Kalah TDD study · Part 29 of 29 · Data

Supporting data: run records

The run records of full-1, rep-1 and rep-2.

Source runs full-1, rep-1 and rep-2 at tdd-experiment c3dd988

Contents

The run records of full-1, rep-1 and rep-2: their settings, cap hits, failed collection steps and unattributed agents.

Run record (full-1)

Values are copied from the harness's own records, so they are verified against args.json, batch.json, each clone's collect.tsv and the workflow's journal.jsonl.

Tag. Source: runs/<tag>.

Value: full-1

Model. Source: costs.json.

Value: claude-opus-5-5

Agent runtime. Source: costs.json.

Value: 2.1.295, system hash 3712685649

Effort. Source: args.json.

Value: medium

Methods. Source: args.json.

Value: 11: classic, noun-verb, mapped-path, functions-first, tpp, coarse-to-fine, annealed, multi-start, mutation-guided, property-based, held-out

Caps. Source: args.json.

Value: {"cycles": 15, "iterations": 15, "starts": 3}

Methods doc sha256. Source: args.json.

Value: 6460695052da555ae469d575fbe2ce5b3ae57df5825abf102a290ce4e65cba21

Started (UTC). Source: batch.json.

Value: 2026-10-15T23:53:12Z

Ended (UTC). Source: batch.json.

Value: 2026-10-16T00:47:49Z

Cap hits

LabelCapValue
tpp:L2-7#1cycles15
mapped-path:L3-8#1cycles15

Failed collection steps

No collection step failed.

Unattributed agents

LabelAgent id
harness:seed-multi-start-1a2e48671b47052495
harness:seed-multi-start-2a074ef43b82e176b2
harness:seed-multi-start-3a7749ec328507544c
harness:winner-multi-startafd24f1f068a61830
review:r04a532205d8e924e7de
review:r05a044cf5dcae5d0bd9
review:r01aec286dbe85601360
review:r06a370b95413d08b9ed
review:r08a07864e742330a97c
review:r09aa0d9e81b269ba99e
review:r10a874c4d9b9d053704
review:r11a4b4c8714b9e69924
review:r07a488bf7f553892624
review:r03a1900bbca92cfdae9
review:r02a7ac4fc2ba83cdd51

Run record (rep-1)

Values are copied from the harness's own records, so they are verified against args.json, batch.json, each clone's collect.tsv and the workflow's journal.jsonl.

Tag. Source: runs/<tag>.

Value: rep-1

Model. Source: costs.json.

Value: claude-opus-5-5

Agent runtime. Source: costs.json.

Value: 2.1.308, system hash 3955976082

Effort. Source: args.json.

Value: medium

Methods. Source: args.json.

Value: 20: rep-1-r01, rep-1-r14, rep-1-r11, rep-1-r07, rep-1-r04, rep-1-r03, rep-1-r10, rep-1-r18, rep-1-r12, rep-1-r20, rep-1-r17, rep-1-r08, rep-1-r13, rep-1-r15, rep-1-r02, rep-1-r05, rep-1-r09, rep-1-r06, rep-1-r16, rep-1-r19

Caps. Source: args.json.

Value: {"cycles": 15, "iterations": 15, "starts": 3}

Methods doc sha256. Source: args.json.

Value: 12f026f6ba5c9c1856e1fb587afa14a66be207151a79d380635b16dd19e857c6

Started (UTC). Source: batch.json.

Value: 2026-10-26T10:02:33Z

Ended (UTC). Source: batch.json.

Value: 2026-10-26T11:14:30Z

Cap hits

No cap hit was recorded.

Failed collection steps

No collection step failed.

Unattributed agents

Table, 20 rows and 2 columns
LabelAgent id
review:r18af4d0785cf68cb281
review:r07a564c02cdb0c85c47
review:r12a0bf98eb3e8097a2e
review:r03ad3c417591735fd40
review:r11a79a056c78816fd61
review:r20a6c04fdfb6442f7f9
review:r10a49ca83aff355436c
review:r01a72a64d6b230fae61
review:r04a7eaa4e719b922b25
review:r14a4ebdbc068964a6d4
review:r15a656a3088378fe617
review:r09a2e41f9ab141604b1
review:r17ad2c0e358a063289c
review:r19a6b137659ee86472a
review:r05aa395a53cc52cc090
review:r16a9e7d0dc2a4565cf6
review:r13a7b468c838a17fd63
review:r06ad2bd20c4ddd0fee3
review:r08a5ac0ff3908780707
review:r02a1db48b9bd84bfea8

Run record (rep-2)

Values are copied from the harness's own records, so they are verified against args.json, batch.json, each clone's collect.tsv and the workflow's journal.jsonl.

Tag. Source: runs/<tag>.

Value: rep-2

Model. Source: costs.json.

Value: claude-opus-5-5

Agent runtime. Source: costs.json.

Value: 2.1.309, system hash 3544498427

Effort. Source: args.json.

Value: medium

Methods. Source: args.json.

Value: 20: rep-2-r07, rep-2-r08, rep-2-r06, rep-2-r05, rep-2-r04, rep-2-r16, rep-2-r19, rep-2-r11, rep-2-r12, rep-2-r09, rep-2-r10, rep-2-r14, rep-2-r15, rep-2-r01, rep-2-r17, rep-2-r03, rep-2-r02, rep-2-r20, rep-2-r13, rep-2-r18

Caps. Source: args.json.

Value: {"cycles": 15, "iterations": 15, "starts": 3}

Methods doc sha256. Source: args.json.

Value: 12f026f6ba5c9c1856e1fb587afa14a66be207151a79d380635b16dd19e857c6

Started (UTC). Source: batch.json.

Value: 2026-10-29T06:42:01Z

Ended (UTC). Source: batch.json.

Value: 2026-10-29T08:40:43Z

Rebuilt clones. Source: rebuild.json.

Value: rep-2-r04, rep-2-r08, rep-2-r14, rep-2-r17, rep-2-r20

Rebuild started (UTC). Source: rebuild.json.

Value: 2026-10-30T05:59:54Z

Rebuild ended (UTC). Source: rebuild.json.

Value: 2026-10-30T06:19:14Z

Rebuild harness commit. Source: rebuild.json.

Value: cc8c8cbb06373ce400666528458a349eb0ae4d7a

Rebuild harness delta sha256. Source: rebuild.json.

Value: 8fde5d73aecf82700e53426dcbca82f2bd86cc1a421e20fa57f0176d86958cdf

Cap hits

No cap hit was recorded.

Failed collection steps

No collection step failed.

Unattributed agents

Table, 20 rows and 2 columns
LabelAgent id
review:r16a125ca2ecf756ac47
review:r06ae76c4b84c05f1cbd
review:r12a78b3b93532c35daa
review:r11a97162b4ce4abf4db
review:r07a709a730770c2a618
review:r05ad878aeb87137da37
review:r19a3dafd9124e09b686
review:r09a84e9463459c52eb2
review:r13a89fe85ccc3d1ac5e
review:r02aae11e1dffda4f13d
review:r01a83d9542b53b4fbdb
review:r03a061775a1c54b4ec9
review:r10a8c21e04f97336337
review:r15a111d54060f87d759
review:r18a7e010a4616446636
review:r04aad90ee9a3ac0580e
review:r08ae163d1b4e1869d6b
review:r14a012ecf23dcc6656d
review:r17a81ac019ba8208c20
review:r20a57bf82e032e6c086