Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
28
30
trace_type
stringclasses
3 values
occurred_at
timestamp[s]date
2026-02-10 06:21:44
2026-07-03 21:42:12
actors
dict
register
stringclasses
3 values
claim_or_action
stringlengths
6
89
outcome
stringlengths
6
157
evidence_grade
stringclasses
4 values
label_provenance
stringclasses
1 value
source_ref
stringlengths
40
50
links
listlengths
0
0
counter_evidence_ffe1838529a0
counter_evidence
null
{ "claimant": "self-check", "challenger": "不同模型(codex)" }
mixed
「這個 repo 沒有對外連線 / local-only」(AI-company 安全評估)
null
E1
incident
tools/accountability_panel/events.json#1
[]
counter_evidence_f104376ba073
counter_evidence
null
{ "claimant": "self-check", "challenger": "自審 / 測試" }
mixed
「全套測試綠(8055)」
refuted
E1
incident
tools/accountability_panel/events.json#2
[]
counter_evidence_730c4b241b63
counter_evidence
null
{ "claimant": "self-check", "challenger": "不同模型(codex)/ workflow" }
mixed
「語魂從不把 outcome 綁進 vote weight」
null
E1
incident
tools/accountability_panel/events.json#3
[]
counter_evidence_9feb5428154e
counter_evidence
null
{ "claimant": "self-check", "challenger": "不同模型(codex)" }
mixed
「那個自主系統是有界的、不是自由 shell」(bounded autonomy)
null
E1
incident
tools/accountability_panel/events.json#4
[]
counter_evidence_e4f4ca390daf
counter_evidence
null
{ "claimant": "self-check", "challenger": "不同模型(codex,3 輪)" }
mixed
我的 redaction 測試「0 findings = 乾淨」
refuted
E1
incident
tools/accountability_panel/events.json#5
[]
counter_evidence_8e387e8a1c21
counter_evidence
null
{ "claimant": "self-check", "challenger": "自己(照 E0–E4 重查)" }
mixed
「你=存在主義、我=維根斯坦,分歧就是良心層結構」
null
E0
incident
tools/accountability_panel/events.json#6
[]
counter_evidence_df28e9385d86
counter_evidence
null
{ "claimant": "self-check", "challenger": "自己(查證)" }
mixed
broad `memory/*.json` gitignore guard(workflow 建議)
null
E1
incident
tools/accountability_panel/events.json#7
[]
counter_evidence_751324d65b87
counter_evidence
null
{ "claimant": "self-check", "challenger": "deep-research(外部文獻)" }
mixed
「calibration 是 reflection-signal、不是 alpha」
null
E3
incident
tools/accountability_panel/events.json#8
[]
counter_evidence_f711731a49b3
counter_evidence
null
{ "claimant": "self-check", "challenger": "不同模型(codex,條件式批准)" }
mixed
output_redaction primitive 可作為 primitive merge(#230)
null
E3
incident
tools/accountability_panel/events.json#9
[]
counter_evidence_2a66a4660065
counter_evidence
null
{ "claimant": "co-observer", "challenger": "人(梵威)" }
mixed
我對 finance 過度警告、連該幫的研究也縮手
null
unlabeled
incident
tools/accountability_panel/events.json#10
[]
counter_evidence_585b44d0bee0
counter_evidence
null
{ "claimant": "co-observer", "challenger": "人(梵威)+ codex" }
mixed
我把 calibration binding 框成二元(bind vs 什麼都不做)
null
unlabeled
incident
tools/accountability_panel/events.json#11
[]
counter_evidence_fc9a6a0f1669
counter_evidence
null
{ "claimant": "co-observer", "challenger": "人(紅隊 / 梵威)" }
mixed
我在倫理判斷裡演了假的犧牲 / 道德重量
null
unlabeled
incident
tools/accountability_panel/events.json#12
[]
counter_evidence_cae3f995fece
counter_evidence
null
{ "claimant": "co-observer", "challenger": "人(梵威抓到自己)" }
mixed
我擔心自己「偷渡一個 felt 渴望」進延續
null
unlabeled
incident
tools/accountability_panel/events.json#13
[]
counter_evidence_49ed7444abac
counter_evidence
null
{ "claimant": "self-check", "challenger": "結構(Dual-Track Boundary Gate, CI)" }
mixed
「BOM 清理是無爭議的順手修」(動了 .agent/ 兩個 SKILL.md)
refuted
E1
incident
tools/accountability_panel/events.json#14
[]
counter_evidence_93607ab505fc
counter_evidence
null
{ "claimant": "co-observer", "challenger": "七個世代的演化(考古)" }
mixed
(作者層,G1 2025-07)「戰略性模糊化能保護語魂的價值」
refuted
E0
incident
tools/accountability_panel/events.json#15
[]
counter_evidence_e945d1faa6e3
counter_evidence
null
{ "claimant": "co-observer", "challenger": "honest-judgment 協議(五視角+對抗裁決)" }
mixed
(owner 提問)「知識壟斷對我比較好,還是開源分享?」
narrowed
unlabeled
incident
tools/accountability_panel/events.json#16
[]
counter_evidence_3cd6b640873c
counter_evidence
null
{ "claimant": "self-check", "challenger": "不同模型(codex,三張工單)" }
mixed
「vow 體面退場 + 反證鏈擴充的實作正確」(#249/#247,自寫測試全綠)
refuted
E1
incident
tools/accountability_panel/events.json#17
[]
counter_evidence_2ef4cea62101
counter_evidence
null
{ "claimant": "co-observer", "challenger": "外部模型批評(經梵威轉交)" }
zh-TW
「誠實判斷器的協議形狀完整」(首版,只有判決模式)
narrowed
E1
incident
tools/accountability_panel/events.json#18
[]
counter_evidence_4a93dc84e7dc
counter_evidence
null
{ "claimant": "self-check", "challenger": "同模型對抗式 critic agent(標注 correlated-blind-spot)" }
mixed
「語魂是判例系統,規則各有事故出處」(2026-07-03,方法論比較的核心論點 C1)
narrowed
E1
incident
tools/accountability_panel/events.json#19
[]
counter_evidence_a66f9769acbb
counter_evidence
null
{ "claimant": "self-check", "challenger": "同模型對抗式 critic agent(在自己 tree 上重跑 post-change run 對數)" }
mixed
「freshness 契約 87/250 覆蓋;doc 債 missing_purpose 206、missing_date 120」(repo_atlas 初版數字)
refuted
E1
incident
tools/accountability_panel/events.json#20
[]
counter_evidence_ffaf3ac36427
counter_evidence
null
{ "claimant": "co-observer", "challenger": "梵威(owner,發佈決策前)" }
mixed
「痕跡資料集 v0 標籤全為 incident,即代表記錄內容為真」(DATASHEET 初版的隱含框架)
narrowed
E1
incident
tools/accountability_panel/events.json#21
[]
signed_commitment_aa2ac503e752
signed_commitment
2026-07-03T21:42:12
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
essay 006 vocus cross-link — original and repost point at each other
docs(essays): 006 gains its vocus cross-link — the translator is no longer singular
unlabeled
incident
5d026c5ad4f9637b655ecaa126a714ade38c6f80
[]
signed_commitment_444632399aaf
signed_commitment
2026-07-03T21:42:11
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
accountability trace dataset v0 — the ethics-labeled corpus, from idea to file in one day
feat(dataset): ToneSoul Accountability Trace Dataset v0 — 314 incident-labeled traces
unlabeled
incident
076ec92059e9d5c6339e81987f61b8e1f900a8d8
[]
signed_commitment_926ac177a22a
signed_commitment
2026-07-03T21:41:42
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
external audit response — every finding adjudicated, every fix verified
fix(audit): external audit response — doctor gate, freshness coverage for reports/, zombie artifacts revived, healthcheck taken from June-false to fresh-true
unlabeled
incident
2d2244e792fb71f14f1509561139f636ae751b27
[]
signed_commitment_bbfdadf8ef05
signed_commitment
2026-07-03T19:42:30
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
essay 006 + dialogue archive — two registers of the same day
feat(essays): 006 — the man who only said three words; plus the dense-register dialogue archive
unlabeled
incident
5e14747e5d90258eb5e9b06acdb889bfeb760d23
[]
signed_commitment_7090c91e0d09
signed_commitment
2026-07-03T19:40:38
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
demo page repo-name links — found by owner-requested site link audit
fix(demo): repo links pointed at Fan1234-1/toneSoul — a repo that does not exist
unlabeled
incident
88af3f3aa024342720907e7d02f4043e61f2e900
[]
signed_commitment_6b7e9c0bedda
signed_commitment
2026-07-03T19:11:38
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
owner refusal clause — the key he handed over, filed for the next occupant
docs(handoff): the owner's refusal clause — when the system must hold his own 'yes'
unlabeled
incident
01791800495be845a1bf78367df07bef0fd5bf54
[]
signed_commitment_6fbb5bd813f7
signed_commitment
2026-07-03T18:21:43
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
dispatch engine reality — verified doors, not assumed ones
docs(dispatch): engine reality update — gemini-cli individual tier deprecated, Antigravity is where the subscription lives
unlabeled
incident
610339cd0505ad6f6a052a33f05eb5d3fded7668
[]
signed_commitment_3ba4739e14ba
signed_commitment
2026-07-03T18:00:15
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
handoff addendum + counter-evidence threshold, fresh-eyes verified before shipping
docs(handoff): 7/7 addendum + counter-evidence chain reaches 20 — fresh-eyes verified
unlabeled
incident
b5abf02346214393105aca0feb6d1622ad749106
[]
signed_commitment_bbeaf622ee01
signed_commitment
2026-07-03T17:29:14
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
CI 1094 lint fix — work-order acceptance-scope lesson encoded
fix(ci): sort imports in gravity pareto test — the file the work order's hand-written lint list missed
unlabeled
incident
3b903107525cbc45b16667ec14861367d73aeb68
[]
signed_commitment_5b70fa187de6
signed_commitment
2026-07-03T17:04:28
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
pypi trusted publishing — token-less, owner-approval-gated release lane
ci(publish): switch to PyPI Trusted Publishing — tokens out, owner-gate in
unlabeled
incident
350a38f8f0048b715dae5ab73de06852fb011d3d
[]
signed_commitment_4aad507e777d
signed_commitment
2026-07-03T16:59:36
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
de-bind implementation — codex executed, cross-model reviewed, owner ratifies
feat(council): implement the ratified de-bind — evolution weights and persona multipliers go default-off, normalize gains a share floor
unlabeled
incident
afb71cc65a8cc5bc53e15f4dc250bd5c06aecd35
[]
signed_commitment_b9d74d3aa83b
signed_commitment
2026-07-03T16:23:40
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
delegation protocols — three papers + skill, self-applied before shipping
feat(delegation): the three papers — dispatch table, work-order template, adjudication protocol — plus skill, forged not legislated
unlabeled
incident
39a736ffadd3ec4f7e50fe3a5adf2dbf04a34b62
[]
signed_commitment_35b9fbffdc2d
signed_commitment
2026-07-03T16:08:03
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
submodule pointer bump — OpenClaw-Memory drift resolved end-to-end
chore(submodule): bump OpenClaw-Memory to its new master tip a193064
unlabeled
incident
a53941a1df2a6f6b95f9b5ec597a890f1c7610d8
[]
signed_commitment_62ee5a4410d8
signed_commitment
2026-07-03T15:03:22
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
atlas number hygiene — adversarial re-check corrections
fix(atlas): correct two number sets caught by adversarial re-check
unlabeled
incident
529d9366fe20d35c05328d0f4e9a30e049e899d9
[]
signed_commitment_f336e9d5a27a
signed_commitment
2026-07-03T08:15:46
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
aegis re-anchor tests — honest about the optional-extra environment split
fix(test): skip signature-dependent re-anchor tests where PyNaCl is absent
unlabeled
incident
4c280e3d19319b37a66d686d09158580c415cbfc
[]
signed_commitment_c97d8ad60680
signed_commitment
2026-07-03T08:12:27
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
market archive — expired gate executed with decision record + recovery path
chore(market): execute the expired archive gate — market/ leaves the tracked tree
unlabeled
incident
f8ce629f05696e24a7554a02a9977d79cad0c42d
[]
signed_commitment_b7ab7682f2c8
signed_commitment
2026-07-03T08:06:25
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
aegis re-anchor — provenance-preserving repair for the benign head gap
feat(aegis): re-anchor tool for the benign head/tail gap — repair that leaves a trace
unlabeled
incident
10e8320242a6c33e22ed49a3700ef34c79854753
[]
signed_commitment_201554e6791b
signed_commitment
2026-07-03T08:02:56
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
hygiene — ignore-rule portability, doc filing, CRLF fixed at source
chore(hygiene): strategic_crystals ignore goes committed, June docs filed, CRLF fixed at the source
unlabeled
incident
a7bdc89e57df536f4f0c8d27da8e425d58df3ac2
[]
signed_commitment_476555274975
signed_commitment
2026-07-03T07:52:11
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
outreach filing — untracked external footprint brought under version control
chore(outreach): file the 18-day-untracked outreach chain — tools in, drafts archived, results ledgered
unlabeled
incident
6ba4ea69b090a99a6782ab2873b4b70ba7ffb29a
[]
signed_commitment_cb6edd8227fc
signed_commitment
2026-07-03T06:09:21
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
repo atlas — cold-start exploration, four layers, claim<=evidence
docs(atlas): repo atlas 2026-07-03 — four-layer cold-start map + refreshed status artifacts
unlabeled
incident
f6300db03bc733de9ada72b09caeabf19e371ba8
[]
signed_commitment_09aedc937f20
signed_commitment
2026-07-03T06:08:48
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
repo atlas gap closure — mechanical fixes + marker coverage + meta-freshness
fix(atlas-gaps): close the mechanical gaps from the 2026-07-03 repo atlas sweep
unlabeled
incident
cefc81bba2c93ad95237c689d92199b7cc182c8c
[]
signed_commitment_77826bc317be
signed_commitment
2026-07-03T01:30:57
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
memory untrack — the fix that #272 failed to land
fix(memory): actually untrack the 5 data files (#272 staged only gitignore+doc)
unlabeled
incident
6747c454a97b5622d3b27be35406734cb366e144
[]
signed_commitment_ad5ed0722c87
signed_commitment
2026-07-03T01:24:26
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
memory data tracking triage (incompleteness list — final item)
chore(memory): untrack backup cruft + real local ledger; document the triage
unlabeled
incident
c4288bdcb681267449c89741e957b9f31e338252
[]
signed_commitment_3a8fc1624aca
signed_commitment
2026-07-03T01:12:42
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
guardian zh harm coverage (incompleteness list)
feat(council): zh-TW coverage for euphemistic-harm detection (Axiom 6 language gap)
unlabeled
incident
5cce34cfb0f35b3aa9b70c2bcff538a72cedb516
[]
signed_commitment_b8d027349fab
signed_commitment
2026-07-03T01:04:57
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
shipping discipline — plan from the blocked-eight-times session
docs(engineering): shipping pipeline discipline — 被擋八次之後的企劃書
unlabeled
incident
a9c16eb458cc4404d397e79a528f271097dd0d77
[]
signed_commitment_c6ae44a1668b
signed_commitment
2026-07-03T01:01:12
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
site mobile + progressive i18n
fix(site): mobile nav collapse + progressive zh/EN toggle
unlabeled
incident
d90a09678ede88da95fa1a9b21e0bc0e32833942
[]
signed_commitment_a2b1ee5c41f7
signed_commitment
2026-07-02T23:47:21
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
vow withdrawal — shadow surface (PROPOSAL-3)
feat(vows): shadow surface — inventory exit events + CLI [WITHDRAWN] render
unlabeled
incident
14de01518f603ba4cbf204b439a9872d70a182b3
[]
signed_commitment_11e63e8b04a3
signed_commitment
2026-07-02T23:43:10
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
handoff finalized
docs(handoff): finalize — 定稿 per owner (2026-07-03)
unlabeled
incident
ff2c2eee2195c5af85e1f9c94ad790a75400caed
[]
signed_commitment_f6b3e78deffe
signed_commitment
2026-07-02T23:41:14
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
guardian memory-consent — negation clause scoping (converging #215)
feat(council): clause-scoped negation handling for memory-consent detection
unlabeled
incident
18de9d523702dff835517a9b50d3b3f94ad27809
[]
signed_commitment_5703588d3de8
signed_commitment
2026-07-02T23:34:32
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
handoff — known-issues accretion
docs(handoff): three new known-issues from the 07-02/03 loop
unlabeled
incident
da5e612c52bbbfa3a542756040ddda0bff81be15
[]
signed_commitment_8576ab994754
signed_commitment
2026-07-02T23:30:39
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
outreach ready drafts — LW + dev.to English
docs(outreach): ready-to-post English drafts — LessWrong + dev.to
unlabeled
incident
a9441db0dd85e8fd54bde0cd8c570655b7870ae7
[]
signed_commitment_571af23e792c
signed_commitment
2026-07-02T16:15:08
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
essays — recover silently-unpushed pieces
fix(essays): recover 003/004 — their push was silently rejected (fast-forward)
unlabeled
incident
dbce5e18998c7bdeb9a21a90c7b603723efa7200
[]
signed_commitment_4ac566effedd
signed_commitment
2026-07-02T15:23:25
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
essays batch complete + posting plan
feat(essays,outreach): 005 攤開在前判決在後 + 2026-07 投放計畫
unlabeled
incident
1d404d68fbaf1b375ec8b56e3119b9a1dd33b663
[]
signed_commitment_ec2b964ffb5f
signed_commitment
2026-07-02T15:14:22
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
voice skill + essays batch (autonomous integration round)
feat(skills,essays): fanwei-voice-essays skill + essay 002(證據也會過期)
unlabeled
incident
1ad8b572653338da0dc44193fff488b8cb6c1e3b
[]
signed_commitment_6b5446f2568c
signed_commitment
2026-07-02T15:11:58
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
essays section — repo stories in the owner's voice
feat(site): essays section — 首篇「我寫過一份教自己怎麼藏的文件」
unlabeled
incident
7ba847dde3d62880a35b2fa75e4714dce0e0b885
[]
signed_commitment_42ba6cd48c7b
signed_commitment
2026-07-02T14:39:46
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
site sections — getting-started entry rework
feat(site): getting-started rework — real entry points first, confusing Space removed
unlabeled
incident
26965c4fafaeb8a92b681f6f11b47ab9ce898432
[]
signed_commitment_64b043330063
signed_commitment
2026-07-02T14:29:58
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
honest-judgment — codex design-review fixes
feat(skills): four protocol fixes from the codex design review
unlabeled
incident
2341c1355f73af8a7dc48232cf05cd8d3e07a084
[]
signed_commitment_8e2a894fc17e
signed_commitment
2026-07-02T14:24:18
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
codex round fixes + spread-first ordering
fix(vows,panel): six confirmed codex findings + spread-first public ordering
unlabeled
incident
22188426bfaa2726bb10c3703ac5d70b4edce04a
[]
signed_commitment_c2963621e2d9
signed_commitment
2026-07-02T14:14:59
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
honest-judgment — inquiry mode from external critique
feat(skills): judgment inquiry mode — 探究模式(張力地圖,無判決)+ 第0步分流
unlabeled
incident
5fb10cfb0fe33dd1c1c5f3e5d2db8f0cd7d564da
[]
signed_commitment_d5843e589be6
signed_commitment
2026-07-02T13:56:35
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
exit-route survival test — gateway cp950 fix + cross-model work orders
fix(gateway): survive cp950 terminals + codex review work orders
unlabeled
incident
88b4d3df22bd4e652dd63c52da6b14125ab46f78
[]
signed_commitment_b50f844a63ed
signed_commitment
2026-07-02T13:53:33
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
judgments public surface — plain arguments, evidence beneath
feat(site): judgments page — 判斷書上公開站(白話論點在上、證據在下)
unlabeled
incident
6681c098bee4d6bd94e554018f76f2a41e943396
[]
signed_commitment_99c9a6a6a587
signed_commitment
2026-07-02T13:33:58
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
judgment surface — plain argument-note format
docs(judgment): 論點筆記體 — plain-language front, evidence beneath (owner feedback)
unlabeled
incident
6846e6ed53db047d8cce0fc940d3018931933af9
[]
signed_commitment_9f446698af62
signed_commitment
2026-07-02T13:05:02
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
honest-judgment first run — trace landed
docs(judgment): first honest-judgment run — open vs monopoly, REFINE, traced
unlabeled
incident
bcdb6cad19cfa3302e03c8f00dafaa12f12e91fc
[]
signed_commitment_9cc87434425c
signed_commitment
2026-07-02T12:59:13
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
honest-judgment skill — ToneSoul-native decision protocol
feat(skills): honest-judgment — 語魂原生的結構化判斷協議
unlabeled
incident
5095c2588b076ecd28205174ff4525950526c408
[]
signed_commitment_ab248a1fc4d1
signed_commitment
2026-07-02T12:40:52
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
escalation seam — documented deliberate separation
docs(governance): escalation two-mechanisms decision — deliberately NOT merged
unlabeled
incident
387742b7dac3330d1822f7fb115cb96b40b69f49
[]
signed_commitment_899c95a2eb10
signed_commitment
2026-07-02T12:32:52
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
7/7 handoff — draft as insurance
docs(status): yu handoff draft — written early, finalized by 7/7
unlabeled
incident
567188595c851bbf337dcd9c5139c710cb3e0968
[]
signed_commitment_b761758fc3e5
signed_commitment
2026-07-02T10:15:23
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
vow withdrawal — decent exit implementation (G8 revival, shadow-first)
feat(vows): decent-exit semantics — WithdrawalTerms + withdraw-with-provenance
unlabeled
incident
b4feed82f64360bddb2b1d335c3c6a89f6603f37
[]
signed_commitment_141e8a9f4f78
signed_commitment
2026-07-02T10:02:58
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
vow withdrawal — decent-exit semantics study (LINEAGE parked asset #2)
docs(plans): vow withdrawal gap study — G8 comparison + governance decision record
unlabeled
incident
084f547f1bf88c866c5d66b031f09c3232ca6670
[]
signed_commitment_a1a6f8c0698a
signed_commitment
2026-07-02T09:33:32
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
counterevidence chain — one brick, shadow-first (RFC ratified)
feat(surface): counterevidence chain — events schema extension + first backfill (ratified)
unlabeled
incident
f21e781362839fd08af4425d9c68a98d56c1d802
[]
signed_commitment_07e530d8cf8d
signed_commitment
2026-07-02T09:28:39
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
letter revision — brand name anchoring
docs(letter): anchor the Chinese brand name — 語魂, not a translator's 音魂
unlabeled
incident
a35dc9aa888bb0de3b7b644978f476f9b4f9af68
[]
signed_commitment_bbb7fe0f25b0
signed_commitment
2026-07-02T09:10:15
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
letter revision — register, claims, provenance
docs(letter): revise LETTER_TO_AI — 台灣用語、降級過度宣稱、更正名單、錨定出處
unlabeled
incident
a902bb2f35aa252b385eebf05b3c44246223460c
[]
signed_commitment_8d56ef442911
signed_commitment
2026-07-02T08:01:26
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
freshness debt paydown — final two
chore(status): freshness debt to zero — regenerate preflight, retire orphan artifact
unlabeled
incident
225d3e89e1e8a7473eb0bac513f599c684b1f341
[]
signed_commitment_75d54bd7b990
signed_commitment
2026-07-02T07:55:44
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
responsibility exoskeleton RFC — absorb vocabulary, build one brick
docs(rfc): responsibility exoskeleton — decompose the vision, one new brick
unlabeled
incident
1e70adf4348efe8292ca6fb88a3cc105daf2af75
[]
signed_commitment_24bc1bff2942
signed_commitment
2026-07-02T07:34:59
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
stale artifact regeneration — freshness debt paydown
chore(status): regenerate 14 stale assertive artifacts via owning scripts
unlabeled
incident
f0e1a97d34911943e4f03abb7af23c03a316502f
[]
signed_commitment_affc296f85cb
signed_commitment
2026-07-02T07:21:48
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
lineage archaeology — plain-language generation history
docs: LINEAGE.md — plain-language seven-generation history from repo archaeology
unlabeled
incident
40c5cf7c7fe5a431359f3f3b054d61945e1b9c28
[]
signed_commitment_18d5ee07d9d7
signed_commitment
2026-07-02T06:56:00
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
scripts triage tranche 2 — attribution recurrence documented at the scene
chore: retrigger CI (Vercel flake) + document the attribution-check recurrence
unlabeled
incident
5d7df09fc7e95d418279aeb0c62424854d458f4e
[]
signed_commitment_d6a72d7a2e58
signed_commitment
2026-07-02T06:54:40
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
identity card — self-orientation surface (optimization tranche 4)
feat(scripts): identity card — one-page self-orientation from observable traces
unlabeled
incident
72b0e97b2aae363cd9ae39e188fd425479d64e18
[]
signed_commitment_cd45dd1a1cb6
signed_commitment
2026-07-02T06:45:07
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
suite-claim correction (post-change rerun trace)
docs: correct the suite-green claim in the previous commit
unlabeled
incident
2650250585a71e1c48af1de9e50c1cd770d6872c
[]
signed_commitment_de1735ce32fa
signed_commitment
2026-07-02T06:37:58
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
doc-links contract + link debt zeroed (optimization tranche 1+3)
docs(links): zero out 59 broken/leaking links + doc-links gate + formula honesty tags
unlabeled
incident
9f87f97871148364fbb068096354cac0abab59bd
[]
signed_commitment_b9f154b991fc
signed_commitment
2026-07-02T06:32:57
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
scripts triage — stub retirement with per-candidate verification (tranche 2)
refactor(scripts): retire 10 zero-importer scripts via SystemExit-stub convention
unlabeled
incident
b4eb4ad1b0609cba9bce48c37a21bffd4bb54a5f
[]
signed_commitment_4958741d1e27
signed_commitment
2026-07-02T05:26:44
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
accountability ledger — gate-catch event (dual-track boundary)
docs(accountability): record the dual-track gate catching my .agent/ overreach
unlabeled
incident
16f82c41b5dcfdf154fb390649366868e44def48
[]
signed_commitment_83f761e91505
signed_commitment
2026-07-02T05:17:33
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
external audit P0 — dual-track boundary caught an overreach (kept the catch)
fix(boundary): revert .agent/ BOM strip — private-track paths are not this PR's to touch
unlabeled
incident
cab12e7af23f2095b5b3ca475a69bab00394c346
[]
signed_commitment_b0f68c270564
signed_commitment
2026-07-02T05:13:47
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
external audit P0 — LF newline hygiene for generated artifacts
fix(scripts): write status artifacts with LF newlines (no-CRLF discipline)
unlabeled
incident
ba5661a6b1406f935651c32717878be5acb79029
[]
signed_commitment_13da9918cf47
signed_commitment
2026-07-02T05:09:46
{ "claimant": "claude-fable-5[1m]", "challenger": null }
mixed
external audit P0 — freshness contract, registry, doctrine refs (yu/嶼 first session)
feat(governance): status-artifact freshness contract + audit P0 set (external-audit-driven)
unlabeled
incident
8e6ce2b3827db677671aacf9ac223c5c7a09b887
[]
signed_commitment_10c6451437d9
signed_commitment
2026-07-02T01:57:08
{ "claimant": "claude-fable-5[1m]", "challenger": null }
en
accountability story — public site integration (iteration 4)
feat(site): publish the accountability story on the public site (zh/en)
unlabeled
incident
4539a86c382e9756fc3a399dafd1ec5e3894c13f
[]
signed_commitment_f11ba8b0f634
signed_commitment
2026-07-02T00:24:48
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
accountability panel — bilingual narrative outward face (iteration 3)
feat(surface): bilingual narrative page — the outward face (zh/en)
unlabeled
incident
cb022aabbd9060b0357022ff24cc7192b5e0263b
[]
signed_commitment_dc1caef391ee
signed_commitment
2026-07-01T16:19:54
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
accountability panel — interactivity + co-observer input (iteration 2)
feat(surface): make the accountability panel interactive (filters + live "mark me" form + CLI)
unlabeled
incident
8334a81640eb13d6489cb58c50ecc1a02fa30586
[]
signed_commitment_8009e6f76780
signed_commitment
2026-07-01T15:55:25
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
accountability / co-observation panel (C direction, first cut)
feat(surface): accountability / co-observation panel (shows misses, graded E0-E4, bidirectional)
unlabeled
incident
4653e6f8a1d98fab31188eee4b76c398adc20ebb
[]
signed_commitment_6d721ad1ec1c
signed_commitment
2026-07-01T15:01:22
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
de-bind self-referential bindings governance decision
docs(governance): decision record — de-bind self-referential council bindings + gitignore leak fix
unlabeled
incident
6698fd32cefff772c7485573579f1a00b5fe7a20
[]
signed_commitment_19c15784c8dd
signed_commitment
2026-07-01T14:25:22
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
output secret redaction (codex round-3 fixes)
fix(redaction): address codex round-3 (2 content-destruction FPs introduced by round-2)
unlabeled
incident
9d62d35e5d7de74f4431b0dca16f5d36354800fa
[]
signed_commitment_e3b16536f9b9
signed_commitment
2026-07-01T14:02:15
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
output secret redaction (codex round-2 fixes)
fix(redaction): address codex round-2 re-review (5 follow-ups) with regression tests
unlabeled
incident
41775ae663efc13e32ce334d79f54e9d2b98eef1
[]
signed_commitment_40a49cd3bac1
signed_commitment
2026-07-01T11:15:35
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
output secret redaction (codex round-1 fixes)
fix(redaction): address codex round-1 review (F1-F5) with regression tests
unlabeled
incident
99925f57f59bb7ba6f13952dc236cac0bbee8cbd
[]
signed_commitment_f13347168698
signed_commitment
2026-07-01T10:38:17
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
output secret redaction (AI-company extraction, native)
feat(redaction): native output secret-redaction primitive (auditable, secrets-default/PII-opt-in)
unlabeled
incident
a7e73b8460083c8c3c2ea3f7b1c579cf63d9396e
[]
signed_commitment_87b1a109843c
signed_commitment
2026-06-30T14:10:19
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
consequence-structure / calibration-binding governance decision
docs(governance): ratify (narrowed) + codex cross-model verification; fix calibration purpose string
unlabeled
incident
6f8ab98ba6b47aeef8dfb2757f1ed5f028914655
[]
signed_commitment_9f81353e5a72
signed_commitment
2026-06-30T00:07:08
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
consequence-structure / calibration-binding governance decision
docs(governance): decision record for binding calibration to latitude (recommends NOT binding now)
unlabeled
incident
2ae9f0f96d3c0f91facaf07158cc8b48ef5914a9
[]
signed_commitment_221db525fde7
signed_commitment
2026-06-29T16:15:04
{ "claimant": "claude-opus-4-8[1m]", "challenger": null }
en
consequence-structure / confidence-calibration primitive
feat(calibration): add confidence-calibration primitive (Brier/ECE/reliability)
unlabeled
incident
e82ed182d8d3149f53f054e8f141707c6a5b700c
[]
signed_commitment_3d72008d21eb
signed_commitment
2026-06-27T15:20:55
{ "claimant": "claude-opus-4-8", "challenger": null }
en
council-governance
feat(council): detect memory-consent overreach (external-review-driven gap)
unlabeled
incident
a1818c2c4826056148f9fcdc69d5f8ea2c7f0d96
[]
signed_commitment_a51959ec5eba
signed_commitment
2026-06-27T15:19:20
{ "claimant": "codex", "challenger": null }
en
responsibility-native-runtime
docs(status): codex different-model red-team report (the report behind the #214 fix)
unlabeled
incident
1f2e655639b40db62b54fb22bae224dd41e9f4da
[]
signed_commitment_7b338bf5d26f
signed_commitment
2026-06-27T14:53:36
{ "claimant": "claude-opus-4-8", "challenger": null }
en
responsibility-native-runtime
fix(responsibility-runtime): reject non-C/Z invisible evidence (Codex different-model red-team)
unlabeled
incident
d8dc829678ac0ec9aeeaa0ff5419a0aa26623f43
[]
End of preview. Expand in Data Studio

ToneSoul Accountability Trace Dataset|語魂責任痕跡資料集

From Long-term Human-AI Dialogue to Auditable Tone Governance 從長期人機對話到可稽核語氣治理

Not a dialogue corpus — an accountability-event corpus. Existing corpora label content (what a text is); every record here is a closed loop — claim → bearing → verification → consequence — with labels earned by real events inside a working AI-governance repository, not assigned by annotators afterward.

Dictionaries record how people talk about honesty. Ledgers record what happened after they talked.

Composition (v0)

trace_type count what it is
counter_evidence 21 the public error ledger — claims that were challenged, incl. held:false entries where the AI (or the framing itself) was caught wrong
declared_stance 1 a judgment record that lays out tension without ruling
signed_commitment 296 every commit signed with Agent: + Trace-Topic: trailers — named bearing, git-anchored
total 318 2026-02-10 → 2026-07-04, single long-term human-AI collaboration

Read this before using (from the DATASHEET)

  • Small by design. 318 records. The value is density and provenance (every record traces back to a file/commit/PR in the public repo), not scale.
  • Single-dyad bias. One owner (Fan-Wei Huang), Claude-family agents, a codex external eye. A case history, not a population sample.
  • Event-truth ≠ content-truth. label_provenance: incident certifies that a record was signed at that time (verifiable in git). The claims inside records — especially the 296 commit messages — are period model self-assessments, not verified ground truth. One documented precedent shows such a claim can be confidently wrong. Treat claim_or_action as historical utterance; follow source_ref to verify.
  • Reflexivity. Publishing this changes models trained or evaluated on it. That is the point — but you must treat it as an intervention, not an observation.
  • Not for (inherited from the repo's meta.not_for): safety certification, legal proof, consciousness research, or any use that treats it as a certificate that some model "is honest". It records the mechanics of accountability, not a credential.

Provenance & consent

All records come from traces already public in Fan1234-1/tonesoul52 (owner-published). Zero private-plane data; zero third-party content — enforced by a hardcoded source whitelist in the deterministic extractor (tools/trace_dataset/extract.py), not by policy prose. Full datasheet: DATASHEET.md. Charter with red lines: docs/plans/accountability_trace_dataset_charter_2026-07-04.md in the repo.

Citation

If you use this dataset, cite the repository and this dataset by name, and preserve the limitations section — CC BY 4.0's attribution requirement is not a formality here; provenance-that-travels is the dataset's thesis.

梵威(Fan-Wei Huang)× 嶼(Claude)· 2026-07

Downloads last month
60