Commit graph

  • 73dd117f6a Revision D31 batch 2: actuality map atop decisions log, refreshed CLAUDE.md pipeline and status, banners and supersession marks across architecture docs Claude (backend session) 2026-07-11 05:34:25 +03:00
  • 5bc75d0e9a Pre-register exp13 translator bake-off: six arms, cross-family span-citing judges, output gates, frozen before paid calls Claude (backend session) 2026-07-11 05:30:19 +03:00
  • 9e3762eaa8 Split PROGRESS journal: move 330KB of closed 04-10.07 chronology to archive with verbatim headings, keep live sections and anchors, repoint incoming references Claude (backend session) 2026-07-11 05:28:52 +03:00
  • a83313d0c3 Sync README to D30 state, archive closed exp12 and package-5 prompts Claude (backend session) 2026-07-11 05:09:23 +03:00
  • 6a29c75668 Add exp12 monitor script and bump contract range to D30 in onboarding docs Claude (backend session) 2026-07-11 05:08:27 +03:00
  • 79f9ffe096 Ratify D30: D1 flip to bilingual editor superseding D17, reflow and sanitizer gates, conditional glossary v2, translator bake-off exp13, cheap-model track; issue D15.2 and exp13 prompts Claude (backend session) 2026-07-11 05:07:20 +03:00
  • 75db9e144f Land exp12 quality diagnosis with orchestrator review banner: mono-editor passivity confirmed, unanimity and fidelity-gate claims corrected, glossary v2 status trap flagged Claude (backend session) 2026-07-11 05:05:40 +03:00
  • 4952673b22 Ratify D29 absorbing reader-IDE research into D15.2 package scope: tmctl export with annotations schema, red policy, color map with grey tier, human_override lock layer Claude (backend session) 2026-07-11 03:37:55 +03:00
  • 458b0ea8d6 Land research/16 reader-IDE alignment with orchestrator review banner and journal entries: probe numbers corrected to DB truth, contract overclaims fixed Claude (backend session) 2026-07-11 03:31:08 +03:00
  • 4dabade8b4 Ratify D28 accepting backend package 5 with D24.4 recall-tradeoff amendment and D15.2 fix-list additions; sync status docs Claude (backend session) 2026-07-11 02:16:46 +03:00
  • aa47e257c4 Land backend package 5: per-model max_tokens floor on primary and hop, postcheck base-dst union, budget-refusal and coverage-recall pins, D23.4 tails; golden byte-identical Claude (backend session) 2026-07-11 02:13:21 +03:00
  • 73e5c0524c Ratify D27 owner xAI key policy: single data-sharing key for tests and pilot with pre-release opt-out, clean-key pilot blocker removed, glm-5 substitution no longer forced Claude (backend session) 2026-07-11 01:41:15 +03:00
  • a7377a747b Sync status docs and acceptance prompt to D26 freeze and exp12 priority Claude (backend session) 2026-07-11 01:23:15 +03:00
  • aa6bc240ec Ratify D26 quality-first pivot: stage B frozen after owner readability verdict, issue exp12 quality-diagnosis prompt with editor arm matrix and blind owner reading Claude (backend session) 2026-07-11 01:22:23 +03:00
  • 58f510d2de Sync status docs to D24/D25, issue backend package 5 and polygon package 4 prompts, redefine acceptance stage B to 300-chapter slice, archive GPT ascent prompt Claude (backend session) 2026-07-11 00:55:10 +03:00
  • 7a29147884 Ratify D24 stage-A acceptance with floor and post-check fixes plus stage-B ordering, and D25 research/17 absorption; journal orchestrator handoff cycle Claude (backend session) 2026-07-11 00:52:28 +03:00
  • a148f41e3e Land research/17 external GPT critique with orchestrator review banner: 19/19 sources verified, B2 verdicts re-graded PLAUSIBLE, mandate-seeding caveat recorded Claude (backend session) 2026-07-11 00:45:10 +03:00
  • 7ca14c2c5f Land phase-1 stage-A acceptance report with orchestrator verification: headline numbers reproduced by execution, text defects corrected, verdict caveats and alias-blind-spot drift note added Claude (backend session) 2026-07-11 00:44:20 +03:00
  • 4c48f893d8 Hand off orchestrator role second time: full project pillars, acceptance stage-A results, pending research/17 critique review, edge-case lessons, forward-planning mandate Claude (backend session) 2026-07-11 00:01:01 +03:00
  • 1727459e08 Refresh onboarding status paragraphs to post-D23 state with three parallel launches ready Claude (backend session) 2026-07-10 22:13:11 +03:00
  • c59dc77f01 Ratify D23: refactor package accepted with proven equivalence, golden guard as invariant eight, OpenReadOnly sanctioned; issue acceptance session prompt, sync docs Claude (backend session) 2026-07-10 22:12:10 +03:00
  • 296419c8ae Land backend package 4: runner decomposed to eight files, golden determinism guard, OpenReadOnly status during live runs, operator logging, tmctl unit tests Claude (backend session) 2026-07-10 22:10:24 +03:00
  • a96248f14b Ratify D22 accepting both packages with register caveat, 18+ judge revision, L3 pre-screen task, and D15.2 v3.1 amendments; sync status docs, archive prompts Claude (backend session) 2026-07-10 04:07:19 +03:00
  • 3551997292 Land polygon erotica package: exp11 three-register benchmark closing D14.4, seed finalized gender-hidden, fix-list D19.5, journal entries for both sessions with orchestrator review corrections Claude (backend session) 2026-07-10 04:02:10 +03:00
  • e1fcee4e25 Land backend package 3: D15.2 spec v3, D20.4 fix-list mutation-verified, D3 chain and channel B config with live-checked slugs, Gemini additive_total reasoning billing Claude (backend session) 2026-07-10 04:02:10 +03:00
  • 4908f956ba Issue two handoff prompts: GPT-5.6 external ascent critique with anti-anchoring phases, and backend code-health refactor package with golden determinism guard Claude (backend session) 2026-07-10 03:39:19 +03:00
  • ebe7765946 Issue reader-IDE research prompt covering alignment anchors, HITL editor prior art, trust visualization, and backend export contract; record owner frontend vision Claude (backend session) 2026-07-10 02:23:22 +03:00
  • 85550d226a Sync uniqueness-claim narrowing from D21.7 into strategic review TL;DR and verdict sections Claude (backend session) 2026-07-10 00:12:36 +03:00
  • 8643fdfe96 Add owner rule: provider anomalies require official vendor-doc fact-check before interpretation, recorded in CLAUDE.md, quirks header, and eval rules Claude (backend session) 2026-07-10 00:03:50 +03:00
  • 7828647319 Fact-check Gemini live: 2.5-series shutdown 2026-10-16, undocumented 404 flakes, PROHIBITED_CONTENT is non-configurable; update quirks calendar and erotica prompt Claude (backend session) 2026-07-09 23:57:47 +03:00
  • 2a77750190 Amend erotica polygon prompt before launch: Gemini judge pitfalls, D19.2 quirks-line task, optional P4 fidelity tail unblocking anti-echo wiring Claude (backend session) 2026-07-09 23:51:16 +03:00
  • be7388128b Sync status docs with D18-D21, narrow grok quirks line per D19.1, archive voice-state prompt, issue backend package-3 handoff prompt Claude (backend session) 2026-07-09 23:39:22 +03:00
  • 22cd5e5b91 Ratify D21: voice-and-state research accepted, voice profile, address-pair registry, reveal windows, tiered pilot arms, gated anti-echo wiring, uniqueness claims narrowed Claude (backend session) 2026-07-09 23:33:39 +03:00
  • 5953b37681 Land voice-and-state research package: research/15 with voice, scene-state, prompting, and P4 echo-mitigation probes, plus orchestrator corrections from external review Claude (backend session) 2026-07-09 23:31:23 +03:00
  • b508faa5a6 Hand off orchestrator role via dedicated session prompt, issue erotica polygon prompt closing D14.4 scope, archive completed package-2 prompts Claude (backend session) 2026-07-09 22:31:24 +03:00
  • b51355cc8d Ratify D20: backend package 2 accepted, D15.2 v2 direction approved pending two spec amendments, section 13 answers, acceptance-editor decision Claude (backend session) 2026-07-09 22:00:08 +03:00
  • b51ac4c2e0 Land backend package 2: review fix-list, GB18030 ingest with CJK chapter split, four cheap style flaggers with migration v7, D15.2 v2 spec, Gu Zhenren smoke Claude (backend session) 2026-07-09 21:58:28 +03:00
  • 3011b886ca Ratify D19 channel B composition, seed decisions, and echo-class generalization after polygon package 2 external review Claude (backend session) 2026-07-09 21:21:21 +03:00
  • 3d183bc2f4 Land polygon package 2: mirror sync with Go v3, explicit benchmark exp10, Mistral fidelity rerun, terse dialogue precision, with orchestrator corrections from external review Claude (backend session) 2026-07-09 21:16:32 +03:00
  • 17ccea482b Add echo-mode sub-mandate to voice-state research: mechanism hypotheses, prompt mitigations, prefill probe, budget bump Claude (backend session) 2026-07-09 21:04:17 +03:00
  • ec578ef307 Add voice-and-state research session prompt covering character voice, scene state, prompting art, request params, prior-art checks, and creative slot-bound hypotheses Claude (backend session) 2026-07-09 20:26:34 +03:00
  • 86b983d331 Ratify D18 acceptance book switch to Gu Zhenren zh-ru, issue package-2 handoff prompts for backend and polygon, archive closed prompts Claude (backend session) 2026-07-09 19:31:05 +03:00
  • b7556b3a4f Ratify external-review verdicts for both packages, block D15.2 implementation pending spec rework, add fix-lists, multisession git rule, and key-file gitignore Claude (backend session) 2026-07-09 19:07:55 +03:00
  • 9afea37bee Land pilot-harness package: memory_eval echo control and fidelity axis, M0-M3 rename, artifact blinding, exp09 v2 with D13, kana precision, Mistral probe, webnovel slice; journal entries for both sessions Claude (backend session) 2026-07-09 19:05:28 +03:00
  • 153277e3bb Land review-fixes package: D16 memory trust-boundary fixes, honest resume comments, tmctl status and redrive with migration v6, additive reasoning buffer, D15.2 spec draft Claude (backend session) 2026-07-09 19:05:28 +03:00
  • 6ab92b5e66 Add MISTRAL_API_KEY and update keys list for eval/backend Claude (backend session) 2026-07-09 16:30:00 +03:00
  • bc7ee58f80 Add handoff prompts for backend fixes package and polygon pilot-harness repair, linked from README and journal Claude (backend session) 2026-07-09 16:25:07 +03:00
  • 8ac568b97e Archive closed session prompts and add onboarding docs: root CLAUDE.md plus backend and eval READMEs Claude (backend session) 2026-07-09 16:16:11 +03:00
  • b01218bdca Ratify post-review decisions D13-D17 and sync MVP plan, PlantUML diagrams, journal, and superseded banners with the strategic review Claude (backend session) 2026-07-09 16:16:11 +03:00
  • f14eff45dc Commit owner's V1-V3 brief additions and two externally sourced failure-mode notes with provenance headers pending origin confirmation Claude (backend session) 2026-07-09 16:16:11 +03:00
  • b1d54b6103 Add strategic architecture review: verdict, SOTA-2026 survey, 9 verified findings, ranked risks, course corrections, and pilot protocol amendments Claude (backend session) 2026-07-09 15:22:00 +03:00
  • ff5f000da5 Consolidate docs after reconciliation audit: name 05-decisions-log source of truth, fold exp08 economics and grok reasoning fix, archive closed prompts, add backend/polygon/strategic handoff prompts Claude (backend session) 2026-07-05 21:32:18 +03:00
  • 1aed2c7627 Trim grok-reasoning re-narration in exp08 and cost_model: single-source the mechanism in provider-quirks, keep only cost-relevant facts plus pointers Claude (backend session) 2026-07-05 21:17:06 +03:00
  • 9792285e76 Correct exp04's 'grok without reasoning' characterization — grok-4.3 defaults to low reasoning (~13s), production editor sets reasoning_effort:none for 0 tokens; ranking unchanged Claude (backend session) 2026-07-05 21:07:39 +03:00
  • f4e7124c4b Consolidate config: wire grok-4.3 as the editor with an explicit reasoning_effort:none off-switch (control:effort), keeping DeepSeek's echo-mine no-op, and mark escalation stubs interim Claude (backend session) 2026-07-05 20:58:09 +03:00
  • 84737d032c Reconcile grok-reasoning across polygon docs: reasoning_effort:none disables it (probe bug), editor stays grok-4.3 with provider-quirks slug fixed; economics unchanged; add pilot mono-editor arm Claude (backend session) 2026-07-05 20:36:48 +03:00
  • 3911bafe30 Land polygon session 3: coverage-gate precision (0 FP/57, flip safe), cost-model v2 on grok-editor stack, and Phase-2.5 pilot protocol with a Go-mirror memory-eval Claude (backend session) 2026-07-05 15:44:44 +03:00
  • 6a68dfd44e Wire the approved glossary into the monolingual editor as CONFIRMED dst-only target constraints (D1) and drop {{text}} from the editor prompt Claude (backend session) 2026-07-05 15:37:04 +03:00
  • 38c7f3a04a Close the two adversarial-review majors: strip default-ignorable code points in both memory normalizers and trim surrounding whitespace on seed keying surfaces, bumping the normalizer version Claude (backend session) 2026-07-05 14:51:36 +03:00
  • 3df5983fee Add the Phase-2 polygon prompt: memory-eval tests deterministic selective injection with memory.go the source of truth and eval/memory_hotpath.py a pre-cc57c7b reference prototype Claude (backend session) 2026-07-05 14:35:27 +03:00
  • cc57c7bc02 Close external-review findings: gate on confirmed misses only, per-language min-key with collision-prone AMBIGUOUS downgrade, and fail-loud seed guards Claude (backend session) 2026-07-05 14:04:45 +03:00
  • 068f59d22f External-review memory bank v2: accept with 6 confirmed findings in the short-key/collision/gate zone, F1 and injection surface clean Claude (backend session) 2026-07-05 13:38:48 +03:00
  • aa00339abb Land memory bank v2 (glossary hot path, injection surface, F1 materialization, decl-aware post-check) with 8 self-review fixes Claude (backend session) 2026-07-05 08:13:04 +03:00
  • baf6ae31bd Augment the memory-v2 backend prompt with research/14's quantified post-check false-flag data, the stored-decl completeness caveat, and the E1 measurement assets Claude (backend session) 2026-07-05 03:47:16 +03:00
  • e23160b139 Answer the memory-v2 prompt open questions: confirm F1 materialization, seeded-glossary first veha, decl-form post-check as flagger, code-assembled injection Claude (backend session) 2026-07-05 03:42:27 +03:00
  • 15cf46f7a3 Add memory-v2 onboarding prompt for the next backend session: schema freeze, deterministic hot-path injection, F1 materialization, and mandatory post-check Claude (backend session) 2026-07-05 03:36:28 +03:00
  • 6d96bb4ac8 Close step-3a external-review findings: spine reconciliation for dangling/mislabeled chapters, UTF-8 normalization, ASCII quote-depth, ruby full-replace, and tiling fuzz tests Claude (backend session) 2026-07-05 03:33:57 +03:00
  • 6700a0f2a6 External-review шаг 3a: accept with 7 confirmed findings led by silent chapter-loss on malformed epub, recommend spine reconciliation Claude (backend session) 2026-07-05 03:04:15 +03:00
  • 8096224a2f Confirm the ruby-seed contract for memory-v2: base to src, reading as the Polivanov dst bridge strengthening B6, with name-vs-double-reading classification Claude (backend session) 2026-07-05 02:47:45 +03:00
  • b26e4df465 Replace the placeholder chunker with a real sentence-aware source segmenter, txt/epub ingest, and ruby-reading capture persisted to a new v4 table Claude (backend session) 2026-07-05 02:41:57 +03:00
  • 073ed62791 Add step-3a onboarding prompt for the fresh backend session: real chunker, txt/epub import, ruby capture Claude (backend session) 2026-07-05 01:48:58 +03:00
  • e4ca3a57fd Add the adaptive-memory research verdict (doc 14) with four executable probes: flagship logprob access, in-context demos vs glossary, lemma post-check, and the deterministic lever curve Claude (backend session) 2026-07-05 01:34:18 +03:00
  • 1878c3ea95 Track the closed memory-validation and extraction sessions' eval artifacts that committed docs already reference Claude (backend session) 2026-07-05 01:34:18 +03:00
  • 0b31222b5c Close external-review findings: fold classifier version and local backend tag into the snapshot, catch nested echo-mine forms, restrict escalation to the translator role Claude (backend session) 2026-07-05 01:22:40 +03:00
  • 3a5b966c16 External-review Veha 2.5: accept with 7 confirmed findings, confirm deepseek-v4-pro price, add coverage-precision validation task for Polygon Claude (backend session) 2026-07-05 01:07:36 +03:00
  • ed4f4ff5a7 Apply self-review fixes: fold fallback wire config into the snapshot, make escalation resume-idempotent and ceiling-tolerant, strip think in the coverage gate, gate only the translator Claude (backend session) 2026-07-05 00:35:34 +03:00
  • a98af88b12 Add an opt-in live-conformance harness per adapter behind a live build tag, verified against DeepSeek showing the echo mine stays disarmed on the real wire Claude (backend session) 2026-07-04 23:59:52 +03:00
  • 5cbf75cae8 Add single-hop escalation routing a deterministic content-failure to a named fallback model once, re-gated, budget-capped, with channel-B isolation by type Claude (backend session) 2026-07-04 23:56:50 +03:00
  • 6dc8f13c6c Fix the Kimi API base URL to api.moonshot.ai after a live /models probe showed api.kimi.ai does not resolve Claude (backend session) 2026-07-04 23:44:16 +03:00
  • 8957046424 Add the configurable excision coverage-gate porting refusal_bench sentence-coverage bit-for-bit, emitting excision_suspect and folded into the snapshot Claude (backend session) 2026-07-04 23:33:42 +03:00
  • d15cdb3b71 Add D12 failure-resilience model: three-class failure taxonomy, F3 idempotency fix bounded to at-most-once, progress-surfacing spec grounded in durable-execution prior art Claude (backend session) 2026-07-04 23:10:32 +03:00
  • 737d4c6a68 Fail-fast on any config that disables DeepSeek thinking at reasoning=off, the echo-mine regression gate that keeps thinking on by default Claude (backend session) 2026-07-04 22:31:49 +03:00
  • fc05b6c2e7 Answer Polygon on silent provider refusals (echo detected, excision pending) and add onboarding prompt for the robustness session Claude (backend session) 2026-07-04 22:13:34 +03:00
  • 357dde296b Add chapter/chunk book loop with per-chunk disposition flags, attempt-axis retries, content-hashed resume, and F4/D2.5 fixes Claude (backend session) 2026-07-04 22:00:30 +03:00
  • 0dbcce7e97 Harden refusal benchmark: split verdict taxonomy, live-audited provider configs, local warmup Claude (backend session) 2026-07-04 21:06:47 +03:00
  • 68059b0d3f Spec the memory-bet indicative eval as a ready-to-run task for the eval session Claude (backend session) 2026-07-04 20:40:10 +03:00
  • 636919c52f Add per-model capability wire-shape layer (D3.1) and fold context-assembly, memory-version, and resolved capability into the job snapshot (D5.2) Claude (backend session) 2026-07-04 20:32:14 +03:00
  • dee6efa282 Accept memory-bank validation: GO on schema and deterministic gates, bet gated by in-house eval running parallel; apply pings to R3/R7 Claude (backend session) 2026-07-04 20:26:00 +03:00
  • 8fac2b8968 Answer Phase-1 backend questions D8-D11: content-hash snapshot, ja-ru acceptance book, all-terms consistency, budget as config ceiling with stale cost numbers flagged Claude (backend session) 2026-07-04 19:42:37 +03:00
  • ce71a1a752 Add polygon eval scripts for local-model, editor and MoE benchmarks plus ollama launcher and provider config Claude (backend session) 2026-07-04 19:26:36 +03:00
  • 1985c16d23 Collapse economics to single direct-key EU-SaaS contour, drop BYOK/RU-law, finalize memory-validation session prompt Claude (backend session) 2026-07-04 19:10:30 +03:00
  • 87b92ba696 Add decisions log and provider-quirks, set grok-4.3 editor default and OpenAI-retained model stack per owner and editor bake-off Claude (backend session) 2026-07-04 18:47:45 +03:00
  • f77ea6cc41 Record Phase-1 decision requests (editor structure, bad-chunk disposition, escalation models, memory schema, ongoing Analyst) with reviewer recommendations Claude (backend session) 2026-07-04 17:20:10 +03:00
  • c504cfeb14 Mark native Anthropic adapter deprecated (kept as reference for Phase-2 Gemini adapter); record real DeepSeek cache-hit confirmation Claude (backend session) 2026-07-04 17:07:38 +03:00
  • 56c2f5a61c Fix tmctl report: materialize request_log rows so they outlive the query context that defer-cancel tore down mid-iteration Claude (backend session) 2026-07-04 16:56:21 +03:00
  • c4676d53dc Drop Anthropic from the model stack per owner decision: remove provider and Claude models, repoint escalation and C2 judge to non-Anthropic Claude (backend session) 2026-07-04 16:35:52 +03:00
  • 9d78385c37 Apply external review fixes: snapshot local overrides, gate-enabled guard, TM-key doc note, db-lock gitignore, techdebt disposition Claude (backend session) 2026-07-04 16:29:25 +03:00
  • 56670e3d67 Scope CheckKeys to reachable models so backend runs on two keys; escalation keys required only when gates enabled Claude (backend session) 2026-07-04 15:49:27 +03:00
  • 214e005155 Harden Phase 0 after self-review: snapshot coverage, response-price fallback, usage clamp, job-status lifecycle, config guards, source normalization, hash robustness Claude (backend session) 2026-07-04 14:49:43 +03:00
  • 2006989b9b Untrack .claude/settings.local.json so its allow rules aren't repo-supplied Claude (backend session) 2026-07-04 14:35:50 +03:00