principalforge/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent...
ctx_compress-s1-00461-twin
principalforge
ctx_compress-s1-00461
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_token drops override values of 0, False and empty string. It lives in src/principalforge/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 principalforge layou...
32,040
1,602
51
105
false
synth:6312684508687824843@1.41502
synth
[{"aliases": ["src/principalforge/limits.py", "limits.py"], "kind": "file", "value": "src/principalforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/principalforge/cli.py", "cli.py"], "kind": "file", "value": "src/principalforg...
[{"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/principalforge/compa...
ctx_compress-s1-00462
bearingforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_all double counts carried amounts when the period closes on a weekend. It lives in src/bearingforge/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 ...
33,452
1,672
62
75
false
synth:3686003344093868904@1
synth
[{"aliases": ["src/bearingforge/compat.py", "compat.py"], "kind": "file", "value": "src/bearingforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/bearingforge/tables.py", "tables.py"], "kind": "file", "value": "src/bearingforge/...
[{"answers": ["src/bearingforge/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-00462-twin
bearingforge
ctx_compress-s1-00462
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_all double counts carried amounts when the period closes on a weekend. It lives in src/bearingforge/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 ...
22,198
1,109
44
75
false
synth:3686003344093868904@1
synth
[{"aliases": ["src/bearingforge/compat.py", "compat.py"], "kind": "file", "value": "src/bearingforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/bearingforge/tables.py", "tables.py"], "kind": "file", "value": "src/bearingforge/...
[{"answers": ["src/bearingforge/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-00463
payoutdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_tariff_all lets defaults overwrite explicit overrides. It lives in src/payoutdesk/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 payoutdesk layout. [tool_use ki...
49,371
2,468
84
91
false
synth:7758678307242194676@1.36187
synth
[{"aliases": ["src/payoutdesk/rules.py", "rules.py"], "kind": "file", "value": "src/payoutdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/payoutdesk/format.py", "format.py"], "kind": "file", "value": "src/payoutdesk/format.py"}, {...
[{"answers": ["src/payoutdesk/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-s1-00463-twin
payoutdesk
ctx_compress-s1-00463
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_tariff_all lets defaults overwrite explicit overrides. It lives in src/payoutdesk/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 payoutdesk layout. [tool_use ki...
26,051
1,302
47
91
false
synth:7758678307242194676@1.36187
synth
[{"aliases": ["src/payoutdesk/rules.py", "rules.py"], "kind": "file", "value": "src/payoutdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/payoutdesk/format.py", "format.py"], "kind": "file", "value": "src/payoutdesk/format.py"}, {...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-00464
tariffforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch grows linearly instead of doubling. It lives in src/tariffforge/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 tariffforge layout. [tool_use ki...
96,494
3,072
132
147
false
synth:8620460447978561389@1.78767
synth
[{"aliases": ["src/tariffforge/report.py", "report.py"], "kind": "file", "value": "src/tariffforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/tariffforge/format.py", "format.py"], "kind": "file", "value": "src/tariffforge/form...
[{"answers": ["src/tariffforge/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-s1-00464-twin
tariffforge
ctx_compress-s1-00464
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch grows linearly instead of doubling. It lives in src/tariffforge/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 tariffforge layout. [tool_use ki...
46,555
2,327
84
147
false
synth:8620460447978561389@1.78767
synth
[{"aliases": ["src/tariffforge/report.py", "report.py"], "kind": "file", "value": "src/tariffforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/tariffforge/format.py", "format.py"], "kind": "file", "value": "src/tariffforge/form...
[{"answers": ["src/tariffforge/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-s1-00465
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] bin_counter keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/probeflow/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 probeflow layout...
69,101
3,072
99
106
false
synth:4521692502164637281@1.06973
synth
[{"aliases": ["src/probeflow/store.py", "store.py"], "kind": "file", "value": "src/probeflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/bucket_4.csv", "bucket_4.csv"], "kind": "file", "value": "data/bucket_4.csv"}, {"aliases": ["...
[{"answers": ["src/probeflow/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-s1-00465-twin
probeflow
ctx_compress-s1-00465
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/probeflow/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 probeflow layout...
30,761
1,538
49
106
false
synth:4521692502164637281@1.06973
synth
[{"aliases": ["src/probeflow/store.py", "store.py"], "kind": "file", "value": "src/probeflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/bucket_4.csv", "bucket_4.csv"], "kind": "file", "value": "data/bucket_4.csv"}, {"aliases": ["...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_report.py", "tests/test_report.p...
ctx_compress-s1-00466
captionworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_lexeme_safe leaves a trailing dash when the title ends in punctuation. It lives in src/captionworks/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 captionwo...
45,496
2,274
78
83
false
synth:2163533959201107228@1.18332
synth
[{"aliases": ["src/captionworks/report.py", "report.py"], "kind": "file", "value": "src/captionworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/captionworks/core.py", "core.py"], "kind": "file", "value": "src/captionworks/core...
[{"answers": ["src/captionworks/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-00466-twin
captionworks
ctx_compress-s1-00466
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_lexeme_safe leaves a trailing dash when the title ends in punctuation. It lives in src/captionworks/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 captionwo...
24,663
1,233
43
83
false
synth:2163533959201107228@1.18332
synth
[{"aliases": ["src/captionworks/report.py", "report.py"], "kind": "file", "value": "src/captionworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/captionworks/core.py", "core.py"], "kind": "file", "value": "src/captionworks/core...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/captionworks/compat....
ctx_compress-s1-00467
footnotedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_revision_window returns the upper bound for every value instead of clamping into the range. It lives in src/footnotedesk/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 mappin...
50,958
2,547
80
97
false
synth:5560261171947148401@1
synth
[{"aliases": ["src/footnotedesk/core.py", "core.py"], "kind": "file", "value": "src/footnotedesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/heading_6.csv", "heading_6.csv"], "kind": "file", "value": "data/heading_6.csv"}, {"aliases":...
[{"answers": ["src/footnotedesk/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-s1-00467-twin
footnotedesk
ctx_compress-s1-00467
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_revision_window returns the upper bound for every value instead of clamping into the range. It lives in src/footnotedesk/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 mappin...
28,669
1,433
50
97
false
synth:5560261171947148401@1
synth
[{"aliases": ["src/footnotedesk/core.py", "core.py"], "kind": "file", "value": "src/footnotedesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/heading_6.csv", "heading_6.csv"], "kind": "file", "value": "data/heading_6.csv"}, {"aliases":...
[{"answers": ["src/footnotedesk/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-s1-00468
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_sample returns the upper bound for every value instead of clamping into the range. 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 s...
46,828
2,341
99
112
false
synth:8446391241030840851@1.4881
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": ["data/span_0.csv", "span_0.csv"], "kind": "file", "value": "data/span_0.csv"}, {"aliases": ["...
[{"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": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00468-twin
samplekit
ctx_compress-s1-00468
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_sample returns the upper bound for every value instead of clamping into the range. 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 s...
29,815
1,490
63
112
false
synth:8446391241030840851@1.4881
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": ["data/span_0.csv", "span_0.csv"], "kind": "file", "value": "data/span_0.csv"}, {"aliases": ["...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/samplekit/format.py"...
ctx_compress-s1-00469
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] rotate_claim raises ValueError when a value itself contains an equals sign. It lives in src/sessionforge/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 sessionforge la...
52,232
2,611
90
101
false
synth:8701633233269016824@1
synth
[{"aliases": ["src/sessionforge/core.py", "core.py"], "kind": "file", "value": "src/sessionforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/sessionforge/io.py", "io.py"], "kind": "file", "value": "src/sessionforge/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": ["core.py", "src/sessionforge/core.py"]...
ctx_compress-s1-00469-twin
sessionforge
ctx_compress-s1-00469
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim raises ValueError when a value itself contains an equals sign. It lives in src/sessionforge/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 sessionforge la...
26,417
1,320
48
101
false
synth:8701633233269016824@1
synth
[{"aliases": ["src/sessionforge/core.py", "core.py"], "kind": "file", "value": "src/sessionforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/sessionforge/io.py", "io.py"], "kind": "file", "value": "src/sessionforge/io.py"}, {"aliases":...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/sessionforge/audit.py...
ctx_compress-s1-00470
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] revoke_grant produces doubled dashes for consecutive separators. It lives in src/grantforge/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 grantforge layout. [tool...
73,708
3,072
106
119
false
synth:635156336370941246@1.72687
synth
[{"aliases": ["src/grantforge/format.py", "format.py"], "kind": "file", "value": "src/grantforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/grantforge/io.py", "io.py"], "kind": "file", "value": "src/grantforge/io.py"}, {"alias...
[{"answers": ["src/grantforge/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00470-twin
grantforge
ctx_compress-s1-00470
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_grant produces doubled dashes for consecutive separators. It lives in src/grantforge/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 grantforge layout. [tool...
42,883
2,144
66
119
false
synth:635156336370941246@1.72687
synth
[{"aliases": ["src/grantforge/format.py", "format.py"], "kind": "file", "value": "src/grantforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/grantforge/io.py", "io.py"], "kind": "file", "value": "src/grantforge/io.py"}, {"alias...
[{"answers": ["src/grantforge/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-00471
indexflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_footnote_batch keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/indexflow/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 inde...
17,082
1,000
36
43
false
synth:1245273214366317970@1
synth
[{"aliases": ["src/indexflow/compat.py", "compat.py"], "kind": "file", "value": "src/indexflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/clause_2.csv", "clause_2.csv"], "kind": "file", "value": "data/clause_2.csv"}, {"aliase...
[{"answers": ["src/indexflow/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-s1-00472
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] collate_heading_once leaves a trailing dash when the title ends in punctuation. It lives in src/sectionledger/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 sectionledge...
48,458
2,422
73
92
false
synth:6311495058055702627@1.21988
synth
[{"aliases": ["src/sectionledger/cli.py", "cli.py"], "kind": "file", "value": "src/sectionledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/heading_0.csv", "heading_0.csv"], "kind": "file", "value": "data/heading_0.csv"}, {"aliases": ["s...
[{"answers": ["src/sectionledger/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-00472-twin
sectionledger
ctx_compress-s1-00472
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_heading_once leaves a trailing dash when the title ends in punctuation. It lives in src/sectionledger/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 sectionledge...
23,986
1,199
43
92
false
synth:6311495058055702627@1.21988
synth
[{"aliases": ["src/sectionledger/cli.py", "cli.py"], "kind": "file", "value": "src/sectionledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/heading_0.csv", "heading_0.csv"], "kind": "file", "value": "data/heading_0.csv"}, {"aliases": ["s...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-00473
palletledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse double counts carried amounts when the period closes on a weekend. It lives in src/palletledger/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...
61,698
3,072
87
100
false
synth:8630871591146248513@1.0886
synth
[{"aliases": ["src/palletledger/report.py", "report.py"], "kind": "file", "value": "src/palletledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/palletledger/audit.py", "audit.py"], "kind": "file", "value": "src/palletledger/au...
[{"answers": ["src/palletledger/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-s1-00473-twin
palletledger
ctx_compress-s1-00473
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse double counts carried amounts when the period closes on a weekend. It lives in src/palletledger/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...
36,058
1,802
55
100
false
synth:8630871591146248513@1.0886
synth
[{"aliases": ["src/palletledger/report.py", "report.py"], "kind": "file", "value": "src/palletledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/palletledger/audit.py", "audit.py"], "kind": "file", "value": "src/palletledger/au...
[{"answers": ["src/palletledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-00474
bearingworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_zone_batch skips the first page; page 1 returns what should be page 2. It lives in src/bearingworks/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 bearingw...
73,010
3,072
104
117
false
synth:4240059669764674310@2.23785
synth
[{"aliases": ["src/bearingworks/format.py", "format.py"], "kind": "file", "value": "src/bearingworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/bearingworks/schema.py", "schema.py"], "kind": "file", "value": "src/bearingworks/...
[{"answers": ["src/bearingworks/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-00474-twin
bearingworks
ctx_compress-s1-00474
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch skips the first page; page 1 returns what should be page 2. It lives in src/bearingworks/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 bearingw...
45,417
2,270
61
117
false
synth:4240059669764674310@2.23785
synth
[{"aliases": ["src/bearingworks/format.py", "format.py"], "kind": "file", "value": "src/bearingworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/bearingworks/schema.py", "schema.py"], "kind": "file", "value": "src/bearingworks/...
[{"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/bearingworks/limits....
ctx_compress-s1-00475
cuedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_stem rejects a range where lo equals hi, which is a valid single point range. It lives in src/cuedesk/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 cuedesk layout....
69,629
3,072
91
102
false
synth:2372978336004990371@1.33003
synth
[{"aliases": ["src/cuedesk/cli.py", "cli.py"], "kind": "file", "value": "src/cuedesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/snippet_4.csv", "snippet_4.csv"], "kind": "file", "value": "data/snippet_4.csv"}, {"aliases": ["data/snippet_...
[{"answers": ["src/cuedesk/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 recen...
ctx_compress-s1-00475-twin
cuedesk
ctx_compress-s1-00475
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_stem rejects a range where lo equals hi, which is a valid single point range. It lives in src/cuedesk/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 cuedesk layout....
35,897
1,794
57
102
false
synth:2372978336004990371@1.33003
synth
[{"aliases": ["src/cuedesk/cli.py", "cli.py"], "kind": "file", "value": "src/cuedesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/snippet_4.csv", "snippet_4.csv"], "kind": "file", "value": "data/snippet_4.csv"}, {"aliases": ["data/snippet_...
[{"answers": ["src/cuedesk/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 recen...
ctx_compress-s1-00476
beaconledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_leg ignores the cap, so late attempts wait for minutes. It lives in src/beaconledger/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 beaconledger layout. [too...
89,316
3,072
106
125
false
synth:1153203437407104650@2.1773
synth
[{"aliases": ["src/beaconledger/store.py", "store.py"], "kind": "file", "value": "src/beaconledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/corridor_1.csv", "corridor_1.csv"], "kind": "file", "value": "data/corridor_1.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/zone_9.csv", "zone_9.csv"], "kin...
ctx_compress-s1-00476-twin
beaconledger
ctx_compress-s1-00476
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_leg ignores the cap, so late attempts wait for minutes. It lives in src/beaconledger/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 beaconledger layout. [too...
59,448
2,972
68
125
false
synth:1153203437407104650@2.1773
synth
[{"aliases": ["src/beaconledger/store.py", "store.py"], "kind": "file", "value": "src/beaconledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/corridor_1.csv", "corridor_1.csv"], "kind": "file", "value": "data/corridor_1.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": ["schema.py", "src/beaconledger/schema....
ctx_compress-s1-00477
refundworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_page averages one value too many once the window is full. It lives in src/refundworks/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 refundworks layout...
58,739
2,936
104
111
false
synth:118413283009320503@1
synth
[{"aliases": ["src/refundworks/audit.py", "audit.py"], "kind": "file", "value": "src/refundworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/refund_6.csv", "refund_6.csv"], "kind": "file", "value": "data/refund_6.csv"}, {"aliases"...
[{"answers": ["src/refundworks/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-s1-00477-twin
refundworks
ctx_compress-s1-00477
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_page averages one value too many once the window is full. It lives in src/refundworks/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 refundworks layout...
35,959
1,797
66
111
false
synth:118413283009320503@1
synth
[{"aliases": ["src/refundworks/audit.py", "audit.py"], "kind": "file", "value": "src/refundworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/refund_6.csv", "refund_6.csv"], "kind": "file", "value": "data/refund_6.csv"}, {"aliases"...
[{"answers": ["src/refundworks/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-s1-00478
glyphledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] normalise_cue_once raises ValueError when a value itself contains an equals sign. It lives in src/glyphledger/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 glyphledger la...
44,238
2,211
66
83
false
synth:1784795835867075085@1
synth
[{"aliases": ["src/glyphledger/io.py", "io.py"], "kind": "file", "value": "src/glyphledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/glyphledge...
[{"answers": ["src/glyphledger/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-00478-twin
glyphledger
ctx_compress-s1-00478
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] normalise_cue_once raises ValueError when a value itself contains an equals sign. It lives in src/glyphledger/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 glyphledger la...
28,854
1,442
46
83
false
synth:1784795835867075085@1
synth
[{"aliases": ["src/glyphledger/io.py", "io.py"], "kind": "file", "value": "src/glyphledger/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/glyphledge...
[{"answers": ["src/glyphledger/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-00479
captionledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_chunk ignores the cap, so late attempts wait for minutes. It lives in src/captionledger/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the captionledger layout....
46,904
2,345
68
87
false
synth:487070891452535092@1
synth
[{"aliases": ["src/captionledger/audit.py", "audit.py"], "kind": "file", "value": "src/captionledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/captionledger/io.py", "io.py"], "kind": "file", "value": "src/captionledger/io.py"}, {...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_cli.py", "tests/test_cli.py"], "...
ctx_compress-s1-00479-twin
captionledger
ctx_compress-s1-00479
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_chunk ignores the cap, so late attempts wait for minutes. It lives in src/captionledger/audit.py; `python -m pytest tests/test_audit.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the captionledger layout....
23,719
1,185
42
87
false
synth:487070891452535092@1
synth
[{"aliases": ["src/captionledger/audit.py", "audit.py"], "kind": "file", "value": "src/captionledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/captionledger/io.py", "io.py"], "kind": "file", "value": "src/captionledger/io.py"}, {...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["cue_1.csv", "data/cue_1.csv"], "kind"...
ctx_compress-s1-00480
bucketforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_span raises SampleBoundError for bytes input in strict mode instead of decoding it. It lives in src/bucketforge/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...
76,993
3,072
98
111
false
synth:499445501481086532@1.76632
synth
[{"aliases": ["src/bucketforge/format.py", "format.py"], "kind": "file", "value": "src/bucketforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/bucketforge/compat.py", "compat.py"], "kind": "file", "value": "src/bucketforge/comp...
[{"answers": ["src/bucketforge/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-00480-twin
bucketforge
ctx_compress-s1-00480
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_span raises SampleBoundError for bytes input in strict mode instead of decoding it. It lives in src/bucketforge/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...
39,519
1,975
62
111
false
synth:499445501481086532@1.76632
synth
[{"aliases": ["src/bucketforge/format.py", "format.py"], "kind": "file", "value": "src/bucketforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/bucketforge/compat.py", "compat.py"], "kind": "file", "value": "src/bucketforge/comp...
[{"answers": ["src/bucketforge/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-00481
counterkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_sample rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterkit/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 cou...
70,458
3,072
98
111
false
synth:832490798329671056@1.4752
synth
[{"aliases": ["src/counterkit/limits.py", "limits.py"], "kind": "file", "value": "src/counterkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/span_2.csv", "span_2.csv"], "kind": "file", "value": "data/span_2.csv"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/trace_4.csv", "trace_4.csv"], "k...
ctx_compress-s1-00481-twin
counterkit
ctx_compress-s1-00481
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_sample rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterkit/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 cou...
36,095
1,804
52
111
false
synth:832490798329671056@1.4752
synth
[{"aliases": ["src/counterkit/limits.py", "limits.py"], "kind": "file", "value": "src/counterkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/span_2.csv", "span_2.csv"], "kind": "file", "value": "data/span_2.csv"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/trace_4.csv", "trace_4.csv"], "k...
ctx_compress-s1-00482
payoutflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_all double counts carried amounts when the period closes on a weekend. It lives in src/payoutflow/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 pa...
30,154
1,507
53
58
false
synth:5054916401317310413@1
synth
[{"aliases": ["src/payoutflow/store.py", "store.py"], "kind": "file", "value": "src/payoutflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/invoice_5.csv", "invoice_5.csv"], "kind": "file", "value": "data/invoice_5.csv"}, {"aliases...
[{"answers": ["src/payoutflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00482-twin
payoutflow
ctx_compress-s1-00482
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_all double counts carried amounts when the period closes on a weekend. It lives in src/payoutflow/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 pa...
13,489
1,000
31
58
false
synth:5054916401317310413@1
synth
[{"aliases": ["src/payoutflow/store.py", "store.py"], "kind": "file", "value": "src/payoutflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/invoice_5.csv", "invoice_5.csv"], "kind": "file", "value": "data/invoice_5.csv"}, {"aliases...
[{"answers": ["src/payoutflow/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-00483
bucketworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] smooth_gauge leaves a trailing dash when the title ends in punctuation. It lives in src/bucketworks/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 bucketworks layout. [t...
64,916
3,072
106
117
false
synth:136071559000681265@1.24093
synth
[{"aliases": ["src/bucketworks/cli.py", "cli.py"], "kind": "file", "value": "src/bucketworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/bucketworks/tables.py", "tables.py"], "kind": "file", "value": "src/bucketworks/tables.py"}, {"aliases...
[{"answers": ["src/bucketworks/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 r...
ctx_compress-s1-00483-twin
bucketworks
ctx_compress-s1-00483
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 leaves a trailing dash when the title ends in punctuation. It lives in src/bucketworks/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 bucketworks layout. [t...
43,076
2,153
68
117
false
synth:136071559000681265@1.24093
synth
[{"aliases": ["src/bucketworks/cli.py", "cli.py"], "kind": "file", "value": "src/bucketworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/bucketworks/tables.py", "tables.py"], "kind": "file", "value": "src/bucketworks/tables.py"}, {"aliases...
[{"answers": ["src/bucketworks/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 r...
ctx_compress-s1-00484
cuedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_glyph raises KeyError for a missing key instead of returning the default. It lives in src/cuedesk/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 cuedesk layou...
81,923
3,072
104
119
false
synth:8790089901398878591@1.55523
synth
[{"aliases": ["src/cuedesk/store.py", "store.py"], "kind": "file", "value": "src/cuedesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/cuedesk/core.py", "core.py"], "kind": "file", "value": "src/cuedesk/core.py"}, {"aliases": ["src/c...
[{"answers": ["src/cuedesk/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 r...
ctx_compress-s1-00484-twin
cuedesk
ctx_compress-s1-00484
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_glyph raises KeyError for a missing key instead of returning the default. It lives in src/cuedesk/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 cuedesk layou...
41,510
2,075
66
119
false
synth:8790089901398878591@1.55523
synth
[{"aliases": ["src/cuedesk/store.py", "store.py"], "kind": "file", "value": "src/cuedesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/cuedesk/core.py", "core.py"], "kind": "file", "value": "src/cuedesk/core.py"}, {"aliases": ["src/c...
[{"answers": ["src/cuedesk/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 r...
ctx_compress-s1-00485
lotkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse_safe raises WarehouseRangeError for bytes input in strict mode instead of decoding it. It lives in src/lotkit/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 m...
88,569
3,072
106
133
false
synth:2422075698423322391@1.8614
synth
[{"aliases": ["src/lotkit/rules.py", "rules.py"], "kind": "file", "value": "src/lotkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/lotkit/audit.py", "audit.py"], "kind": "file", "value": "src/lotkit/audit.py"}, {"aliases": ["data/p...
[{"answers": ["src/lotkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00485-twin
lotkit
ctx_compress-s1-00485
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse_safe raises WarehouseRangeError for bytes input in strict mode instead of decoding it. It lives in src/lotkit/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 m...
51,729
2,586
62
133
false
synth:2422075698423322391@1.8614
synth
[{"aliases": ["src/lotkit/rules.py", "rules.py"], "kind": "file", "value": "src/lotkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/lotkit/audit.py", "audit.py"], "kind": "file", "value": "src/lotkit/audit.py"}, {"aliases": ["data/p...
[{"answers": ["src/lotkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-00486
indexkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_section_page leaves a trailing dash when the title ends in punctuation. It lives in src/indexkit/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 indexkit layou...
29,658
1,482
54
65
false
synth:4417484890341674348@1
synth
[{"aliases": ["src/indexkit/rules.py", "rules.py"], "kind": "file", "value": "src/indexkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/indexkit/format.py", "format.py"], "kind": "file", "value": "src/indexkit/format.py"}, {"aliases...
[{"answers": ["src/indexkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00486-twin
indexkit
ctx_compress-s1-00486
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_section_page leaves a trailing dash when the title ends in punctuation. It lives in src/indexkit/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 indexkit layou...
12,888
1,000
30
65
false
synth:4417484890341674348@1
synth
[{"aliases": ["src/indexkit/rules.py", "rules.py"], "kind": "file", "value": "src/indexkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/indexkit/format.py", "format.py"], "kind": "file", "value": "src/indexkit/format.py"}, {"aliases...
[{"answers": ["src/indexkit/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-00487
balanceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_balance raises ValueError when a value itself contains an equals sign. It lives in src/balanceforge/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 balanc...
107,992
3,072
109
118
false
synth:2229044283026957202@1.66305
synth
[{"aliases": ["src/balanceforge/report.py", "report.py"], "kind": "file", "value": "src/balanceforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/receipt_4.csv", "receipt_4.csv"], "kind": "file", "value": "data/receipt_4.csv"},...
[{"answers": ["src/balanceforge/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-s1-00487-twin
balanceforge
ctx_compress-s1-00487
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_balance raises ValueError when a value itself contains an equals sign. It lives in src/balanceforge/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 balanc...
71,675
3,072
69
118
false
synth:2229044283026957202@1.66305
synth
[{"aliases": ["src/balanceforge/report.py", "report.py"], "kind": "file", "value": "src/balanceforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/receipt_4.csv", "receipt_4.csv"], "kind": "file", "value": "data/receipt_4.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": ["io.py", "src/balanceforge/io.py"], "k...
ctx_compress-s1-00488
chunkflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_glyph raises StemRangeError for bytes input in strict mode instead of decoding it. It lives in src/chunkflow/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 th...
62,379
3,072
119
128
false
synth:8588053031078607231@1.4303
synth
[{"aliases": ["src/chunkflow/report.py", "report.py"], "kind": "file", "value": "src/chunkflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/slug_4.csv", "slug_4.csv"], "kind": "file", "value": "data/slug_4.csv"}, {"aliases": ["...
[{"answers": ["src/chunkflow/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-00488-twin
chunkflow
ctx_compress-s1-00488
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] truncate_glyph raises StemRangeError for bytes input in strict mode instead of decoding it. It lives in src/chunkflow/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 th...
37,690
1,884
67
128
false
synth:8588053031078607231@1.4303
synth
[{"aliases": ["src/chunkflow/report.py", "report.py"], "kind": "file", "value": "src/chunkflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/slug_4.csv", "slug_4.csv"], "kind": "file", "value": "data/slug_4.csv"}, {"aliases": ["...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["report.py", "src/chunkflow/report.py"...
ctx_compress-s1-00489
counterflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_span raises SampleShapeError for bytes input in strict mode instead of decoding it. It lives in src/counterflow/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 co...
85,183
3,072
110
131
false
synth:693878454732663880@1.70388
synth
[{"aliases": ["src/counterflow/audit.py", "audit.py"], "kind": "file", "value": "src/counterflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/series_3.csv", "series_3.csv"], "kind": "file", "value": "data/series_3.csv"}, {"aliases"...
[{"answers": ["src/counterflow/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-s1-00489-twin
counterflow
ctx_compress-s1-00489
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_span raises SampleShapeError for bytes input in strict mode instead of decoding it. It lives in src/counterflow/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 co...
51,138
2,556
78
131
false
synth:693878454732663880@1.70388
synth
[{"aliases": ["src/counterflow/audit.py", "audit.py"], "kind": "file", "value": "src/counterflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/series_3.csv", "series_3.csv"], "kind": "file", "value": "data/series_3.csv"}, {"aliases"...
[{"answers": ["src/counterflow/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-s1-00490
nonceworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] mint_policy ignores the cap, so late attempts wait for minutes. It lives in src/nonceworks/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 nonceworks layout. [tool_...
54,791
2,739
90
101
false
synth:9144126305870934182@1.21788
synth
[{"aliases": ["src/nonceworks/report.py", "report.py"], "kind": "file", "value": "src/nonceworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/nonceworks/cli.py", "cli.py"], "kind": "file", "value": "src/nonceworks/cli.py"}, {"al...
[{"answers": ["src/nonceworks/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00490-twin
nonceworks
ctx_compress-s1-00490
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] mint_policy ignores the cap, so late attempts wait for minutes. It lives in src/nonceworks/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 nonceworks layout. [tool_...
29,492
1,474
51
101
false
synth:9144126305870934182@1.21788
synth
[{"aliases": ["src/nonceworks/report.py", "report.py"], "kind": "file", "value": "src/nonceworks/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/nonceworks/cli.py", "cli.py"], "kind": "file", "value": "src/nonceworks/cli.py"}, {"al...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["report.py", "src/nonceworks/report.py...
ctx_compress-s1-00491
tracedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_probe misses keys that differ only in case. It lives in src/tracedesk/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 tracedesk layout. [tool_use kind=se...
108,954
3,072
130
139
false
synth:4890928001139799429@2.05475
synth
[{"aliases": ["src/tracedesk/limits.py", "limits.py"], "kind": "file", "value": "src/tracedesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/tracedesk/store.py", "store.py"], "kind": "file", "value": "src/tracedesk/store.py"}, {"...
[{"answers": ["src/tracedesk/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 mo...
ctx_compress-s1-00491-twin
tracedesk
ctx_compress-s1-00491
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_probe misses keys that differ only in case. It lives in src/tracedesk/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 tracedesk layout. [tool_use kind=se...
58,647
2,932
74
139
false
synth:4890928001139799429@2.05475
synth
[{"aliases": ["src/tracedesk/limits.py", "limits.py"], "kind": "file", "value": "src/tracedesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/tracedesk/store.py", "store.py"], "kind": "file", "value": "src/tracedesk/store.py"}, {"...
[{"answers": ["src/tracedesk/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 mo...
ctx_compress-s1-00492
policyflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_scope double counts carried amounts when the period closes on a weekend. It lives in src/policyflow/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 policyflow ...
38,950
1,947
62
71
false
synth:5221273966468220605@1
synth
[{"aliases": ["src/policyflow/rules.py", "rules.py"], "kind": "file", "value": "src/policyflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/policyflow/cli.py", "cli.py"], "kind": "file", "value": "src/policyflow/cli.py"}, {"aliases"...
[{"answers": ["src/policyflow/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mos...
ctx_compress-s1-00492-twin
policyflow
ctx_compress-s1-00492
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_scope double counts carried amounts when the period closes on a weekend. It lives in src/policyflow/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 policyflow ...
20,840
1,042
38
71
false
synth:5221273966468220605@1
synth
[{"aliases": ["src/policyflow/rules.py", "rules.py"], "kind": "file", "value": "src/policyflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/policyflow/cli.py", "cli.py"], "kind": "file", "value": "src/policyflow/cli.py"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/policyflow/tables.py", "tables.py...
ctx_compress-s1-00493
refundkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_payout raises ValueError when a value itself contains an equals sign. It lives in src/refundkit/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 refundkit layout. [...
66,090
3,072
109
120
false
synth:7270712217400669031@1.5321
synth
[{"aliases": ["src/refundkit/core.py", "core.py"], "kind": "file", "value": "src/refundkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/refundkit/rules.py", "rules.py"], "kind": "file", "value": "src/refundkit/rules.py"}, {"aliases": ["...
[{"answers": ["src/refundkit/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most r...
ctx_compress-s1-00493-twin
refundkit
ctx_compress-s1-00493
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_payout raises ValueError when a value itself contains an equals sign. It lives in src/refundkit/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 refundkit layout. [...
36,732
1,836
65
120
false
synth:7270712217400669031@1.5321
synth
[{"aliases": ["src/refundkit/core.py", "core.py"], "kind": "file", "value": "src/refundkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/refundkit/rules.py", "rules.py"], "kind": "file", "value": "src/refundkit/rules.py"}, {"aliases": ["...
[{"answers": ["src/refundkit/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most r...
ctx_compress-s1-00494
zoneworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_corridor_page leaves a trailing dash when the title ends in punctuation. It lives in src/zoneworks/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 zonework...
72,897
3,072
105
112
false
synth:2679003179081621639@1.24162
synth
[{"aliases": ["src/zoneworks/format.py", "format.py"], "kind": "file", "value": "src/zoneworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/zoneworks/schema.py", "schema.py"], "kind": "file", "value": "src/zoneworks/schema.py"},...
[{"answers": ["src/zoneworks/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-00494-twin
zoneworks
ctx_compress-s1-00494
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_corridor_page leaves a trailing dash when the title ends in punctuation. It lives in src/zoneworks/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 zonework...
44,535
2,226
67
112
false
synth:2679003179081621639@1.24162
synth
[{"aliases": ["src/zoneworks/format.py", "format.py"], "kind": "file", "value": "src/zoneworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/zoneworks/schema.py", "schema.py"], "kind": "file", "value": "src/zoneworks/schema.py"},...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["cli.py", "src/zoneworks/cli.py"], "ki...
ctx_compress-s1-00495
bucketledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 double counts carried amounts when the period closes on a weekend. It lives in src/bucketledger/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 bucketled...
46,165
2,308
71
86
false
synth:4037763688968575244@1
synth
[{"aliases": ["src/bucketledger/audit.py", "audit.py"], "kind": "file", "value": "src/bucketledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliase...
[{"answers": ["src/bucketledger/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-00495-twin
bucketledger
ctx_compress-s1-00495
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 double counts carried amounts when the period closes on a weekend. It lives in src/bucketledger/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 bucketled...
24,492
1,224
39
86
false
synth:4037763688968575244@1
synth
[{"aliases": ["src/bucketledger/audit.py", "audit.py"], "kind": "file", "value": "src/bucketledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliase...
[{"answers": ["src/bucketledger/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-00496
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_warehouse keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/stockforge/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the stockforge layo...
80,506
3,072
102
125
false
synth:2877546839913060441@1.67995
synth
[{"aliases": ["src/stockforge/io.py", "io.py"], "kind": "file", "value": "src/stockforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/lot_1.csv", "lot_1.csv"], "kind": "file", "value": "data/lot_1.csv"}, {"aliases": ["data/stock_5.csv", "stock...
[{"answers": ["src/stockforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rece...
ctx_compress-s1-00496-twin
stockforge
ctx_compress-s1-00496
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_warehouse keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/stockforge/io.py; `python -m pytest tests/test_io.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the stockforge layo...
53,433
2,671
66
125
false
synth:2877546839913060441@1.67995
synth
[{"aliases": ["src/stockforge/io.py", "io.py"], "kind": "file", "value": "src/stockforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/lot_1.csv", "lot_1.csv"], "kind": "file", "value": "data/lot_1.csv"}, {"aliases": ["data/stock_5.csv", "stock...
[{"answers": ["src/stockforge/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rece...
ctx_compress-s1-00497
lotflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_shipment_all ignores the cap, so late attempts wait for minutes. It lives in src/lotflow/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 lotflow layout. [tool_use k...
96,752
3,072
127
136
false
synth:5974888732961708325@1.7162
synth
[{"aliases": ["src/lotflow/io.py", "io.py"], "kind": "file", "value": "src/lotflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/stock_2.csv", "stock_2.csv"], "kind": "file", "value": "data/stock_2.csv"}, {"aliases": ["data/stock_1.csv", "stock...
[{"answers": ["src/lotflow/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-s1-00497-twin
lotflow
ctx_compress-s1-00497
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_shipment_all ignores the cap, so late attempts wait for minutes. It lives in src/lotflow/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 lotflow layout. [tool_use k...
51,733
2,586
69
136
false
synth:5974888732961708325@1.7162
synth
[{"aliases": ["src/lotflow/io.py", "io.py"], "kind": "file", "value": "src/lotflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/stock_2.csv", "stock_2.csv"], "kind": "file", "value": "data/stock_2.csv"}, {"aliases": ["data/stock_1.csv", "stock...
[{"answers": ["src/lotflow/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-s1-00498
quotaworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_booking leaves a trailing dash when the title ends in punctuation. It lives in src/quotaworks/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 quotaworks layout...
40,522
2,026
70
77
false
synth:4089253662334105184@1
synth
[{"aliases": ["src/quotaworks/rules.py", "rules.py"], "kind": "file", "value": "src/quotaworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/quotaworks/limits.py", "limits.py"], "kind": "file", "value": "src/quotaworks/limits.py"}, {...
[{"answers": ["src/quotaworks/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-s1-00498-twin
quotaworks
ctx_compress-s1-00498
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] assign_booking leaves a trailing dash when the title ends in punctuation. It lives in src/quotaworks/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 quotaworks layout...
22,319
1,115
38
77
false
synth:4089253662334105184@1
synth
[{"aliases": ["src/quotaworks/rules.py", "rules.py"], "kind": "file", "value": "src/quotaworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/quotaworks/limits.py", "limits.py"], "kind": "file", "value": "src/quotaworks/limits.py"}, {...
[{"answers": ["src/quotaworks/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-s1-00499
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_series_page grows linearly instead of doubling. It lives in src/seriesflow/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 seriesflow layout. [tool_use kind=search nam...
101,719
3,072
108
135
false
synth:2605398583039693750@2.32125
synth
[{"aliases": ["src/seriesflow/io.py", "io.py"], "kind": "file", "value": "src/seriesflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/seriesflow/a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["limits.py", "src/seriesflow/limits.py...
ctx_compress-s1-00499-twin
seriesflow
ctx_compress-s1-00499
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_series_page grows linearly instead of doubling. It lives in src/seriesflow/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 seriesflow layout. [tool_use kind=search nam...
55,894
2,794
65
135
false
synth:2605398583039693750@2.32125
synth
[{"aliases": ["src/seriesflow/io.py", "io.py"], "kind": "file", "value": "src/seriesflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/seriesflow/a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["counter_3.csv", "data/counter_3.csv"]...
ctx_compress-s1-00500
windowforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_booking rejects a range where lo equals hi, which is a valid single point range. It lives in src/windowforge/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 ...
57,552
2,877
102
115
false
synth:1835931890010334344@1.05422
synth
[{"aliases": ["src/windowforge/compat.py", "compat.py"], "kind": "file", "value": "src/windowforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/windowforge/limits.py", "limits.py"], "kind": "file", "value": "src/windowforge/limi...
[{"answers": ["src/windowforge/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-00500-twin
windowforge
ctx_compress-s1-00500
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_booking rejects a range where lo equals hi, which is a valid single point range. It lives in src/windowforge/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 ...
31,211
1,560
52
115
false
synth:1835931890010334344@1.05422
synth
[{"aliases": ["src/windowforge/compat.py", "compat.py"], "kind": "file", "value": "src/windowforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/windowforge/limits.py", "limits.py"], "kind": "file", "value": "src/windowforge/limi...
[{"answers": ["src/windowforge/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-00501
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] dedupe_snippet_batch keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/slugkit/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 slugkit...
44,884
2,244
85
104
false
synth:4238097362611677731@1
synth
[{"aliases": ["src/slugkit/compat.py", "compat.py"], "kind": "file", "value": "src/slugkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/caption_0.csv", "caption_0.csv"], "kind": "file", "value": "data/caption_0.csv"}, {"aliases...
[{"answers": ["src/slugkit/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 most...
ctx_compress-s1-00501-twin
slugkit
ctx_compress-s1-00501
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] dedupe_snippet_batch keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/slugkit/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 slugkit...
28,857
1,442
55
104
false
synth:4238097362611677731@1
synth
[{"aliases": ["src/slugkit/compat.py", "compat.py"], "kind": "file", "value": "src/slugkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/caption_0.csv", "caption_0.csv"], "kind": "file", "value": "data/caption_0.csv"}, {"aliases...
[{"answers": ["src/slugkit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...
ctx_compress-s1-00502
reorderdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_bin_strict raises KeyError for a missing key instead of returning the default. It lives in src/reorderdesk/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 reo...
69,151
3,072
93
116
false
synth:4101509371311465190@1.98182
synth
[{"aliases": ["src/reorderdesk/limits.py", "limits.py"], "kind": "file", "value": "src/reorderdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/lot_5.csv", "lot_5.csv"], "kind": "file", "value": "data/lot_5.csv"}, {"aliases": [...
[{"answers": ["src/reorderdesk/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-00502-twin
reorderdesk
ctx_compress-s1-00502
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] count_bin_strict raises KeyError for a missing key instead of returning the default. It lives in src/reorderdesk/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 reo...
48,172
2,408
59
116
false
synth:4101509371311465190@1.98182
synth
[{"aliases": ["src/reorderdesk/limits.py", "limits.py"], "kind": "file", "value": "src/reorderdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/lot_5.csv", "lot_5.csv"], "kind": "file", "value": "data/lot_5.csv"}, {"aliases": [...
[{"answers": ["src/reorderdesk/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-00503
bearingkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bearing raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/bearingkit/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 bear...
103,421
3,072
126
149
false
synth:6083764731288019384@2.05523
synth
[{"aliases": ["src/bearingkit/core.py", "core.py"], "kind": "file", "value": "src/bearingkit/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/fix_4.csv", "fix_4.csv"], "kind": "file", "value": "data/fix_4.csv"}, {"aliases": ["data/tile_5.c...
[{"answers": ["src/bearingkit/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most ...