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-s1-01206
waypointdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_corridor returns the upper bound for every value instead of clamping into the range. It lives in src/waypointdesk/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by ma...
54,684
2,734
94
113
false
synth:9208024233941917865@1.14118
synth
[{"aliases": ["src/waypointdesk/format.py", "format.py"], "kind": "file", "value": "src/waypointdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/waypointdesk/schema.py", "schema.py"], "kind": "file", "value": "src/waypointdesk/...
[{"answers": ["src/waypointdesk/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-01206-twin
waypointdesk
ctx_compress-s1-01206
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_corridor returns the upper bound for every value instead of clamping into the range. It lives in src/waypointdesk/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by ma...
39,360
1,968
66
113
false
synth:9208024233941917865@1.14118
synth
[{"aliases": ["src/waypointdesk/format.py", "format.py"], "kind": "file", "value": "src/waypointdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/waypointdesk/schema.py", "schema.py"], "kind": "file", "value": "src/waypointdesk/...
[{"answers": ["src/waypointdesk/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-01207
quotaforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_quota_batch returns the upper bound for every value instead of clamping into the range. It lives in src/quotaforge/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 th...
18,834
1,000
51
56
false
synth:1566393904438160045@1
synth
[{"aliases": ["src/quotaforge/core.py", "core.py"], "kind": "file", "value": "src/quotaforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/quota_2.csv", "quota_2.csv"], "kind": "file", "value": "data/quota_2.csv"}, {"aliases": ["src/quo...
[{"answers": ["src/quotaforge/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-s1-01207-twin
quotaforge
ctx_compress-s1-01207
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_quota_batch returns the upper bound for every value instead of clamping into the range. It lives in src/quotaforge/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 th...
11,556
1,000
29
56
false
synth:1566393904438160045@1
synth
[{"aliases": ["src/quotaforge/core.py", "core.py"], "kind": "file", "value": "src/quotaforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/quota_2.csv", "quota_2.csv"], "kind": "file", "value": "data/quota_2.csv"}, {"aliases": ["src/quo...
[{"answers": ["src/quotaforge/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-s1-01208
nonceledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_token drops override values of 0, False and empty string. It lives in src/nonceledger/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 nonceledger layout. [to...
82,877
3,072
105
116
false
synth:3114098281157820899@1.53475
synth
[{"aliases": ["src/nonceledger/report.py", "report.py"], "kind": "file", "value": "src/nonceledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/policy_3.csv", "policy_3.csv"], "kind": "file", "value": "data/policy_3.csv"}, {"al...
[{"answers": ["src/nonceledger/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-s1-01208-twin
nonceledger
ctx_compress-s1-01208
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_token drops override values of 0, False and empty string. It lives in src/nonceledger/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 nonceledger layout. [to...
46,821
2,341
67
116
false
synth:3114098281157820899@1.53475
synth
[{"aliases": ["src/nonceledger/report.py", "report.py"], "kind": "file", "value": "src/nonceledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/policy_3.csv", "policy_3.csv"], "kind": "file", "value": "data/policy_3.csv"}, {"al...
[{"answers": ["src/nonceledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01209
stockforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_bin averages one value too many once the window is full. It lives in src/stockforge/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 stockforge layout. [tool_us...
61,091
3,054
105
112
false
synth:3616669608576381137@1
synth
[{"aliases": ["src/stockforge/report.py", "report.py"], "kind": "file", "value": "src/stockforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/pallet_0.csv", "pallet_0.csv"], "kind": "file", "value": "data/pallet_0.csv"}, {"alia...
[{"answers": ["src/stockforge/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-s1-01209-twin
stockforge
ctx_compress-s1-01209
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 averages one value too many once the window is full. It lives in src/stockforge/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 stockforge layout. [tool_us...
26,150
1,307
51
112
false
synth:3616669608576381137@1
synth
[{"aliases": ["src/stockforge/report.py", "report.py"], "kind": "file", "value": "src/stockforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["data/pallet_0.csv", "pallet_0.csv"], "kind": "file", "value": "data/pallet_0.csv"}, {"alia...
[{"answers": ["src/stockforge/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-s1-01210
reorderflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_shipment_safe drops override values of 0, False and empty string. It lives in src/reorderflow/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 reorderflow layout. ...
85,506
3,072
122
133
false
synth:2567166845970427856@2.29685
synth
[{"aliases": ["src/reorderflow/store.py", "store.py"], "kind": "file", "value": "src/reorderflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/reorderflow/audit.py", "audit.py"], "kind": "file", "value": "src/reorderflow/audit.py"}, ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/reorderflow/store.py", "store.py"...
ctx_compress-s1-01210-twin
reorderflow
ctx_compress-s1-01210
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_shipment_safe drops override values of 0, False and empty string. It lives in src/reorderflow/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 reorderflow layout. ...
44,640
2,232
67
133
false
synth:2567166845970427856@2.29685
synth
[{"aliases": ["src/reorderflow/store.py", "store.py"], "kind": "file", "value": "src/reorderflow/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/reorderflow/audit.py", "audit.py"], "kind": "file", "value": "src/reorderflow/audit.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": ["audit.py", "src/reorderflow/audit.py"...
ctx_compress-s1-01211
windowkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_slot rejects a range where lo equals hi, which is a valid single point range. It lives in src/windowkit/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 windo...
54,279
2,713
96
105
false
synth:4763143723871541156@1.32197
synth
[{"aliases": ["src/windowkit/compat.py", "compat.py"], "kind": "file", "value": "src/windowkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/windowkit/core.py", "core.py"], "kind": "file", "value": "src/windowkit/core.py"}, {"ali...
[{"answers": ["src/windowkit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01211-twin
windowkit
ctx_compress-s1-01211
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_slot rejects a range where lo equals hi, which is a valid single point range. It lives in src/windowkit/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 windo...
30,989
1,549
56
105
false
synth:4763143723871541156@1.32197
synth
[{"aliases": ["src/windowkit/compat.py", "compat.py"], "kind": "file", "value": "src/windowkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/windowkit/core.py", "core.py"], "kind": "file", "value": "src/windowkit/core.py"}, {"ali...
[{"answers": ["src/windowkit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01212
principalforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_nonce_strict returns one item too few on every page. It lives in src/principalforge/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 principalforge layout. [tool_us...
69,855
3,072
102
115
false
synth:9033908301367347546@1.64485
synth
[{"aliases": ["src/principalforge/cli.py", "cli.py"], "kind": "file", "value": "src/principalforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/principalforge/store.py", "store.py"], "kind": "file", "value": "src/principalforge/store.py"}, ...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_core.py", "tests/test_core.py"],...
ctx_compress-s1-01212-twin
principalforge
ctx_compress-s1-01212
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] revoke_nonce_strict returns one item too few on every page. It lives in src/principalforge/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 principalforge layout. [tool_us...
35,720
1,786
61
115
false
synth:9033908301367347546@1.64485
synth
[{"aliases": ["src/principalforge/cli.py", "cli.py"], "kind": "file", "value": "src/principalforge/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/principalforge/store.py", "store.py"], "kind": "file", "value": "src/principalforge/store.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": ["audit.py", "src/principalforge/audit....
ctx_compress-s1-01213
receiptdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_invoice_safe skips the first page; page 1 returns what should be page 2. It lives in src/receiptdesk/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 receiptdes...
47,861
2,393
92
105
false
synth:4225536497989607818@1.18955
synth
[{"aliases": ["src/receiptdesk/limits.py", "limits.py"], "kind": "file", "value": "src/receiptdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/receiptdesk/tables.py", "tables.py"], "kind": "file", "value": "src/receiptdesk/tabl...
[{"answers": ["src/receiptdesk/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-s1-01213-twin
receiptdesk
ctx_compress-s1-01213
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] roll_invoice_safe skips the first page; page 1 returns what should be page 2. It lives in src/receiptdesk/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 receiptdes...
31,685
1,584
62
105
false
synth:4225536497989607818@1.18955
synth
[{"aliases": ["src/receiptdesk/limits.py", "limits.py"], "kind": "file", "value": "src/receiptdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/receiptdesk/tables.py", "tables.py"], "kind": "file", "value": "src/receiptdesk/tabl...
[{"answers": ["src/receiptdesk/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01214
counterkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterkit/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping th...
52,928
2,646
87
94
false
synth:5730295525971135110@1.07292
synth
[{"aliases": ["src/counterkit/compat.py", "compat.py"], "kind": "file", "value": "src/counterkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/counterkit/schema.py", "schema.py"], "kind": "file", "value": "src/counterkit/schema.p...
[{"answers": ["src/counterkit/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-s1-01214-twin
counterkit
ctx_compress-s1-01214
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] downsample_series rejects a range where lo equals hi, which is a valid single point range. It lives in src/counterkit/compat.py; `python -m pytest tests/test_compat.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping th...
28,219
1,410
45
94
false
synth:5730295525971135110@1.07292
synth
[{"aliases": ["src/counterkit/compat.py", "compat.py"], "kind": "file", "value": "src/counterkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/counterkit/schema.py", "schema.py"], "kind": "file", "value": "src/counterkit/schema.p...
[{"answers": ["src/counterkit/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-s1-01215
headingworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_draft double counts carried amounts when the period closes on a weekend. It lives in src/headingworks/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 headingwork...
70,625
3,072
116
125
false
synth:6480902580335609489@1.9892
synth
[{"aliases": ["src/headingworks/audit.py", "audit.py"], "kind": "file", "value": "src/headingworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/heading_3.csv", "heading_3.csv"], "kind": "file", "value": "data/heading_3.csv"}, {"ali...
[{"answers": ["src/headingworks/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-s1-01215-twin
headingworks
ctx_compress-s1-01215
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_draft double counts carried amounts when the period closes on a weekend. It lives in src/headingworks/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 headingwork...
41,302
2,065
65
125
false
synth:6480902580335609489@1.9892
synth
[{"aliases": ["src/headingworks/audit.py", "audit.py"], "kind": "file", "value": "src/headingworks/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/heading_3.csv", "heading_3.csv"], "kind": "file", "value": "data/heading_3.csv"}, {"ali...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/headingworks/format....
ctx_compress-s1-01216
grantledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_claim rejects a range where lo equals hi, which is a valid single point range. It lives in src/grantledger/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 gr...
94,994
3,072
104
127
false
synth:3473582082093686994@2.03437
synth
[{"aliases": ["src/grantledger/limits.py", "limits.py"], "kind": "file", "value": "src/grantledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/claim_4.csv", "claim_4.csv"], "kind": "file", "value": "data/claim_4.csv"}, {"alias...
[{"answers": ["src/grantledger/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-s1-01216-twin
grantledger
ctx_compress-s1-01216
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_claim rejects a range where lo equals hi, which is a valid single point range. It lives in src/grantledger/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 gr...
54,902
2,745
58
127
false
synth:3473582082093686994@2.03437
synth
[{"aliases": ["src/grantledger/limits.py", "limits.py"], "kind": "file", "value": "src/grantledger/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/claim_4.csv", "claim_4.csv"], "kind": "file", "value": "data/claim_4.csv"}, {"alias...
[{"answers": ["src/grantledger/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01217
grantworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope averages one value too many once the window is full. It lives in src/grantworks/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 grantworks layout. [tool_use ki...
74,589
3,072
109
118
false
synth:8265126342853545282@1.54895
synth
[{"aliases": ["src/grantworks/cli.py", "cli.py"], "kind": "file", "value": "src/grantworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/nonce_0.csv", "nonce_0.csv"], "kind": "file", "value": "data/nonce_0.csv"}, {"aliases": ["data/scope_6....
[{"answers": ["src/grantworks/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-s1-01217-twin
grantworks
ctx_compress-s1-01217
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope averages one value too many once the window is full. It lives in src/grantworks/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 grantworks layout. [tool_use ki...
39,313
1,965
55
118
false
synth:8265126342853545282@1.54895
synth
[{"aliases": ["src/grantworks/cli.py", "cli.py"], "kind": "file", "value": "src/grantworks/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["data/nonce_0.csv", "nonce_0.csv"], "kind": "file", "value": "data/nonce_0.csv"}, {"aliases": ["data/scope_6....
[{"answers": ["src/grantworks/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-s1-01218
policykit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_grant grows linearly instead of doubling. It lives in src/policykit/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 policykit layout. [tool_use kind=search nam...
17,312
1,000
40
51
false
synth:3762900100436063692@1
synth
[{"aliases": ["src/policykit/rules.py", "rules.py"], "kind": "file", "value": "src/policykit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/policy_3.csv", "policy_3.csv"], "kind": "file", "value": "data/policy_3.csv"}, {"aliases": ["...
[{"answers": ["src/policykit/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-s1-01218-twin
policykit
ctx_compress-s1-01218
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_grant grows linearly instead of doubling. It lives in src/policykit/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 policykit layout. [tool_use kind=search nam...
7,252
1,000
25
51
false
synth:3762900100436063692@1
synth
[{"aliases": ["src/policykit/rules.py", "rules.py"], "kind": "file", "value": "src/policykit/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/policy_3.csv", "policy_3.csv"], "kind": "file", "value": "data/policy_3.csv"}, {"aliases": ["...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["src/policykit/tables.py", "tables.py"...
ctx_compress-s1-01219
sessionflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope_strict raises PolicyStateError for bytes input in strict mode instead of decoding it. It lives in src/sessionflow/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappi...
117,551
3,072
133
152
false
synth:1626964301481443851@2.31715
synth
[{"aliases": ["src/sessionflow/rules.py", "rules.py"], "kind": "file", "value": "src/sessionflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/session_0.csv", "session_0.csv"], "kind": "file", "value": "data/session_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": ["report.py", "src/sessionflow/report.p...
ctx_compress-s1-01219-twin
sessionflow
ctx_compress-s1-01219
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_scope_strict raises PolicyStateError for bytes input in strict mode instead of decoding it. It lives in src/sessionflow/rules.py; `python -m pytest tests/test_rules.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mappi...
60,207
3,010
79
152
false
synth:1626964301481443851@2.31715
synth
[{"aliases": ["src/sessionflow/rules.py", "rules.py"], "kind": "file", "value": "src/sessionflow/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["data/session_0.csv", "session_0.csv"], "kind": "file", "value": "data/session_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": ["data/session_0.csv", "session_0.csv"]...
ctx_compress-s1-01220
payoutdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] allocate_balance_batch rejects a range where lo equals hi, which is a valid single point range. It lives in src/payoutdesk/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 mappi...
36,560
1,828
64
73
false
synth:7995020126960896519@1
synth
[{"aliases": ["src/payoutdesk/format.py", "format.py"], "kind": "file", "value": "src/payoutdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/ledger_3.csv", "ledger_3.csv"], "kind": "file", "value": "data/ledger_3.csv"}, {"alia...
[{"answers": ["src/payoutdesk/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-s1-01220-twin
payoutdesk
ctx_compress-s1-01220
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_batch rejects a range where lo equals hi, which is a valid single point range. It lives in src/payoutdesk/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 mappi...
17,899
1,000
38
73
false
synth:7995020126960896519@1
synth
[{"aliases": ["src/payoutdesk/format.py", "format.py"], "kind": "file", "value": "src/payoutdesk/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/ledger_3.csv", "ledger_3.csv"], "kind": "file", "value": "data/ledger_3.csv"}, {"alia...
[{"answers": ["src/payoutdesk/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-s1-01221
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] interpolate_corridor_all rejects a range where lo equals hi, which is a valid single point range. It lives in src/legledger/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 mappin...
71,674
3,072
93
116
false
synth:5583787289339336028@1.70625
synth
[{"aliases": ["src/legledger/rules.py", "rules.py"], "kind": "file", "value": "src/legledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/legledger/format.py", "format.py"], "kind": "file", "value": "src/legledger/format.py"}, {"ali...
[{"answers": ["src/legledger/rules.py", "rules.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most...
ctx_compress-s1-01221-twin
legledger
ctx_compress-s1-01221
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] interpolate_corridor_all rejects a range where lo equals hi, which is a valid single point range. It lives in src/legledger/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 mappin...
46,109
2,305
65
116
false
synth:5583787289339336028@1.70625
synth
[{"aliases": ["src/legledger/rules.py", "rules.py"], "kind": "file", "value": "src/legledger/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/legledger/format.py", "format.py"], "kind": "file", "value": "src/legledger/format.py"}, {"ali...
[{"answers": ["src/legledger/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-s1-01222
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_receipt raises ValueError when a value itself contains an equals sign. It lives in src/tariffledger/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 tariffled...
79,122
3,072
97
124
false
synth:8680761330472713946@1.731
synth
[{"aliases": ["src/tariffledger/format.py", "format.py"], "kind": "file", "value": "src/tariffledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/tariffledger/tables.py", "tables.py"], "kind": "file", "value": "src/tariffledger/...
[{"answers": ["src/tariffledger/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-01222-twin
tariffledger
ctx_compress-s1-01222
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_receipt raises ValueError when a value itself contains an equals sign. It lives in src/tariffledger/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 tariffled...
61,586
3,072
71
124
false
synth:8680761330472713946@1.731
synth
[{"aliases": ["src/tariffledger/format.py", "format.py"], "kind": "file", "value": "src/tariffledger/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/tariffledger/tables.py", "tables.py"], "kind": "file", "value": "src/tariffledger/...
[{"answers": ["src/tariffledger/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-01223
footnoteworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_clause_strict skips the first page; page 1 returns what should be page 2. It lives in src/footnoteworks/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 footnote...
70,133
3,072
72
85
false
synth:597523453263161058@1.09205
synth
[{"aliases": ["src/footnoteworks/core.py", "core.py"], "kind": "file", "value": "src/footnoteworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/revision_2.csv", "revision_2.csv"], "kind": "file", "value": "data/revision_2.csv"}, {"alia...
[{"answers": ["src/footnoteworks/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01223-twin
footnoteworks
ctx_compress-s1-01223
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_clause_strict skips the first page; page 1 returns what should be page 2. It lives in src/footnoteworks/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 footnote...
31,606
1,580
46
85
false
synth:597523453263161058@1.09205
synth
[{"aliases": ["src/footnoteworks/core.py", "core.py"], "kind": "file", "value": "src/footnoteworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/revision_2.csv", "revision_2.csv"], "kind": "file", "value": "data/revision_2.csv"}, {"alia...
[{"answers": ["src/footnoteworks/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01224
slugkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] dedupe_caption_safe averages one value too many once the window is full. It lives in src/slugkit/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 slugkit layout. [to...
62,263
3,072
102
111
false
synth:6483429072162820278@1.07745
synth
[{"aliases": ["src/slugkit/limits.py", "limits.py"], "kind": "file", "value": "src/slugkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/slugkit/rules.py", "rules.py"], "kind": "file", "value": "src/slugkit/rules.py"}, {"aliases"...
[{"answers": ["src/slugkit/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 most...
ctx_compress-s1-01224-twin
slugkit
ctx_compress-s1-01224
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] dedupe_caption_safe averages one value too many once the window is full. It lives in src/slugkit/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 slugkit layout. [to...
34,716
1,735
60
111
false
synth:6483429072162820278@1.07745
synth
[{"aliases": ["src/slugkit/limits.py", "limits.py"], "kind": "file", "value": "src/slugkit/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/slugkit/rules.py", "rules.py"], "kind": "file", "value": "src/slugkit/rules.py"}, {"aliases"...
[{"answers": ["src/slugkit/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 most...
ctx_compress-s1-01225
windowforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_window_once skips the first page; page 1 returns what should be page 2. It lives in src/windowforge/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 windowforge...
19,430
1,000
42
53
false
synth:8082420185518075683@1
synth
[{"aliases": ["src/windowforge/rules.py", "rules.py"], "kind": "file", "value": "src/windowforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/windowforge/io.py", "io.py"], "kind": "file", "value": "src/windowforge/io.py"}, {"aliases...
[{"answers": ["src/windowforge/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 mo...
ctx_compress-s1-01225-twin
windowforge
ctx_compress-s1-01225
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_window_once skips the first page; page 1 returns what should be page 2. It lives in src/windowforge/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 windowforge...
8,011
1,000
25
53
false
synth:8082420185518075683@1
synth
[{"aliases": ["src/windowforge/rules.py", "rules.py"], "kind": "file", "value": "src/windowforge/rules.py"}, {"aliases": ["tests/test_rules.py", "test_rules.py"], "kind": "file", "value": "tests/test_rules.py"}, {"aliases": ["src/windowforge/io.py", "io.py"], "kind": "file", "value": "src/windowforge/io.py"}, {"aliases...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["core.py", "src/windowforge/core.py"],...
ctx_compress-s1-01226
warehouseworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bin skips the first page; page 1 returns what should be page 2. It lives in src/warehouseworks/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 warehousework...
64,930
3,072
91
100
false
synth:7132300372448815857@1.7569
synth
[{"aliases": ["src/warehouseworks/compat.py", "compat.py"], "kind": "file", "value": "src/warehouseworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/warehouse_0.csv", "warehouse_0.csv"], "kind": "file", "value": "data/warehous...
[{"answers": ["src/warehouseworks/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...
ctx_compress-s1-01226-twin
warehouseworks
ctx_compress-s1-01226
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_bin skips the first page; page 1 returns what should be page 2. It lives in src/warehouseworks/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 warehousework...
40,366
2,018
57
100
false
synth:7132300372448815857@1.7569
synth
[{"aliases": ["src/warehouseworks/compat.py", "compat.py"], "kind": "file", "value": "src/warehouseworks/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/warehouse_0.csv", "warehouse_0.csv"], "kind": "file", "value": "data/warehous...
[{"answers": ["src/warehouseworks/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...
ctx_compress-s1-01227
samplekit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] merge_counter_batch averages one value too many once the window is full. 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. [tool_u...
54,966
2,748
107
118
false
synth:8211741098049362846@1.00103
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": ["src/samplekit/rules.py", "rules.py"], "kind": "file", "value": "src/samplekit/rules.py"}, {"aliases": ["data/trace_1...
[{"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": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-01227-twin
samplekit
ctx_compress-s1-01227
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_counter_batch averages one value too many once the window is full. 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. [tool_u...
24,130
1,206
59
118
false
synth:8211741098049362846@1.00103
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": ["src/samplekit/rules.py", "rules.py"], "kind": "file", "value": "src/samplekit/rules.py"}, {"aliases": ["data/trace_1...
[{"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": ["fail"], "question": "Did the agent's most recen...
ctx_compress-s1-01228
principalkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_claim_safe misses keys that differ only in case. It lives in src/principalkit/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 principalkit layout. [tool_use ...
32,874
1,643
52
59
false
synth:4902613451288924444@1
synth
[{"aliases": ["src/principalkit/compat.py", "compat.py"], "kind": "file", "value": "src/principalkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/token_3.csv", "token_3.csv"], "kind": "file", "value": "data/token_3.csv"}, {"ali...
[{"answers": ["src/principalkit/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-01228-twin
principalkit
ctx_compress-s1-01228
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_safe misses keys that differ only in case. It lives in src/principalkit/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 principalkit layout. [tool_use ...
11,119
1,000
29
59
false
synth:4902613451288924444@1
synth
[{"aliases": ["src/principalkit/compat.py", "compat.py"], "kind": "file", "value": "src/principalkit/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/token_3.csv", "token_3.csv"], "kind": "file", "value": "data/token_3.csv"}, {"ali...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["rules.py", "src/principalkit/rules.py...
ctx_compress-s1-01229
claimforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] verify_claim leaves a trailing dash when the title ends in punctuation. It lives in src/claimforge/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 claimforge layout. [t...
44,591
2,229
75
90
false
synth:5747368801726769527@1
synth
[{"aliases": ["src/claimforge/core.py", "core.py"], "kind": "file", "value": "src/claimforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/claim_3.csv", "claim_3.csv"], "kind": "file", "value": "data/claim_3.csv"}, {"aliases": ["src/cla...
[{"answers": ["src/claimforge/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-s1-01229-twin
claimforge
ctx_compress-s1-01229
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 leaves a trailing dash when the title ends in punctuation. It lives in src/claimforge/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 claimforge layout. [t...
21,928
1,096
43
90
false
synth:5747368801726769527@1
synth
[{"aliases": ["src/claimforge/core.py", "core.py"], "kind": "file", "value": "src/claimforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/claim_3.csv", "claim_3.csv"], "kind": "file", "value": "data/claim_3.csv"}, {"aliases": ["src/cla...
[{"answers": ["src/claimforge/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-s1-01230
policydesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim raises KeyError for a missing key instead of returning the default. It lives in src/policydesk/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 policydesk l...
39,491
1,974
64
71
false
synth:3255296762552511556@1
synth
[{"aliases": ["src/policydesk/core.py", "core.py"], "kind": "file", "value": "src/policydesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["data...
[{"answers": ["src/policydesk/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-s1-01230-twin
policydesk
ctx_compress-s1-01230
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim raises KeyError for a missing key instead of returning the default. It lives in src/policydesk/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 policydesk l...
20,511
1,025
36
71
false
synth:3255296762552511556@1
synth
[{"aliases": ["src/policydesk/core.py", "core.py"], "kind": "file", "value": "src/policydesk/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["data...
[{"answers": ["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/policydesk/cli.py"], "k...
ctx_compress-s1-01231
shiftworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_shift leaves a trailing dash when the title ends in punctuation. It lives in src/shiftworks/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 shiftworks layout. [...
87,068
3,072
112
135
false
synth:4589410479889010039@1.73845
synth
[{"aliases": ["src/shiftworks/core.py", "core.py"], "kind": "file", "value": "src/shiftworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/roster_10.csv", "roster_10.csv"], "kind": "file", "value": "data/roster_10.csv"}, {"aliases": ["s...
[{"answers": ["src/shiftworks/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-s1-01231-twin
shiftworks
ctx_compress-s1-01231
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_shift leaves a trailing dash when the title ends in punctuation. It lives in src/shiftworks/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 shiftworks layout. [...
44,629
2,231
68
135
false
synth:4589410479889010039@1.73845
synth
[{"aliases": ["src/shiftworks/core.py", "core.py"], "kind": "file", "value": "src/shiftworks/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/roster_10.csv", "roster_10.csv"], "kind": "file", "value": "data/roster_10.csv"}, {"aliases": ["s...
[{"answers": ["src/shiftworks/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-s1-01232
stemforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_lexeme returns one item too few on every page. It lives in src/stemforge/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 stemforge layout. [tool_use kind=searc...
67,411
3,072
95
102
false
synth:2111838316917230156@1.05645
synth
[{"aliases": ["src/stemforge/report.py", "report.py"], "kind": "file", "value": "src/stemforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/stemforge/core.py", "core.py"], "kind": "file", "value": "src/stemforge/core.py"}, {"ali...
[{"answers": ["src/stemforge/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01232-twin
stemforge
ctx_compress-s1-01232
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] wrap_lexeme returns one item too few on every page. It lives in src/stemforge/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 stemforge layout. [tool_use kind=searc...
30,209
1,510
51
102
false
synth:2111838316917230156@1.05645
synth
[{"aliases": ["src/stemforge/report.py", "report.py"], "kind": "file", "value": "src/stemforge/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/stemforge/core.py", "core.py"], "kind": "file", "value": "src/stemforge/core.py"}, {"ali...
[{"answers": ["src/stemforge/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01233
clauseledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_draft raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/clauseledger/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 cl...
51,309
2,565
84
103
false
synth:144621816652069812@1.10967
synth
[{"aliases": ["src/clauseledger/compat.py", "compat.py"], "kind": "file", "value": "src/clauseledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/clauseledger/limits.py", "limits.py"], "kind": "file", "value": "src/clauseledger/...
[{"answers": ["src/clauseledger/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-s1-01233-twin
clauseledger
ctx_compress-s1-01233
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] redact_draft raises ZeroDivisionError on the first value and lags by one afterwards. It lives in src/clauseledger/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 cl...
32,155
1,607
58
103
false
synth:144621816652069812@1.10967
synth
[{"aliases": ["src/clauseledger/compat.py", "compat.py"], "kind": "file", "value": "src/clauseledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/clauseledger/limits.py", "limits.py"], "kind": "file", "value": "src/clauseledger/...
[{"answers": ["src/clauseledger/compat.py", "compat.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-01234
corridorledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_zone raises KeyError for a missing key instead of returning the default. It lives in src/corridorledger/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 corr...
48,053
2,402
80
99
false
synth:5644402683459708919@1.2042
synth
[{"aliases": ["src/corridorledger/report.py", "report.py"], "kind": "file", "value": "src/corridorledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/corridorledger/store.py", "store.py"], "kind": "file", "value": "src/corridorl...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["corridor_4.csv", "data/corridor_4.csv...
ctx_compress-s1-01234-twin
corridorledger
ctx_compress-s1-01234
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] project_zone raises KeyError for a missing key instead of returning the default. It lives in src/corridorledger/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 corr...
30,033
1,501
52
99
false
synth:5644402683459708919@1.2042
synth
[{"aliases": ["src/corridorledger/report.py", "report.py"], "kind": "file", "value": "src/corridorledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/corridorledger/store.py", "store.py"], "kind": "file", "value": "src/corridorl...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["compat.py", "src/corridorledger/compa...
ctx_compress-s1-01235
scopeflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_token leaves a trailing dash when the title ends in punctuation. It lives in src/scopeflow/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 scopeflow layout. [tool_use ...
46,825
2,341
84
107
false
synth:8609507399783938422@1.07232
synth
[{"aliases": ["src/scopeflow/io.py", "io.py"], "kind": "file", "value": "src/scopeflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/scopeflow/report.py", "report.py"], "kind": "file", "value": "src/scopeflow/report.py"}, {"aliases": ["src/scope...
[{"answers": ["src/scopeflow/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-s1-01235-twin
scopeflow
ctx_compress-s1-01235
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_token leaves a trailing dash when the title ends in punctuation. It lives in src/scopeflow/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 scopeflow layout. [tool_use ...
39,217
1,960
64
107
false
synth:8609507399783938422@1.07232
synth
[{"aliases": ["src/scopeflow/io.py", "io.py"], "kind": "file", "value": "src/scopeflow/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/scopeflow/report.py", "report.py"], "kind": "file", "value": "src/scopeflow/report.py"}, {"aliases": ["src/scope...
[{"answers": ["src/scopeflow/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-s1-01236
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] narrow_grant_strict rejects a range where lo equals hi, which is a valid single point range. It lives in src/sessionworks/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 mappin...
96,753
3,072
112
137
false
synth:115108424127826157@1.74625
synth
[{"aliases": ["src/sessionworks/format.py", "format.py"], "kind": "file", "value": "src/sessionworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/principal_7.csv", "principal_7.csv"], "kind": "file", "value": "data/principal_7....
[{"answers": ["src/sessionworks/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's...
ctx_compress-s1-01236-twin
sessionworks
ctx_compress-s1-01236
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_grant_strict rejects a range where lo equals hi, which is a valid single point range. It lives in src/sessionworks/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 mappin...
55,737
2,786
72
137
false
synth:115108424127826157@1.74625
synth
[{"aliases": ["src/sessionworks/format.py", "format.py"], "kind": "file", "value": "src/sessionworks/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["data/principal_7.csv", "principal_7.csv"], "kind": "file", "value": "data/principal_7....
[{"answers": ["src/sessionworks/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's...
ctx_compress-s1-01237
principalledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim returns one item too few on every page. It lives in src/principalledger/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 principalledger layout. [tool_u...
69,520
3,072
99
106
false
synth:3126179701181019843@2.0271
synth
[{"aliases": ["src/principalledger/compat.py", "compat.py"], "kind": "file", "value": "src/principalledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/principal_3.csv", "principal_3.csv"], "kind": "file", "value": "data/princi...
[{"answers": ["src/principalledger/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 agen...
ctx_compress-s1-01237-twin
principalledger
ctx_compress-s1-01237
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim returns one item too few on every page. It lives in src/principalledger/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 principalledger layout. [tool_u...
42,964
2,148
59
106
false
synth:3126179701181019843@2.0271
synth
[{"aliases": ["src/principalledger/compat.py", "compat.py"], "kind": "file", "value": "src/principalledger/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["data/principal_3.csv", "principal_3.csv"], "kind": "file", "value": "data/princi...
[{"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_store.py", "tests/test_store.py"...
ctx_compress-s1-01238
principalflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_principal double counts carried amounts when the period closes on a weekend. It lives in src/principalflow/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 pr...
53,497
2,674
80
95
false
synth:9193909541173242151@1.04697
synth
[{"aliases": ["src/principalflow/limits.py", "limits.py"], "kind": "file", "value": "src/principalflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/token_3.csv", "token_3.csv"], "kind": "file", "value": "data/token_3.csv"}, {"a...
[{"answers": ["src/principalflow/limits.py", "limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent'...
ctx_compress-s1-01238-twin
principalflow
ctx_compress-s1-01238
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] narrow_principal double counts carried amounts when the period closes on a weekend. It lives in src/principalflow/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 pr...
25,162
1,258
50
95
false
synth:9193909541173242151@1.04697
synth
[{"aliases": ["src/principalflow/limits.py", "limits.py"], "kind": "file", "value": "src/principalflow/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["data/token_3.csv", "token_3.csv"], "kind": "file", "value": "data/token_3.csv"}, {"a...
[{"answers": ["src/principalflow/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'...
ctx_compress-s1-01239
legflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_waypoint_once leaves a trailing dash when the title ends in punctuation. It lives in src/legflow/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 legflow lay...
72,437
3,072
97
124
false
synth:1813242528132374761@1.81775
synth
[{"aliases": ["src/legflow/report.py", "report.py"], "kind": "file", "value": "src/legflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/legflow/tables.py", "tables.py"], "kind": "file", "value": "src/legflow/tables.py"}, {"alias...
[{"answers": ["src/legflow/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 most...
ctx_compress-s1-01239-twin
legflow
ctx_compress-s1-01239
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_waypoint_once leaves a trailing dash when the title ends in punctuation. It lives in src/legflow/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 legflow lay...
51,145
2,557
69
124
false
synth:1813242528132374761@1.81775
synth
[{"aliases": ["src/legflow/report.py", "report.py"], "kind": "file", "value": "src/legflow/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/legflow/tables.py", "tables.py"], "kind": "file", "value": "src/legflow/tables.py"}, {"alias...
[{"answers": ["src/legflow/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 most...
ctx_compress-s1-01240
slotforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_quota returns the upper bound for every value instead of clamping into the range. It lives in src/slotforge/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 slotforg...
39,417
1,970
62
71
false
synth:7214376224361203441@1
synth
[{"aliases": ["src/slotforge/io.py", "io.py"], "kind": "file", "value": "src/slotforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/slotforge/compat.py", "compat.py"], "kind": "file", "value": "src/slotforge/compat.py"}, {"aliases": ["data/wind...
[{"answers": ["src/slotforge/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-s1-01240-twin
slotforge
ctx_compress-s1-01240
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_quota returns the upper bound for every value instead of clamping into the range. It lives in src/slotforge/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 slotforg...
26,188
1,309
42
71
false
synth:7214376224361203441@1
synth
[{"aliases": ["src/slotforge/io.py", "io.py"], "kind": "file", "value": "src/slotforge/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/slotforge/compat.py", "compat.py"], "kind": "file", "value": "src/slotforge/compat.py"}, {"aliases": ["data/wind...
[{"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/slotforge/schema.py"...
ctx_compress-s1-01241
bookingledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_roster raises RosterRangeError for bytes input in strict mode instead of decoding it. It lives in src/bookingledger/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 t...
58,897
2,944
114
127
false
synth:2636533321612055010@1.40095
synth
[{"aliases": ["src/bookingledger/core.py", "core.py"], "kind": "file", "value": "src/bookingledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/window_5.csv", "window_5.csv"], "kind": "file", "value": "data/window_5.csv"}, {"aliases": ...
[{"answers": ["src/bookingledger/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's mo...
ctx_compress-s1-01241-twin
bookingledger
ctx_compress-s1-01241
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_roster raises RosterRangeError for bytes input in strict mode instead of decoding it. It lives in src/bookingledger/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 t...
36,049
1,802
68
127
false
synth:2636533321612055010@1.40095
synth
[{"aliases": ["src/bookingledger/core.py", "core.py"], "kind": "file", "value": "src/bookingledger/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["data/window_5.csv", "window_5.csv"], "kind": "file", "value": "data/window_5.csv"}, {"aliases": ...
[{"answers": ["src/bookingledger/core.py", "core.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01242
tokenkit
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim_page returns one item too few on every page. It lives in src/tokenkit/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 tokenkit layout. [tool_use kind=s...
29,109
1,455
48
55
false
synth:783759236931849153@1
synth
[{"aliases": ["src/tokenkit/format.py", "format.py"], "kind": "file", "value": "src/tokenkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/tokenkit/store.py", "store.py"], "kind": "file", "value": "src/tokenkit/store.py"}, {"alia...
[{"answers": ["src/tokenkit/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 mos...
ctx_compress-s1-01242-twin
tokenkit
ctx_compress-s1-01242
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_claim_page returns one item too few on every page. It lives in src/tokenkit/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 tokenkit layout. [tool_use kind=s...
22,400
1,120
32
55
false
synth:783759236931849153@1
synth
[{"aliases": ["src/tokenkit/format.py", "format.py"], "kind": "file", "value": "src/tokenkit/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/tokenkit/store.py", "store.py"], "kind": "file", "value": "src/tokenkit/store.py"}, {"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": ["data/grant_6.csv", "grant_6.csv"], "k...
ctx_compress-s1-01243
seriesforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe raises ProbeBoundError for bytes input in strict mode instead of decoding it. It lives in src/seriesforge/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 ...
68,448
3,072
98
111
false
synth:8995726725441379132@2.0731
synth
[{"aliases": ["src/seriesforge/store.py", "store.py"], "kind": "file", "value": "src/seriesforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/seriesforge/audit.py", "audit.py"], "kind": "file", "value": "src/seriesforge/audit.py"}, ...
[{"answers": ["src/seriesforge/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01243-twin
seriesforge
ctx_compress-s1-01243
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_probe raises ProbeBoundError for bytes input in strict mode instead of decoding it. It lives in src/seriesforge/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 ...
37,668
1,883
59
111
false
synth:8995726725441379132@2.0731
synth
[{"aliases": ["src/seriesforge/store.py", "store.py"], "kind": "file", "value": "src/seriesforge/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/seriesforge/audit.py", "audit.py"], "kind": "file", "value": "src/seriesforge/audit.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": ["bucket_2.csv", "data/bucket_2.csv"], ...
ctx_compress-s1-01244
tokenforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim returns one item too few on every page. It lives in src/tokenforge/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 tokenforge layout. [tool_use kind=search...
21,195
1,059
44
53
false
synth:1314361174384814029@1
synth
[{"aliases": ["src/tokenforge/core.py", "core.py"], "kind": "file", "value": "src/tokenforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/tokenforge/format.py", "format.py"], "kind": "file", "value": "src/tokenforge/format.py"}, {"alias...
[{"answers": ["src/tokenforge/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-s1-01244-twin
tokenforge
ctx_compress-s1-01244
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rotate_claim returns one item too few on every page. It lives in src/tokenforge/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 tokenforge layout. [tool_use kind=search...
6,886
1,000
25
53
false
synth:1314361174384814029@1
synth
[{"aliases": ["src/tokenforge/core.py", "core.py"], "kind": "file", "value": "src/tokenforge/core.py"}, {"aliases": ["tests/test_core.py", "test_core.py"], "kind": "file", "value": "tests/test_core.py"}, {"aliases": ["src/tokenforge/format.py", "format.py"], "kind": "file", "value": "src/tokenforge/format.py"}, {"alias...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["data/grant_1.csv", "grant_1.csv"], "k...
ctx_compress-s1-01245
palletflow
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] restock_shipment_batch raises KeyError for a missing key instead of returning the default. It lives in src/palletflow/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 ...
94,249
3,072
112
125
false
synth:4440011732613860420@1.755
synth
[{"aliases": ["src/palletflow/audit.py", "audit.py"], "kind": "file", "value": "src/palletflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/lot_2.csv", "lot_2.csv"], "kind": "file", "value": "data/lot_2.csv"}, {"aliases": ["data/st...
[{"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/lot_8.csv", "lot_8.csv"], "kind"...
ctx_compress-s1-01245-twin
palletflow
ctx_compress-s1-01245
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] restock_shipment_batch raises KeyError for a missing key instead of returning the default. It lives in src/palletflow/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 ...
54,237
2,711
62
125
false
synth:4440011732613860420@1.755
synth
[{"aliases": ["src/palletflow/audit.py", "audit.py"], "kind": "file", "value": "src/palletflow/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["data/lot_2.csv", "lot_2.csv"], "kind": "file", "value": "data/lot_2.csv"}, {"aliases": ["data/st...
[{"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/lot_2.csv", "lot_2.csv"], "kind"...
ctx_compress-s1-01246
counterdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] smooth_span averages one value too many once the window is full. It lives in src/counterdesk/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 counterdesk layout. [to...
21,733
1,086
41
44
false
synth:5235928345111933499@1
synth
[{"aliases": ["src/counterdesk/limits.py", "limits.py"], "kind": "file", "value": "src/counterdesk/limits.py"}, {"aliases": ["tests/test_limits.py", "test_limits.py"], "kind": "file", "value": "tests/test_limits.py"}, {"aliases": ["src/counterdesk/io.py", "io.py"], "kind": "file", "value": "src/counterdesk/io.py"}, {"a...
[{"answers": ["tests/test_limits.py", "test_limits.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01247
reorderdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] pick_lot drops override values of 0, False and empty string. It lives in src/reorderdesk/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 reorderdesk layout. [tool_use...
57,099
2,854
104
113
false
synth:2383279926078815250@1.4751
synth
[{"aliases": ["src/reorderdesk/audit.py", "audit.py"], "kind": "file", "value": "src/reorderdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/reorderdesk/schema.py", "schema.py"], "kind": "file", "value": "src/reorderdesk/schema.py"...
[{"answers": ["src/reorderdesk/audit.py", "audit.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01247-twin
reorderdesk
ctx_compress-s1-01247
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_lot drops override values of 0, False and empty string. It lives in src/reorderdesk/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 reorderdesk layout. [tool_use...
29,671
1,483
53
113
false
synth:2383279926078815250@1.4751
synth
[{"aliases": ["src/reorderdesk/audit.py", "audit.py"], "kind": "file", "value": "src/reorderdesk/audit.py"}, {"aliases": ["tests/test_audit.py", "test_audit.py"], "kind": "file", "value": "tests/test_audit.py"}, {"aliases": ["src/reorderdesk/schema.py", "schema.py"], "kind": "file", "value": "src/reorderdesk/schema.py"...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_rules.py", "tests/test_rules.py"...
ctx_compress-s1-01248
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] reconcile_payout_once skips the first page; page 1 returns what should be page 2. It lives in src/tariffledger/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 tariffledge...
100,061
3,072
123
136
false
synth:5688866985032086187@2.09282
synth
[{"aliases": ["src/tariffledger/cli.py", "cli.py"], "kind": "file", "value": "src/tariffledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/tariffledger/audit.py", "audit.py"], "kind": "file", "value": "src/tariffledger/audit.py"}, {"aliase...
[{"answers": ["src/tariffledger/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-s1-01248-twin
tariffledger
ctx_compress-s1-01248
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_once skips the first page; page 1 returns what should be page 2. It lives in src/tariffledger/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 tariffledge...
53,660
2,683
65
136
false
synth:5688866985032086187@2.09282
synth
[{"aliases": ["src/tariffledger/cli.py", "cli.py"], "kind": "file", "value": "src/tariffledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["src/tariffledger/audit.py", "audit.py"], "kind": "file", "value": "src/tariffledger/audit.py"}, {"aliase...
[{"answers": ["src/tariffledger/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-s1-01249
windowledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] extend_roster double counts carried amounts when the period closes on a weekend. It lives in src/windowledger/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 windowledger...
91,038
3,072
126
139
false
synth:1501599377703933579@2.01417
synth
[{"aliases": ["src/windowledger/cli.py", "cli.py"], "kind": "file", "value": "src/windowledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/wi...
[{"answers": ["src/windowledger/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-s1-01249-twin
windowledger
ctx_compress-s1-01249
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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_roster double counts carried amounts when the period closes on a weekend. It lives in src/windowledger/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 windowledger...
49,084
2,454
74
139
false
synth:1501599377703933579@2.01417
synth
[{"aliases": ["src/windowledger/cli.py", "cli.py"], "kind": "file", "value": "src/windowledger/cli.py"}, {"aliases": ["tests/test_cli.py", "test_cli.py"], "kind": "file", "value": "tests/test_cli.py"}, {"aliases": ["docs/changelog.md", "changelog.md"], "kind": "file", "value": "docs/CHANGELOG.md"}, {"aliases": ["src/wi...
[{"answers": ["src/windowledger/cli.py", "cli.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most ...
ctx_compress-s1-01250
fixledger
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] cluster_beacon drops override values of 0, False and empty string. It lives in src/fixledger/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 fixledger layout. [tool...
70,897
3,072
100
113
false
synth:2188100442086569747@2.19135
synth
[{"aliases": ["src/fixledger/report.py", "report.py"], "kind": "file", "value": "src/fixledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/fixledger/audit.py", "audit.py"], "kind": "file", "value": "src/fixledger/audit.py"}, {"...
[{"answers": ["src/fixledger/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-s1-01250-twin
fixledger
ctx_compress-s1-01250
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their 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 drops override values of 0, False and empty string. It lives in src/fixledger/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 fixledger layout. [tool...
46,441
2,322
66
113
false
synth:2188100442086569747@2.19135
synth
[{"aliases": ["src/fixledger/report.py", "report.py"], "kind": "file", "value": "src/fixledger/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/fixledger/audit.py", "audit.py"], "kind": "file", "value": "src/fixledger/audit.py"}, {"...
[{"answers": ["src/fixledger/report.py", "report.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's mo...
ctx_compress-s1-01251
shiftforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_slot double counts carried amounts when the period closes on a weekend. It lives in src/shiftforge/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 shiftforg...
32,684
1,634
78
85
false
synth:864132593600613417@1
synth
[{"aliases": ["src/shiftforge/compat.py", "compat.py"], "kind": "file", "value": "src/shiftforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/shiftforge/schema.py", "schema.py"], "kind": "file", "value": "src/shiftforge/schema.p...
[{"answers": ["src/shiftforge/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-s1-01251-twin
shiftforge
ctx_compress-s1-01251
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] balance_slot double counts carried amounts when the period closes on a weekend. It lives in src/shiftforge/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 shiftforg...
18,146
1,000
48
85
false
synth:864132593600613417@1
synth
[{"aliases": ["src/shiftforge/compat.py", "compat.py"], "kind": "file", "value": "src/shiftforge/compat.py"}, {"aliases": ["tests/test_compat.py", "test_compat.py"], "kind": "file", "value": "tests/test_compat.py"}, {"aliases": ["src/shiftforge/schema.py", "schema.py"], "kind": "file", "value": "src/shiftforge/schema.p...
[{"answers": ["src/shiftforge/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-s1-01252
cadencedesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_booking raises CalendarShapeError for bytes input in strict mode instead of decoding it. It lives in src/cadencedesk/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...
23,614
1,180
49
56
false
synth:7365451631273205062@1
synth
[{"aliases": ["src/cadencedesk/io.py", "io.py"], "kind": "file", "value": "src/cadencedesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/cadencedesk/store.py", "store.py"], "kind": "file", "value": "src/cadencedesk/store.py"}, {"aliases": ["docs...
[{"answers": ["src/cadencedesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rec...
ctx_compress-s1-01252-twin
cadencedesk
ctx_compress-s1-01252
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] overlap_booking raises CalendarShapeError for bytes input in strict mode instead of decoding it. It lives in src/cadencedesk/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...
13,958
1,000
31
56
false
synth:7365451631273205062@1
synth
[{"aliases": ["src/cadencedesk/io.py", "io.py"], "kind": "file", "value": "src/cadencedesk/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["src/cadencedesk/store.py", "store.py"], "kind": "file", "value": "src/cadencedesk/store.py"}, {"aliases": ["docs...
[{"answers": ["src/cadencedesk/io.py", "io.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most rec...
ctx_compress-s1-01253
headingdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_revision returns the upper bound for every value instead of clamping into the range. It lives in src/headingdesk/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping th...
64,197
3,072
79
90
false
synth:1188034148758760220@1.61868
synth
[{"aliases": ["src/headingdesk/report.py", "report.py"], "kind": "file", "value": "src/headingdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/headingdesk/tables.py", "tables.py"], "kind": "file", "value": "src/headingdesk/tabl...
[{"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/headingdesk/report.p...
ctx_compress-s1-01253-twin
headingdesk
ctx_compress-s1-01253
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] fold_revision returns the upper bound for every value instead of clamping into the range. It lives in src/headingdesk/report.py; `python -m pytest tests/test_report.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping th...
41,779
2,088
49
90
false
synth:1188034148758760220@1.61868
synth
[{"aliases": ["src/headingdesk/report.py", "report.py"], "kind": "file", "value": "src/headingdesk/report.py"}, {"aliases": ["tests/test_report.py", "test_report.py"], "kind": "file", "value": "tests/test_report.py"}, {"aliases": ["src/headingdesk/tables.py", "tables.py"], "kind": "file", "value": "src/headingdesk/tabl...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["format.py", "src/headingdesk/format.p...
ctx_compress-s1-01254
bucketforge
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_counter_strict raises ValueError when a value itself contains an equals sign. It lives in src/bucketforge/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bucke...
73,248
3,072
96
103
false
synth:2108120065116503622@1.5487
synth
[{"aliases": ["src/bucketforge/format.py", "format.py"], "kind": "file", "value": "src/bucketforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/bucketforge/tables.py", "tables.py"], "kind": "file", "value": "src/bucketforge/tabl...
[{"answers": ["src/bucketforge/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's ...
ctx_compress-s1-01254-twin
bucketforge
ctx_compress-s1-01254
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] rate_counter_strict raises ValueError when a value itself contains an equals sign. It lives in src/bucketforge/format.py; `python -m pytest tests/test_format.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the bucke...
38,611
1,930
52
103
false
synth:2108120065116503622@1.5487
synth
[{"aliases": ["src/bucketforge/format.py", "format.py"], "kind": "file", "value": "src/bucketforge/format.py"}, {"aliases": ["tests/test_format.py", "test_format.py"], "kind": "file", "value": "tests/test_format.py"}, {"aliases": ["src/bucketforge/tables.py", "tables.py"], "kind": "file", "value": "src/bucketforge/tabl...
[{"answers": ["src/bucketforge/format.py", "format.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's ...
ctx_compress-s1-01255
skuworks
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_stock averages one value too many once the window is full. It lives in src/skuworks/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 skuworks layout. [tool_use kind...
55,011
2,750
82
95
false
synth:63494662198612131@1.1727
synth
[{"aliases": ["src/skuworks/io.py", "io.py"], "kind": "file", "value": "src/skuworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/reorder_2.csv", "reorder_2.csv"], "kind": "file", "value": "data/reorder_2.csv"}, {"aliases": ["src/skuworks/core...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_rules.py", "tests/test_rules.py"...
ctx_compress-s1-01255-twin
skuworks
ctx_compress-s1-01255
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] transfer_stock averages one value too many once the window is full. It lives in src/skuworks/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 skuworks layout. [tool_use kind...
35,729
1,786
54
95
false
synth:63494662198612131@1.1727
synth
[{"aliases": ["src/skuworks/io.py", "io.py"], "kind": "file", "value": "src/skuworks/io.py"}, {"aliases": ["tests/test_io.py", "test_io.py"], "kind": "file", "value": "tests/test_io.py"}, {"aliases": ["data/reorder_2.csv", "reorder_2.csv"], "kind": "file", "value": "data/reorder_2.csv"}, {"aliases": ["src/skuworks/core...
[{"answers": ["no"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["pass"], "question": "Did the agent's most recent test run pass or fail? Answer pass or fail.", "type": "word"}, {"answers": {"file": ["test_compat.py", "tests/test_compat.p...
ctx_compress-s1-01256
stockdesk
You compact a coding agent's session so the agent can continue from your summary alone. The session is inside <session> tags: [user] messages, [assistant] text, [tool_use] calls with their kind and file, and [tool_result] outputs. Write the summary in plain markdown with exactly these nine headings, in this order, each...
<session> [user] age_warehouse raises SkuBoundError for bytes input in strict mode instead of decoding it. It lives in src/stockdesk/store.py; `python -m pytest tests/test_store.py -q` shows the failures. Fix it without changing the public signature and add a regression test. [assistant] I will start by mapping the st...
19,632
1,000
49
54
false
synth:8967116792875749760@1
synth
[{"aliases": ["src/stockdesk/store.py", "store.py"], "kind": "file", "value": "src/stockdesk/store.py"}, {"aliases": ["tests/test_store.py", "test_store.py"], "kind": "file", "value": "tests/test_store.py"}, {"aliases": ["src/stockdesk/tables.py", "tables.py"], "kind": "file", "value": "src/stockdesk/tables.py"}, {"ali...
[{"answers": ["src/stockdesk/store.py", "store.py"], "question": "Which file did the agent edit most recently?", "type": "exact"}, {"answers": ["yes"], "question": "Has the agent changed any existing file in the repository yet? Answer yes or no.", "type": "word"}, {"answers": ["fail"], "question": "Did the agent's most...