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-s1-01507
invoiceledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_invoice lets defaults overwrite explicit overrides. It lives in src/invoiceledger/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 invoiceledger layout. [tool...
36,127
1,806
62
73
false
synth:969294658202589944@1
synth
[{"aliases": ["src/invoiceledger/audit.py", "audit.py"], "kind": "file", "value": "src/invoiceledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/ledger_0.csv", "ledger_0.csv"], "kind": "file", "value": "data/ledger_0.csv"}, {"alia...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["audit.py", "src/invoiceledger/audit.p...
ctx_compress-s1-01507-twin
invoiceledger
ctx_compress-s1-01507
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_invoice lets defaults overwrite explicit overrides. It lives in src/invoiceledger/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 invoiceledger layout. [tool...
21,958
1,097
40
73
false
synth:969294658202589944@1
synth
[{"aliases": ["src/invoiceledger/audit.py", "audit.py"], "kind": "file", "value": "src/invoiceledger/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/ledger_0.csv", "ledger_0.csv"], "kind": "file", "value": "data/ledger_0.csv"}, {"alia...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["rules.py", "src/invoiceledger/rules.p...
ctx_compress-s1-01508
stemflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_stem raises ValueError when a value itself contains an equals sign. It lives in src/stemflow/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 stemflow layout. [too...
89,740
3,072
116
139
false
synth:6986553880555258721@1.66482
synth
[{"aliases": ["src/stemflow/io.py", "io.py"], "kind": "file", "value": "src/stemflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/chunk_9.csv", "chunk_9.csv"], "kind": "file", "value": "data/chunk_9.csv"}, {"aliases": ["data/slug_2.csv", "slug...
[{"answers": ["src/stemflow/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-s1-01508-twin
stemflow
ctx_compress-s1-01508
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_stem raises ValueError when a value itself contains an equals sign. It lives in src/stemflow/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 stemflow layout. [too...
47,013
2,350
68
139
false
synth:6986553880555258721@1.66482
synth
[{"aliases": ["src/stemflow/io.py", "io.py"], "kind": "file", "value": "src/stemflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/chunk_9.csv", "chunk_9.csv"], "kind": "file", "value": "data/chunk_9.csv"}, {"aliases": ["data/slug_2.csv", "slug...
[{"answers": ["src/stemflow/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-s1-01509
bearingflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_tile grows linearly instead of doubling. It lives in src/bearingflow/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bearingflow layout. [tool_use kind=sear...
36,263
1,813
61
78
false
synth:2832014456229569504@1
synth
[{"aliases": ["src/bearingflow/compat.py", "compat.py"], "kind": "file", "value": "src/bearingflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"al...
[{"answers": ["src/bearingflow/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01509-twin
bearingflow
ctx_compress-s1-01509
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_tile grows linearly instead of doubling. It lives in src/bearingflow/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bearingflow layout. [tool_use kind=sear...
23,558
1,177
43
78
false
synth:2832014456229569504@1
synth
[{"aliases": ["src/bearingflow/compat.py", "compat.py"], "kind": "file", "value": "src/bearingflow/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"al...
[{"answers": ["src/bearingflow/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01510
sectionforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_heading skips the first page; page 1 returns what should be page 2. It lives in src/sectionforge/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 sectionforge layout....
93,109
3,072
96
115
false
synth:4441934793515325256@1.83005
synth
[{"aliases": ["src/sectionforge/io.py", "io.py"], "kind": "file", "value": "src/sectionforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/margin_1.csv", "margin_1.csv"], "kind": "file", "value": "data/margin_1.csv"}, {"aliases": ["data/draft_5...
[{"answers": ["src/sectionforge/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 re...
ctx_compress-s1-01510-twin
sectionforge
ctx_compress-s1-01510
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_heading skips the first page; page 1 returns what should be page 2. It lives in src/sectionforge/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 sectionforge layout....
53,603
2,680
64
115
false
synth:4441934793515325256@1.83005
synth
[{"aliases": ["src/sectionforge/io.py", "io.py"], "kind": "file", "value": "src/sectionforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/margin_1.csv", "margin_1.csv"], "kind": "file", "value": "data/margin_1.csv"}, {"aliases": ["data/draft_5...
[{"answers": ["src/sectionforge/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 re...
ctx_compress-s1-01511
probeledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_probe raises BucketStateError for bytes input in strict mode instead of decoding it. It lives in src/probeledger/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...
90,789
3,072
99
124
false
synth:4036797708547815607@2.09585
synth
[{"aliases": ["src/probeledger/format.py", "format.py"], "kind": "file", "value": "src/probeledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/gauge_0.csv", "gauge_0.csv"], "kind": "file", "value": "data/gauge_0.csv"}, {"alias...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["rules.py", "src/probeledger/rules.py"...
ctx_compress-s1-01511-twin
probeledger
ctx_compress-s1-01511
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] bin_probe raises BucketStateError for bytes input in strict mode instead of decoding it. It lives in src/probeledger/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...
54,942
2,747
65
124
false
synth:4036797708547815607@2.09585
synth
[{"aliases": ["src/probeledger/format.py", "format.py"], "kind": "file", "value": "src/probeledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/gauge_0.csv", "gauge_0.csv"], "kind": "file", "value": "data/gauge_0.csv"}, {"alias...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/probeledger/tables.py", "tables.p...
ctx_compress-s1-01512
lotdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_warehouse_once rejects a range where lo equals hi, which is a valid single point range. It lives in src/lotdesk/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...
61,006
3,050
98
117
false
synth:8627816516421554739@2.33252
synth
[{"aliases": ["src/lotdesk/format.py", "format.py"], "kind": "file", "value": "src/lotdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/lotdesk/audit.py", "audit.py"], "kind": "file", "value": "src/lotdesk/audit.py"}, {"aliases"...
[{"answers": ["src/lotdesk/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 most...
ctx_compress-s1-01512-twin
lotdesk
ctx_compress-s1-01512
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_warehouse_once rejects a range where lo equals hi, which is a valid single point range. It lives in src/lotdesk/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...
40,656
2,032
61
117
false
synth:8627816516421554739@2.33252
synth
[{"aliases": ["src/lotdesk/format.py", "format.py"], "kind": "file", "value": "src/lotdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/lotdesk/audit.py", "audit.py"], "kind": "file", "value": "src/lotdesk/audit.py"}, {"aliases"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/stock_2.csv", "stock_2.csv"], "k...
ctx_compress-s1-01513
grantdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_nonce double counts carried amounts when the period closes on a weekend. It lives in src/grantdesk/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 grantdesk la...
22,005
1,100
38
41
false
synth:1260591707118607186@1
synth
[{"aliases": ["src/grantdesk/rules.py", "rules.py"], "kind": "file", "value": "src/grantdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/grantdesk/cli.py", "cli.py"], "kind": "file", "value": "src/grantdesk/cli.py"}, {"aliases": ["...
[{"answers": ["tests/test_rules.py", "test_rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01513-twin
grantdesk
ctx_compress-s1-01513
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_nonce double counts carried amounts when the period closes on a weekend. It lives in src/grantdesk/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 grantdesk la...
14,469
1,000
22
41
false
synth:1260591707118607186@1
synth
[{"aliases": ["src/grantdesk/rules.py", "rules.py"], "kind": "file", "value": "src/grantdesk/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/grantdesk/cli.py", "cli.py"], "kind": "file", "value": "src/grantdesk/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": ["format.py", "src/grantdesk/format.py"...
ctx_compress-s1-01514
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] rate_bucket double counts carried amounts when the period closes on a weekend. It lives in src/samplekit/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 samplekit layout. [...
81,633
3,072
103
128
false
synth:6761863117240998018@1.9332
synth
[{"aliases": ["src/samplekit/io.py", "io.py"], "kind": "file", "value": "src/samplekit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/sample_5.csv", "sample_5.csv"], "kind": "file", "value": "data/sample_5.csv"}, {"aliases": ["src/samplekit/stor...
[{"answers": ["src/samplekit/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recen...
ctx_compress-s1-01514-twin
samplekit
ctx_compress-s1-01514
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_bucket double counts carried amounts when the period closes on a weekend. It lives in src/samplekit/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 samplekit layout. [...
43,401
2,170
61
128
false
synth:6761863117240998018@1.9332
synth
[{"aliases": ["src/samplekit/io.py", "io.py"], "kind": "file", "value": "src/samplekit/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/sample_5.csv", "sample_5.csv"], "kind": "file", "value": "data/sample_5.csv"}, {"aliases": ["src/samplekit/stor...
[{"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/samplekit/store.py", "store.py"],...
ctx_compress-s1-01515
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] verify_claim lets defaults overwrite explicit overrides. It lives in src/sessionforge/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 sessionforge layout. [tool_use...
72,365
3,072
85
106
false
synth:2055872925013128565@1.9638
synth
[{"aliases": ["src/sessionforge/compat.py", "compat.py"], "kind": "file", "value": "src/sessionforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/sessionforge/tables.py", "tables.py"], "kind": "file", "value": "src/sessionforge/...
[{"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-s1-01515-twin
sessionforge
ctx_compress-s1-01515
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_claim lets defaults overwrite explicit overrides. It lives in src/sessionforge/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 sessionforge layout. [tool_use...
47,521
2,376
55
106
false
synth:2055872925013128565@1.9638
synth
[{"aliases": ["src/sessionforge/compat.py", "compat.py"], "kind": "file", "value": "src/sessionforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/sessionforge/tables.py", "tables.py"], "kind": "file", "value": "src/sessionforge/...
[{"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/scope_3.csv", "scope_3.csv"], "k...
ctx_compress-s1-01516
corridorforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_beacon leaves a trailing dash when the title ends in punctuation. It lives in src/corridorforge/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 corridorforge layout...
103,975
3,072
120
151
false
synth:2522438587602931288@2.25378
synth
[{"aliases": ["src/corridorforge/io.py", "io.py"], "kind": "file", "value": "src/corridorforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/corridor_8.csv", "corridor_8.csv"], "kind": "file", "value": "data/corridor_8.csv"}, {"aliases": ["data...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["bearing_5.csv", "data/bearing_5.csv"]...
ctx_compress-s1-01516-twin
corridorforge
ctx_compress-s1-01516
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_beacon leaves a trailing dash when the title ends in punctuation. It lives in src/corridorforge/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 corridorforge layout...
74,433
3,072
84
151
false
synth:2522438587602931288@2.25378
synth
[{"aliases": ["src/corridorforge/io.py", "io.py"], "kind": "file", "value": "src/corridorforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/corridor_8.csv", "corridor_8.csv"], "kind": "file", "value": "data/corridor_8.csv"}, {"aliases": ["data...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["bearing_10.csv", "data/bearing_10.csv...
ctx_compress-s1-01517
palletforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_bin_batch returns one item too few on every page. It lives in src/palletforge/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 palletforge layout. [tool_use kin...
72,861
3,072
110
123
false
synth:7976553571322905249@2.2123
synth
[{"aliases": ["src/palletforge/limits.py", "limits.py"], "kind": "file", "value": "src/palletforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/palletforge/io.py", "io.py"], "kind": "file", "value": "src/palletforge/io.py"}, {"a...
[{"answers": ["src/palletforge/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-s1-01517-twin
palletforge
ctx_compress-s1-01517
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_bin_batch returns one item too few on every page. It lives in src/palletforge/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 palletforge layout. [tool_use kin...
51,258
2,562
73
123
false
synth:7976553571322905249@2.2123
synth
[{"aliases": ["src/palletforge/limits.py", "limits.py"], "kind": "file", "value": "src/palletforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/palletforge/io.py", "io.py"], "kind": "file", "value": "src/palletforge/io.py"}, {"a...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_audit.py", "tests/test_audit.py"...
ctx_compress-s1-01518
clauseforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_footnote raises MarginBoundError for bytes input in strict mode instead of decoding it. It lives in src/clauseforge/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 th...
103,683
3,072
112
125
false
synth:4181852123179993528@2.1635
synth
[{"aliases": ["src/clauseforge/cli.py", "cli.py"], "kind": "file", "value": "src/clauseforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/revision_8.csv", "revision_8.csv"], "kind": "file", "value": "data/revision_8.csv"}, {"aliases": ["da...
[{"answers": ["src/clauseforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most r...
ctx_compress-s1-01518-twin
clauseforge
ctx_compress-s1-01518
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] collate_footnote raises MarginBoundError for bytes input in strict mode instead of decoding it. It lives in src/clauseforge/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 th...
47,757
2,387
58
125
false
synth:4181852123179993528@2.1635
synth
[{"aliases": ["src/clauseforge/cli.py", "cli.py"], "kind": "file", "value": "src/clauseforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/revision_8.csv", "revision_8.csv"], "kind": "file", "value": "data/revision_8.csv"}, {"aliases": ["da...
[{"answers": ["src/clauseforge/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-s1-01519
slotworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_window_strict raises BookingDriftError for bytes input in strict mode instead of decoding it. It lives in src/slotworks/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 map...
19,079
1,000
49
58
false
synth:6811523536305674586@1
synth
[{"aliases": ["src/slotworks/rules.py", "rules.py"], "kind": "file", "value": "src/slotworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slotworks/limits.py", "limits.py"], "kind": "file", "value": "src/slotworks/limits.py"}, {"ali...
[{"answers": ["src/slotworks/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-s1-01519-twin
slotworks
ctx_compress-s1-01519
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window_strict raises BookingDriftError for bytes input in strict mode instead of decoding it. It lives in src/slotworks/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 map...
12,498
1,000
33
58
false
synth:6811523536305674586@1
synth
[{"aliases": ["src/slotworks/rules.py", "rules.py"], "kind": "file", "value": "src/slotworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/slotworks/limits.py", "limits.py"], "kind": "file", "value": "src/slotworks/limits.py"}, {"ali...
[{"answers": ["src/slotworks/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-s1-01520
headingkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_margin raises KeyError for a missing key instead of returning the default. It lives in src/headingkit/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 heading...
39,349
1,967
76
89
false
synth:5441271877556175168@1.16623
synth
[{"aliases": ["src/headingkit/rules.py", "rules.py"], "kind": "file", "value": "src/headingkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/headingkit/cli.py", "cli.py"], "kind": "file", "value": "src/headingkit/cli.py"}, {"aliases"...
[{"answers": ["src/headingkit/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-s1-01520-twin
headingkit
ctx_compress-s1-01520
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] renumber_margin raises KeyError for a missing key instead of returning the default. It lives in src/headingkit/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 heading...
28,257
1,412
50
89
false
synth:5441271877556175168@1.16623
synth
[{"aliases": ["src/headingkit/rules.py", "rules.py"], "kind": "file", "value": "src/headingkit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/headingkit/cli.py", "cli.py"], "kind": "file", "value": "src/headingkit/cli.py"}, {"aliases"...
[{"answers": ["src/headingkit/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-s1-01521
snippetworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_chunk grows linearly instead of doubling. It lives in src/snippetworks/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 snippetworks layout. [tool_use kind=searc...
86,142
3,072
105
132
false
synth:6733137418306570803@1.87512
synth
[{"aliases": ["src/snippetworks/cli.py", "cli.py"], "kind": "file", "value": "src/snippetworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/slug_5.csv", "slug_5.csv"], "kind": "file", "value": "data/slug_5.csv"}, {"aliases": ["data/stem_0....
[{"answers": ["src/snippetworks/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-s1-01521-twin
snippetworks
ctx_compress-s1-01521
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_chunk grows linearly instead of doubling. It lives in src/snippetworks/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 snippetworks layout. [tool_use kind=searc...
52,653
2,632
67
132
false
synth:6733137418306570803@1.87512
synth
[{"aliases": ["src/snippetworks/cli.py", "cli.py"], "kind": "file", "value": "src/snippetworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/slug_5.csv", "slug_5.csv"], "kind": "file", "value": "data/slug_5.csv"}, {"aliases": ["data/stem_0....
[{"answers": ["src/snippetworks/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-s1-01522
stockdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_sku_window skips the first page; page 1 returns what should be page 2. It lives in src/stockdesk/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 stockdesk layout. [t...
84,632
3,072
115
146
false
synth:2417354227113280582@2.17903
synth
[{"aliases": ["src/stockdesk/cli.py", "cli.py"], "kind": "file", "value": "src/stockdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/stockdesk/compat.py", "compat.py"], "kind": "file", "value": "src/stockdesk/compat.py"}, {"aliases": ["dat...
[{"answers": ["src/stockdesk/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-01522-twin
stockdesk
ctx_compress-s1-01522
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_sku_window skips the first page; page 1 returns what should be page 2. It lives in src/stockdesk/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 stockdesk layout. [t...
56,839
2,841
73
146
false
synth:2417354227113280582@2.17903
synth
[{"aliases": ["src/stockdesk/cli.py", "cli.py"], "kind": "file", "value": "src/stockdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/stockdesk/compat.py", "compat.py"], "kind": "file", "value": "src/stockdesk/compat.py"}, {"aliases": ["dat...
[{"answers": ["src/stockdesk/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most rec...
ctx_compress-s1-01523
slotledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window_page leaves a trailing dash when the title ends in punctuation. It lives in src/slotledger/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 slotledger layo...
73,535
3,072
110
129
false
synth:3163514370143836232@2.36218
synth
[{"aliases": ["src/slotledger/core.py", "core.py"], "kind": "file", "value": "src/slotledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/slotledger/tables.py", "tables.py"], "kind": "file", "value": "src/slotledger/tables.py"}, {"alias...
[{"answers": ["src/slotledger/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-s1-01523-twin
slotledger
ctx_compress-s1-01523
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window_page leaves a trailing dash when the title ends in punctuation. It lives in src/slotledger/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 slotledger layo...
48,127
2,406
66
129
false
synth:3163514370143836232@2.36218
synth
[{"aliases": ["src/slotledger/core.py", "core.py"], "kind": "file", "value": "src/slotledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/slotledger/tables.py", "tables.py"], "kind": "file", "value": "src/slotledger/tables.py"}, {"alias...
[{"answers": ["src/slotledger/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-s1-01524
scopeforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim rejects a range where lo equals hi, which is a valid single point range. It lives in src/scopeforge/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 scope...
85,849
3,072
113
124
false
synth:8553294553388684783@2.0248
synth
[{"aliases": ["src/scopeforge/report.py", "report.py"], "kind": "file", "value": "src/scopeforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/scope_4.csv", "scope_4.csv"], "kind": "file", "value": "data/scope_4.csv"}, {"aliases...
[{"answers": ["src/scopeforge/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01524-twin
scopeforge
ctx_compress-s1-01524
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim rejects a range where lo equals hi, which is a valid single point range. It lives in src/scopeforge/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 scope...
48,253
2,412
65
124
false
synth:8553294553388684783@2.0248
synth
[{"aliases": ["src/scopeforge/report.py", "report.py"], "kind": "file", "value": "src/scopeforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/scope_4.csv", "scope_4.csv"], "kind": "file", "value": "data/scope_4.csv"}, {"aliases...
[{"answers": ["src/scopeforge/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-s1-01525
seriesworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] smooth_sample raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/seriesworks/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 seri...
63,196
3,072
86
99
false
synth:428960939929522723@1.3975
synth
[{"aliases": ["src/seriesworks/rules.py", "rules.py"], "kind": "file", "value": "src/seriesworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/span_5.csv", "span_5.csv"], "kind": "file", "value": "data/span_5.csv"}, {"aliases": ["do...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/span_2.csv", "span_2.csv"], "kin...
ctx_compress-s1-01525-twin
seriesworks
ctx_compress-s1-01525
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_sample raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/seriesworks/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 seri...
37,372
1,868
50
99
false
synth:428960939929522723@1.3975
synth
[{"aliases": ["src/seriesworks/rules.py", "rules.py"], "kind": "file", "value": "src/seriesworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/span_5.csv", "span_5.csv"], "kind": "file", "value": "data/span_5.csv"}, {"aliases": ["do...
[{"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-s1-01526
nonceworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_nonce_all grows linearly instead of doubling. It lives in src/nonceworks/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the nonceworks layout. [tool_use kind=se...
54,393
2,719
75
90
false
synth:3887716004324841401@1.44887
synth
[{"aliases": ["src/nonceworks/format.py", "format.py"], "kind": "file", "value": "src/nonceworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/token_2.csv", "token_2.csv"], "kind": "file", "value": "data/token_2.csv"}, {"aliases...
[{"answers": ["src/nonceworks/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01526-twin
nonceworks
ctx_compress-s1-01526
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_nonce_all grows linearly instead of doubling. It lives in src/nonceworks/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the nonceworks layout. [tool_use kind=se...
32,795
1,639
45
90
false
synth:3887716004324841401@1.44887
synth
[{"aliases": ["src/nonceworks/format.py", "format.py"], "kind": "file", "value": "src/nonceworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/token_2.csv", "token_2.csv"], "kind": "file", "value": "data/token_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": ["changelog.md", "docs/changelog.md"], ...
ctx_compress-s1-01527
ledgerkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_refund_safe misses keys that differ only in case. It lives in src/ledgerkit/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 ledgerkit layout. [tool_use ki...
52,289
2,614
95
110
false
synth:3345796469045556746@1.22618
synth
[{"aliases": ["src/ledgerkit/report.py", "report.py"], "kind": "file", "value": "src/ledgerkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/credit_5.csv", "credit_5.csv"], "kind": "file", "value": "data/credit_5.csv"}, {"aliase...
[{"answers": ["src/ledgerkit/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-s1-01527-twin
ledgerkit
ctx_compress-s1-01527
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_refund_safe misses keys that differ only in case. It lives in src/ledgerkit/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 ledgerkit layout. [tool_use ki...
29,960
1,498
51
110
false
synth:3345796469045556746@1.22618
synth
[{"aliases": ["src/ledgerkit/report.py", "report.py"], "kind": "file", "value": "src/ledgerkit/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/credit_5.csv", "credit_5.csv"], "kind": "file", "value": "data/credit_5.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["credit_5.csv", "data/credit_5.csv"], ...
ctx_compress-s1-01528
invoiceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_payout rejects a range where lo equals hi, which is a valid single point range. It lives in src/invoiceforge/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 in...
62,341
3,072
90
115
false
synth:8366617647451066898@1.44755
synth
[{"aliases": ["src/invoiceforge/audit.py", "audit.py"], "kind": "file", "value": "src/invoiceforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/invoiceforge/format.py", "format.py"], "kind": "file", "value": "src/invoiceforge/format...
[{"answers": ["src/invoiceforge/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-s1-01528-twin
invoiceforge
ctx_compress-s1-01528
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_payout rejects a range where lo equals hi, which is a valid single point range. It lives in src/invoiceforge/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 in...
37,417
1,870
53
115
false
synth:8366617647451066898@1.44755
synth
[{"aliases": ["src/invoiceforge/audit.py", "audit.py"], "kind": "file", "value": "src/invoiceforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/invoiceforge/format.py", "format.py"], "kind": "file", "value": "src/invoiceforge/format...
[{"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/invoiceforge/audit.py...
ctx_compress-s1-01529
sessionworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_nonce_once keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/sessionworks/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 session...
53,103
2,655
97
104
false
synth:6454328781543768804@1.47513
synth
[{"aliases": ["src/sessionworks/rules.py", "rules.py"], "kind": "file", "value": "src/sessionworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/sessionworks/schema.py", "schema.py"], "kind": "file", "value": "src/sessionworks/schema...
[{"answers": ["src/sessionworks/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01529-twin
sessionworks
ctx_compress-s1-01529
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_nonce_once keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/sessionworks/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 session...
25,599
1,279
47
104
false
synth:6454328781543768804@1.47513
synth
[{"aliases": ["src/sessionworks/rules.py", "rules.py"], "kind": "file", "value": "src/sessionworks/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/sessionworks/schema.py", "schema.py"], "kind": "file", "value": "src/sessionworks/schema...
[{"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/sessionworks/format....
ctx_compress-s1-01530
principaldesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope_window skips the first page; page 1 returns what should be page 2. It lives in src/principaldesk/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 principaldes...
48,138
2,406
94
111
false
synth:1333751607163831755@1
synth
[{"aliases": ["src/principaldesk/cli.py", "cli.py"], "kind": "file", "value": "src/principaldesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/claim_1.csv", "claim_1.csv"], "kind": "file", "value": "data/claim_1.csv"}, {"aliases": ["src/pri...
[{"answers": ["src/principaldesk/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-s1-01530-twin
principaldesk
ctx_compress-s1-01530
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope_window skips the first page; page 1 returns what should be page 2. It lives in src/principaldesk/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 principaldes...
24,257
1,212
54
111
false
synth:1333751607163831755@1
synth
[{"aliases": ["src/principaldesk/cli.py", "cli.py"], "kind": "file", "value": "src/principaldesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/claim_1.csv", "claim_1.csv"], "kind": "file", "value": "data/claim_1.csv"}, {"aliases": ["src/pri...
[{"answers": ["src/principaldesk/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-s1-01531
beaconflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_zone double counts carried amounts when the period closes on a weekend. It lives in src/beaconflow/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 beaconflo...
53,130
2,656
93
104
false
synth:3250946469834497000@1.26297
synth
[{"aliases": ["src/beaconflow/limits.py", "limits.py"], "kind": "file", "value": "src/beaconflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/beaconflow/audit.py", "audit.py"], "kind": "file", "value": "src/beaconflow/audit.py"}...
[{"answers": ["src/beaconflow/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01531-twin
beaconflow
ctx_compress-s1-01531
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 double counts carried amounts when the period closes on a weekend. It lives in src/beaconflow/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 beaconflo...
30,928
1,546
61
104
false
synth:3250946469834497000@1.26297
synth
[{"aliases": ["src/beaconflow/limits.py", "limits.py"], "kind": "file", "value": "src/beaconflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/beaconflow/audit.py", "audit.py"], "kind": "file", "value": "src/beaconflow/audit.py"}...
[{"answers": ["src/beaconflow/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01532
gaugekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_trace double counts carried amounts when the period closes on a weekend. It lives in src/gaugekit/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 gaugekit ...
70,971
3,072
98
113
false
synth:8583950488032009219@2.05357
synth
[{"aliases": ["src/gaugekit/store.py", "store.py"], "kind": "file", "value": "src/gaugekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/gaugekit/audit.py", "audit.py"], "kind": "file", "value": "src/gaugekit/audit.py"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_cli.py", "tests/test_cli.py"], "...
ctx_compress-s1-01532-twin
gaugekit
ctx_compress-s1-01532
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_trace double counts carried amounts when the period closes on a weekend. It lives in src/gaugekit/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 gaugekit ...
34,996
1,749
53
113
false
synth:8583950488032009219@2.05357
synth
[{"aliases": ["src/gaugekit/store.py", "store.py"], "kind": "file", "value": "src/gaugekit/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/gaugekit/audit.py", "audit.py"], "kind": "file", "value": "src/gaugekit/audit.py"}, {"aliases": ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/gauge_4.csv", "gauge_4.csv"], "k...
ctx_compress-s1-01533
tariffledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_tariff drops override values of 0, False and empty string. It lives in src/tariffledger/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 tariffledger layout. [t...
70,965
3,072
96
113
false
synth:8261801348863368538@1.331
synth
[{"aliases": ["src/tariffledger/rules.py", "rules.py"], "kind": "file", "value": "src/tariffledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/payout_2.csv", "payout_2.csv"], "kind": "file", "value": "data/payout_2.csv"}, {"aliase...
[{"answers": ["src/tariffledger/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01533-twin
tariffledger
ctx_compress-s1-01533
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_tariff drops override values of 0, False and empty string. It lives in src/tariffledger/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 tariffledger layout. [t...
35,545
1,777
54
113
false
synth:8261801348863368538@1.331
synth
[{"aliases": ["src/tariffledger/rules.py", "rules.py"], "kind": "file", "value": "src/tariffledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/payout_2.csv", "payout_2.csv"], "kind": "file", "value": "data/payout_2.csv"}, {"aliase...
[{"answers": ["src/tariffledger/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01534
warehousekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bin_once misses keys that differ only in case. It lives in src/warehousekit/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 warehousekit layout. [tool_use k...
97,258
3,072
121
146
false
synth:4305148116771319580@2.31818
synth
[{"aliases": ["src/warehousekit/limits.py", "limits.py"], "kind": "file", "value": "src/warehousekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/warehousekit/report.py", "report.py"], "kind": "file", "value": "src/warehousekit/...
[{"answers": ["src/warehousekit/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-s1-01534-twin
warehousekit
ctx_compress-s1-01534
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bin_once misses keys that differ only in case. It lives in src/warehousekit/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 warehousekit layout. [tool_use k...
54,011
2,700
67
146
false
synth:4305148116771319580@2.31818
synth
[{"aliases": ["src/warehousekit/limits.py", "limits.py"], "kind": "file", "value": "src/warehousekit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/warehousekit/report.py", "report.py"], "kind": "file", "value": "src/warehousekit/...
[{"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/pallet_4.csv", "pallet_4.csv"], ...
ctx_compress-s1-01535
invoiceforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_refund_strict leaves a trailing dash when the title ends in punctuation. It lives in src/invoiceforge/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 invoicefo...
26,887
1,344
55
66
false
synth:979796509758735815@1
synth
[{"aliases": ["src/invoiceforge/audit.py", "audit.py"], "kind": "file", "value": "src/invoiceforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/credit_0.csv", "credit_0.csv"], "kind": "file", "value": "data/credit_0.csv"}, {"aliase...
[{"answers": ["src/invoiceforge/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-s1-01535-twin
invoiceforge
ctx_compress-s1-01535
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] settle_refund_strict leaves a trailing dash when the title ends in punctuation. It lives in src/invoiceforge/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 invoicefo...
15,998
1,000
33
66
false
synth:979796509758735815@1
synth
[{"aliases": ["src/invoiceforge/audit.py", "audit.py"], "kind": "file", "value": "src/invoiceforge/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/credit_0.csv", "credit_0.csv"], "kind": "file", "value": "data/credit_0.csv"}, {"aliase...
[{"answers": ["src/invoiceforge/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-s1-01536
stockdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reserve_lot_window skips the first page; page 1 returns what should be page 2. It lives in src/stockdesk/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 stockdesk layout....
104,700
3,072
128
139
false
synth:8970427492473883610@2.0363
synth
[{"aliases": ["src/stockdesk/cli.py", "cli.py"], "kind": "file", "value": "src/stockdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/warehouse_8.csv", "warehouse_8.csv"], "kind": "file", "value": "data/warehouse_8.csv"}, {"aliases": ["src...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["cli.py", "src/stockdesk/cli.py"], "ki...
ctx_compress-s1-01536-twin
stockdesk
ctx_compress-s1-01536
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_lot_window skips the first page; page 1 returns what should be page 2. It lives in src/stockdesk/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 stockdesk layout....
59,288
2,964
82
139
false
synth:8970427492473883610@2.0363
synth
[{"aliases": ["src/stockdesk/cli.py", "cli.py"], "kind": "file", "value": "src/stockdesk/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/warehouse_8.csv", "warehouse_8.csv"], "kind": "file", "value": "data/warehouse_8.csv"}, {"aliases": ["src...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/stock_9.csv", "stock_9.csv"], "k...
ctx_compress-s1-01537
binkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_pallet_safe raises LotStateError for bytes input in strict mode instead of decoding it. It lives in src/binkit/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 ...
73,370
3,072
77
84
false
synth:6638216773262743142@1.1394
synth
[{"aliases": ["src/binkit/compat.py", "compat.py"], "kind": "file", "value": "src/binkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/pallet_6.csv", "pallet_6.csv"], "kind": "file", "value": "data/pallet_6.csv"}, {"aliases": ["...
[{"answers": ["src/binkit/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-s1-01537-twin
binkit
ctx_compress-s1-01537
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_pallet_safe raises LotStateError for bytes input in strict mode instead of decoding it. It lives in src/binkit/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 ...
41,557
2,077
49
84
false
synth:6638216773262743142@1.1394
synth
[{"aliases": ["src/binkit/compat.py", "compat.py"], "kind": "file", "value": "src/binkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/pallet_6.csv", "pallet_6.csv"], "kind": "file", "value": "data/pallet_6.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": ["bin_3.csv", "data/bin_3.csv"], "kind"...
ctx_compress-s1-01538
marginkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_section raises ValueError when a value itself contains an equals sign. It lives in src/marginkit/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 marginkit ...
81,577
3,072
83
106
false
synth:3736212754290485889@1.51695
synth
[{"aliases": ["src/marginkit/format.py", "format.py"], "kind": "file", "value": "src/marginkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/margin_1.csv", "margin_1.csv"], "kind": "file", "value": "data/margin_1.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/index_10.csv", "index_10.csv"], ...
ctx_compress-s1-01538-twin
marginkit
ctx_compress-s1-01538
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] paginate_section raises ValueError when a value itself contains an equals sign. It lives in src/marginkit/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 marginkit ...
49,408
2,470
63
106
false
synth:3736212754290485889@1.51695
synth
[{"aliases": ["src/marginkit/format.py", "format.py"], "kind": "file", "value": "src/marginkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/margin_1.csv", "margin_1.csv"], "kind": "file", "value": "data/margin_1.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/margin_1.csv", "margin_1.csv"], ...
ctx_compress-s1-01539
ledgerforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_receipt returns the upper bound for every value instead of clamping into the range. It lives in src/ledgerforge/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...
53,207
2,660
86
95
false
synth:5632791680319045824@1.46855
synth
[{"aliases": ["src/ledgerforge/compat.py", "compat.py"], "kind": "file", "value": "src/ledgerforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/ledgerforge/format.py", "format.py"], "kind": "file", "value": "src/ledgerforge/form...
[{"answers": ["src/ledgerforge/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-s1-01539-twin
ledgerforge
ctx_compress-s1-01539
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 returns the upper bound for every value instead of clamping into the range. It lives in src/ledgerforge/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...
23,368
1,168
43
95
false
synth:5632791680319045824@1.46855
synth
[{"aliases": ["src/ledgerforge/compat.py", "compat.py"], "kind": "file", "value": "src/ledgerforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/ledgerforge/format.py", "format.py"], "kind": "file", "value": "src/ledgerforge/form...
[{"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/ledgerforge/io.py"], "ki...
ctx_compress-s1-01540
gaugeforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_sample_once raises KeyError for a missing key instead of returning the default. It lives in src/gaugeforge/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 gaugeforge...
67,541
3,072
105
114
false
synth:895175546796591494@1.13043
synth
[{"aliases": ["src/gaugeforge/cli.py", "cli.py"], "kind": "file", "value": "src/gaugeforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/gaugeforge/schema.py", "schema.py"], "kind": "file", "value": "src/gaugeforge/schema.py"}, {"aliases": [...
[{"answers": ["src/gaugeforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most re...
ctx_compress-s1-01540-twin
gaugeforge
ctx_compress-s1-01540
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] clip_sample_once raises KeyError for a missing key instead of returning the default. It lives in src/gaugeforge/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 gaugeforge...
32,305
1,615
59
114
false
synth:895175546796591494@1.13043
synth
[{"aliases": ["src/gaugeforge/cli.py", "cli.py"], "kind": "file", "value": "src/gaugeforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/gaugeforge/schema.py", "schema.py"], "kind": "file", "value": "src/gaugeforge/schema.py"}, {"aliases": [...
[{"answers": ["src/gaugeforge/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most re...
ctx_compress-s1-01541
scopeworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_policy double counts carried amounts when the period closes on a weekend. It lives in src/scopeworks/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 scopeworks...
49,648
2,482
74
79
false
synth:7141642307248708638@1
synth
[{"aliases": ["src/scopeworks/audit.py", "audit.py"], "kind": "file", "value": "src/scopeworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/session_3.csv", "session_3.csv"], "kind": "file", "value": "data/session_3.csv"}, {"aliases...
[{"answers": ["src/scopeworks/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01541-twin
scopeworks
ctx_compress-s1-01541
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_policy double counts carried amounts when the period closes on a weekend. It lives in src/scopeworks/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 scopeworks...
32,364
1,618
42
79
false
synth:7141642307248708638@1
synth
[{"aliases": ["src/scopeworks/audit.py", "audit.py"], "kind": "file", "value": "src/scopeworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/session_3.csv", "session_3.csv"], "kind": "file", "value": "data/session_3.csv"}, {"aliases...
[{"answers": ["src/scopeworks/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mos...
ctx_compress-s1-01542
counterforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_sample rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterforge/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 c...
85,113
3,072
97
118
false
synth:5735506220108200959@1.62583
synth
[{"aliases": ["src/counterforge/limits.py", "limits.py"], "kind": "file", "value": "src/counterforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/series_3.csv", "series_3.csv"], "kind": "file", "value": "data/series_3.csv"}, {"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["counter_9.csv", "data/counter_9.csv"]...
ctx_compress-s1-01542-twin
counterforge
ctx_compress-s1-01542
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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/counterforge/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 c...
43,803
2,190
55
118
false
synth:5735506220108200959@1.62583
synth
[{"aliases": ["src/counterforge/limits.py", "limits.py"], "kind": "file", "value": "src/counterforge/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/series_3.csv", "series_3.csv"], "kind": "file", "value": "data/series_3.csv"}, {"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["report.py", "src/counterforge/report....
ctx_compress-s1-01543
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_refund rejects a range where lo equals hi, which is a valid single point range. It lives in src/tariffforge/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 t...
36,734
1,836
46
55
false
synth:5582759779008798757@1
synth
[{"aliases": ["src/tariffforge/compat.py", "compat.py"], "kind": "file", "value": "src/tariffforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/tariffforge/report.py", "report.py"], "kind": "file", "value": "src/tariffforge/repo...
[{"answers": ["src/tariffforge/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-s1-01543-twin
tariffforge
ctx_compress-s1-01543
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_refund rejects a range where lo equals hi, which is a valid single point range. It lives in src/tariffforge/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 t...
14,034
1,000
27
55
false
synth:5582759779008798757@1
synth
[{"aliases": ["src/tariffforge/compat.py", "compat.py"], "kind": "file", "value": "src/tariffforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/tariffforge/report.py", "report.py"], "kind": "file", "value": "src/tariffforge/repo...
[{"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/refund_2.csv", "refund_2.csv"], ...
ctx_compress-s1-01544
balanceworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_balance raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/balanceworks/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 ba...
58,039
2,901
94
117
false
synth:1668235668860885940@2.3599
synth
[{"aliases": ["src/balanceworks/audit.py", "audit.py"], "kind": "file", "value": "src/balanceworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/refund_0.csv", "refund_0.csv"], "kind": "file", "value": "data/refund_0.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/refund_5.csv", "refund_5.csv"], ...
ctx_compress-s1-01544-twin
balanceworks
ctx_compress-s1-01544
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] accrue_balance raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/balanceworks/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 ba...
34,348
1,717
53
117
false
synth:1668235668860885940@2.3599
synth
[{"aliases": ["src/balanceworks/audit.py", "audit.py"], "kind": "file", "value": "src/balanceworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/refund_0.csv", "refund_0.csv"], "kind": "file", "value": "data/refund_0.csv"}, {"aliase...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["core.py", "src/balanceworks/core.py"]...
ctx_compress-s1-01545
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_trace raises TraceShapeError for bytes input in strict mode instead of decoding it. It lives in src/bucketledger/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...
97,029
3,072
94
115
false
synth:5605813428080151845@1.74432
synth
[{"aliases": ["src/bucketledger/store.py", "store.py"], "kind": "file", "value": "src/bucketledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/counter_12.csv", "counter_12.csv"], "kind": "file", "value": "data/counter_12.csv"}, {"...
[{"answers": ["src/bucketledger/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's m...
ctx_compress-s1-01545-twin
bucketledger
ctx_compress-s1-01545
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_trace raises TraceShapeError for bytes input in strict mode instead of decoding it. It lives in src/bucketledger/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...
58,015
2,900
62
115
false
synth:5605813428080151845@1.74432
synth
[{"aliases": ["src/bucketledger/store.py", "store.py"], "kind": "file", "value": "src/bucketledger/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["data/counter_12.csv", "counter_12.csv"], "kind": "file", "value": "data/counter_12.csv"}, {"...
[{"answers": ["src/bucketledger/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01546
receiptkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] net_ledger raises ValueError when a value itself contains an equals sign. It lives in src/receiptkit/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 receiptkit layo...
54,132
2,706
84
105
false
synth:4987950959248650228@1.35462
synth
[{"aliases": ["src/receiptkit/compat.py", "compat.py"], "kind": "file", "value": "src/receiptkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/receiptkit/tables.py", "tables.py"], "kind": "file", "value": "src/receiptkit/tables.p...
[{"answers": ["src/receiptkit/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-s1-01546-twin
receiptkit
ctx_compress-s1-01546
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_ledger raises ValueError when a value itself contains an equals sign. It lives in src/receiptkit/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 receiptkit layo...
39,301
1,965
58
105
false
synth:4987950959248650228@1.35462
synth
[{"aliases": ["src/receiptkit/compat.py", "compat.py"], "kind": "file", "value": "src/receiptkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/receiptkit/tables.py", "tables.py"], "kind": "file", "value": "src/receiptkit/tables.p...
[{"answers": ["src/receiptkit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01547
legledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_tile raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/legledger/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 legle...
72,522
3,072
82
93
false
synth:5233203375366159652@2.06307
synth
[{"aliases": ["src/legledger/format.py", "format.py"], "kind": "file", "value": "src/legledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/leg_2.csv", "leg_2.csv"], "kind": "file", "value": "data/leg_2.csv"}, {"aliases": ["dat...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/legledger/format.py"...
ctx_compress-s1-01547-twin
legledger
ctx_compress-s1-01547
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_tile raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/legledger/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 legle...
34,693
1,734
47
93
false
synth:5233203375366159652@2.06307
synth
[{"aliases": ["src/legledger/format.py", "format.py"], "kind": "file", "value": "src/legledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/leg_2.csv", "leg_2.csv"], "kind": "file", "value": "data/leg_2.csv"}, {"aliases": ["dat...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/leg_2.csv", "leg_2.csv"], "kind"...
ctx_compress-s1-01548
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] hyphenate_chunk raises GlyphBoundError for bytes input in strict mode instead of decoding it. It lives in src/captionledger/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 mapp...
43,810
2,190
75
92
false
synth:8656922656467348502@1
synth
[{"aliases": ["src/captionledger/compat.py", "compat.py"], "kind": "file", "value": "src/captionledger/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"}, ...
[{"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": ["caption_0.csv", "data/caption_0.csv"]...
ctx_compress-s1-01548-twin
captionledger
ctx_compress-s1-01548
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] hyphenate_chunk raises GlyphBoundError for bytes input in strict mode instead of decoding it. It lives in src/captionledger/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 mapp...
29,703
1,485
49
92
false
synth:8656922656467348502@1
synth
[{"aliases": ["src/captionledger/compat.py", "compat.py"], "kind": "file", "value": "src/captionledger/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"}, ...
[{"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-s1-01549
sectiondesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] anchor_section keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/sectiondesk/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 sectiondesk layou...
43,566
2,178
68
85
false
synth:7991851111220463597@1
synth
[{"aliases": ["src/sectiondesk/io.py", "io.py"], "kind": "file", "value": "src/sectiondesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/sectiondesk/limits.py", "limits.py"], "kind": "file", "value": "src/sectiondesk/limits.py"}, {"aliases": ["s...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["schema.py", "src/sectiondesk/schema.p...
ctx_compress-s1-01549-twin
sectiondesk
ctx_compress-s1-01549
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 keeps leading spaces in keys so lookups after a semicolon fail. It lives in src/sectiondesk/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 sectiondesk layou...
31,930
1,596
48
85
false
synth:7991851111220463597@1
synth
[{"aliases": ["src/sectiondesk/io.py", "io.py"], "kind": "file", "value": "src/sectiondesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/sectiondesk/limits.py", "limits.py"], "kind": "file", "value": "src/sectiondesk/limits.py"}, {"aliases": ["s...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/sectiondesk/format.p...
ctx_compress-s1-01550
creditledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] reconcile_payout_page leaves a trailing dash when the title ends in punctuation. It lives in src/creditledger/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 creditledg...
67,767
3,072
105
122
false
synth:7161023347465411480@1.86075
synth
[{"aliases": ["src/creditledger/core.py", "core.py"], "kind": "file", "value": "src/creditledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/creditledger/format.py", "format.py"], "kind": "file", "value": "src/creditledger/format.py"},...
[{"answers": ["src/creditledger/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-s1-01550-twin
creditledger
ctx_compress-s1-01550
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_payout_page leaves a trailing dash when the title ends in punctuation. It lives in src/creditledger/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 creditledg...
40,904
2,045
65
122
false
synth:7161023347465411480@1.86075
synth
[{"aliases": ["src/creditledger/core.py", "core.py"], "kind": "file", "value": "src/creditledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/creditledger/format.py", "format.py"], "kind": "file", "value": "src/creditledger/format.py"},...
[{"answers": ["src/creditledger/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-s1-01551
refundforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_receipt lets defaults overwrite explicit overrides. It lives in src/refundforge/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 refundforge layout. [tool_use kind=se...
53,801
2,690
78
93
false
synth:2442577323885738179@1
synth
[{"aliases": ["src/refundforge/io.py", "io.py"], "kind": "file", "value": "src/refundforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/refundforge/cli.py", "cli.py"], "kind": "file", "value": "src/refundforge/cli.py"}, {"aliases": ["src/refund...
[{"answers": ["src/refundforge/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-s1-01551-twin
refundforge
ctx_compress-s1-01551
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_receipt lets defaults overwrite explicit overrides. It lives in src/refundforge/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 refundforge layout. [tool_use kind=se...
31,513
1,575
50
93
false
synth:2442577323885738179@1
synth
[{"aliases": ["src/refundforge/io.py", "io.py"], "kind": "file", "value": "src/refundforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/refundforge/cli.py", "cli.py"], "kind": "file", "value": "src/refundforge/cli.py"}, {"aliases": ["src/refund...
[{"answers": ["src/refundforge/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-s1-01552
ledgerdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_balance_once returns the upper bound for every value instead of clamping into the range. It lives in src/ledgerdesk/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 map...
27,755
1,387
58
65
false
synth:9192254261746437786@1
synth
[{"aliases": ["src/ledgerdesk/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/credit_3.csv", "credit_3.csv"], "kind": "file", "value": "data/credit_3.csv"}, {"alia...
[{"answers": ["src/ledgerdesk/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01552-twin
ledgerdesk
ctx_compress-s1-01552
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_balance_once returns the upper bound for every value instead of clamping into the range. It lives in src/ledgerdesk/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 map...
14,044
1,000
38
65
false
synth:9192254261746437786@1
synth
[{"aliases": ["src/ledgerdesk/limits.py", "limits.py"], "kind": "file", "value": "src/ledgerdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/credit_3.csv", "credit_3.csv"], "kind": "file", "value": "data/credit_3.csv"}, {"alia...
[{"answers": ["src/ledgerdesk/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's m...
ctx_compress-s1-01553
bearingledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_tile_once drops override values of 0, False and empty string. It lives in src/bearingledger/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 bearingledger l...
99,582
3,072
126
139
false
synth:1940606308235170121@2.0762
synth
[{"aliases": ["src/bearingledger/report.py", "report.py"], "kind": "file", "value": "src/bearingledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/zone_4.csv", "zone_4.csv"], "kind": "file", "value": "data/zone_4.csv"}, {"alia...
[{"answers": ["src/bearingledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent'...
ctx_compress-s1-01553-twin
bearingledger
ctx_compress-s1-01553
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] simplify_tile_once drops override values of 0, False and empty string. It lives in src/bearingledger/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 bearingledger l...
59,866
2,993
80
139
false
synth:1940606308235170121@2.0762
synth
[{"aliases": ["src/bearingledger/report.py", "report.py"], "kind": "file", "value": "src/bearingledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/zone_4.csv", "zone_4.csv"], "kind": "file", "value": "data/zone_4.csv"}, {"alia...
[{"answers": ["src/bearingledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent'...
ctx_compress-s1-01554
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_refund_batch keeps leading spaces in keys so lookups after a semicolon fail. 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 ...
53,718
2,685
95
102
false
synth:6469561423412916046@1.00368
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": ["src/balanceforge/io.py", "io.py"], "kind": "file", "value": "src/balanceforge/io.py"},...
[{"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": ["fail"], "question": "Did the agent's...
ctx_compress-s1-01554-twin
balanceforge
ctx_compress-s1-01554
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_refund_batch keeps leading spaces in keys so lookups after a semicolon fail. 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 ...
31,097
1,554
55
102
false
synth:6469561423412916046@1.00368
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": ["src/balanceforge/io.py", "io.py"], "kind": "file", "value": "src/balanceforge/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": ["credit_3.csv", "data/credit_3.csv"], ...
ctx_compress-s1-01555
stemdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 drops override values of 0, False and empty string. It lives in src/stemdesk/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 stemdesk layout. [tool_use ...
110,569
3,072
117
146
false
synth:5784819671897980883@2.32072
synth
[{"aliases": ["src/stemdesk/audit.py", "audit.py"], "kind": "file", "value": "src/stemdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/stemdesk/compat.py", "compat.py"], "kind": "file", "value": "src/stemdesk/compat.py"}, {"aliases...
[{"answers": ["src/stemdesk/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-s1-01555-twin
stemdesk
ctx_compress-s1-01555
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 drops override values of 0, False and empty string. It lives in src/stemdesk/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 stemdesk layout. [tool_use ...
66,055
3,072
71
146
false
synth:5784819671897980883@2.32072
synth
[{"aliases": ["src/stemdesk/audit.py", "audit.py"], "kind": "file", "value": "src/stemdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/stemdesk/compat.py", "compat.py"], "kind": "file", "value": "src/stemdesk/compat.py"}, {"aliases...
[{"answers": ["src/stemdesk/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-s1-01556
glyphforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cue_once double counts carried amounts when the period closes on a weekend. It lives in src/glyphforge/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 glyp...
83,855
3,072
114
127
false
synth:6776117107001483896@2.25727
synth
[{"aliases": ["src/glyphforge/format.py", "format.py"], "kind": "file", "value": "src/glyphforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/glyphforge/io.py", "io.py"], "kind": "file", "value": "src/glyphforge/io.py"}, {"alias...
[{"answers": ["src/glyphforge/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-s1-01556-twin
glyphforge
ctx_compress-s1-01556
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_cue_once double counts carried amounts when the period closes on a weekend. It lives in src/glyphforge/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 glyp...
47,798
2,389
72
127
false
synth:6776117107001483896@2.25727
synth
[{"aliases": ["src/glyphforge/format.py", "format.py"], "kind": "file", "value": "src/glyphforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/glyphforge/io.py", "io.py"], "kind": "file", "value": "src/glyphforge/io.py"}, {"alias...
[{"answers": ["src/glyphforge/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...