- The last pass drops a sentence-level fact error (sav_total_as_strength and nine others) instead of the
whole section; in 依据与边界 a bracketed gloss or a 「某某层:」 label is not a SAV strength verdict; the
writer rule says how to write the Ashtakavarga total.
- The computed basis names a planet once when it is also the lagna lord.
- 现在这几年: the major period is at most a short background clause; the two periods are the current and
next sub-period (notice G-23).
- Chat basis completion uses only the question's own domain routes (parents never gets the family D7).
- 判断依据 display: 保留为参考窗口, detector and 本方法 notes, 原书 wording.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
- chat-basis-completeness.ts: after the wording guards, compare the route's
on-card Narayana / MK / PiK / DK / PK / AmK / divisional charts with the
判断依据 block and append the missing ones from the card's own sentences
(no model call, body untouched)
- chat-basis-clean.ts: drop 只作辅助, 单一系统推断, 两套运程分别计算 and
undated 本轮未交付 clauses
- reading_facts.py: reader words for the self-reference witness and the
unused PD reason; contract G-22 (parameter-sensitivity once, no 激活,
no stability-evaluation line)
- personal-report-reading-wording.ts basisReaderParagraphs at display
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Since the 2026-11-16 merge, answers on identical input dropped these four groups.
The basis rule now asks for them when the card has them and they bear on the
question; the body is unchanged. The display cleaner also drops 「不构成两种方法一致」
and calibration notes note by note, keeps a following note and the period, and a
basis written twice keeps its first copy. The backtest script defaults to the
product's single answer mode.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
PLAIN_TERMS takes the report's house rule (READING_HOUSE_RULE): the plain
voice no longer says 「说这颗星管哪几个宫、落在哪」 next to 「不写宫位编号」.
Red lines say which part of life a planet stands for instead of 哪一宫;
dates go to the month, exact days only for key turning points. After a
简明解读 answer, body lines that still write a house (question names no
house) join the existing single rewrite call (BUG-1948); a rewrite that keeps
a house, drops a date or changes the line's lead is refused and the clause
holding the house is dropped. 深度解读 is unchanged. Shared house pattern
in lib/house-mechanics.ts. Contract notice G-20.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Owner 2026-11-22 「去掉」: a self-chosen scenario is treated like a rectified
time. Personal, annual and synastry pages and exports no longer write the
self-chosen statement ending 「其他可能的出生时间尚未作比较。」, and the
personal opening drops 「提到的时间段是有条件的推测」 for such reports.
Stored reports follow at display. Contract notice G-17 records the
suspension of the astrologer's 11-17 / 11-19 self-choice rules pending her
confirmation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Round 28: the section failed its repair on fact_with_inference and fell back
to a computed text that only said 「所需依据已算出」, so the planets, signs,
houses and periods BUG-2143 moved there were lost.
- writer rule (contract G-17): this section is fact paragraphs, no inference
words; the repair note names the words; the last pass drops inferring
clauses of a fact paragraph like reader wording (salvage reason
fact_with_inference)
- computed 依据与边界 lists the chart evidence per theme (sign, house, lord,
dignity, D10/D2/D9, A10/AmK/UL/DK, health lords, Badhaka, current and next
period to the month), only audit rows with content, birth-scenario limits
only for an uncertain time (BUG-2141), packed into the stored 4 paragraphs
- chat 判断依据: drop 「只是条件性窗口」「只作当年结构」 tails and 「本轮只解释到…」,
「本轮未交付」 → 「这次没有展开」; numbers kept (655 answers, 46 lines → 0)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Owner 2026-11-22: 「木星管第五宫和第八宫,落在第八宫自己的星座」 every one or two sentences and runs of
exact days made the reading unreadable. The report rules now keep house numbers and exact days in
「依据与边界」 (evidence moved, not deleted; the skill's evidence-first is the working order), 「现在这几年」
covers the current and the next period linked by cause, and the in-report questions take the same
plain terms. New reader-wording checks house_mechanics and date_density go to the one repair; the
last pass drops house clauses and shortens dates to the month.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Owner 2026-11-22 「同意」: users saw 「本节关于……尚未作比较。」 at the end of almost every
section, because every time not confirmed in 生时纠正 was trial-read at ±30 minutes.
- birth-time-trust.ts: uncertain = source approximate / period_only / unknown, record kind
approximate / derived, 「说不准」, or a declared range > 0 (same declaration the rectification
page reads). Everything else is read as entered: personal report confirmed_minute, annual
report ±1 confirmed_minute (was known:false → ±30).
- No section line in the writer, computed fallback or annual sections; stored lines are taken
out on the page, export and question material (依据与边界 keeps its own limit sentence); the
opening clause and appendix mark about birth time are gone.
- Uncertain and not settled in 生时纠正: one opening sentence with a /rectification link
(page and export, personal and annual); 「按你提供的出生时间,」 on dated conclusions and the
question-material lead only in that case.
- Contract R-04 / §AM rewritten, notice G-15; annual writing rule and both chat rules updated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Astrologer's Yoga manual 2026-11-21, P0. asraya_partial and asraya_yoga are
disabled with an accuracy note (ids kept); Aasraya proper stays as
Rajju/Musala/Nala. Two Kaalanirdesat progeny rules get conditional Chinese and
English effects. Topics file rebuilt; enabled count 408 -> 406.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXp2U3BH1SxwAGyhhbe1
The reading facts gave the writer 「PVR 表 30」 as the functional-profile label and
Narayana readings with page and table citations; the annual material and its
program-written notes named author versions and table numbers. Facts now carry
plain labels (ids kept), every report writer and in-report answer gets one
no-citation clause, the reading and the answers are checked, and stored text
loses its citations where it is shown.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
The reading facts gain timing.period_connections from jyotisha-20's indirect_activation (BUG-2121): per
house only the matched paths, their lords and levels, and one fact line; built even when the data
chapters were not rendered. The writer rule and the question instruction call it an index, not an
activation verdict; the eight body sections hold back 激活 / 引动 / 被点亮 / 被触发.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Round twenty-four: 「事业」 and 「感情」 fell back to three bare computed lines. A section
was sent back with its first reason only; the repair fixed that one and the dated
sentence without its birth-time condition survived. The condition is now inserted by
the program, the repair note names every reader-wording rule, and after the repair a
section failing only on reader wording drops the offending sentences and keeps the
rest when at least 60% and two sentences stay. Also: richer computed 事业/感情, the
basis rule names the Narayana, witness and same-fact fields, report scenario note and
annual Bhrigu note in the 11-21 wording, and a claim about every planet's strength is
unsupported when the material lists one below the line.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Per the astrologer's 2026-11-21 review (items 1 and 2): one line per
section in her wording, naming only the affected claim types; a dated
conclusion in an affected paragraph says it is by the chosen birth time;
question material keeps the line; stored readings show the new line;
dignity is never mapped to 施展不开 / 最能发挥 / 问题已经消失.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
The astrologer (2026-11-21) ruled that ordinary readings and forecasts never
use the user's experiences or feedback; only the rectification flow may.
Her reference files still told the model to back-calculate the user's known
events first, stop predicting below 3/5 matches, and grade A by such
verification. Until she edits them upstream, a product layer cuts those
passages out of every text sent to a model: the personal reading's quoted
references, the chat's methodology excerpts, and every chat prompt before
it reaches the provider (covering skill_read). The files are not edited;
a test names any rule that stops matching once she changes them.
Grade A no longer asks for a historical back-test. The status
user_history_verification_required is no longer a gap: no verification
questions, no penalty in ranking, never shown as 「需要过往经历核对」, not
offered to the report or synastry writers. 「未见阻断项」 is caught in the
reading and in report-question answers. Contract §AU announces the layer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
The personal report reading wrote UL, 入庙, 小运, 婚姻煞 and the like in the
sections written for ordinary readers. Owner 2026-11-20: the body uses plain
words, technique names stay in 「依据与边界」 and the data chapters.
- Rules (product contract, report-reading block, notice G-9): the eight
sections take the chat's PLAIN_TERMS and time-naming rule; dignity is judged
from dignity_zh but told by its meaning; 「现在这几年」 gives dates and the
ruling planet, no dasha names, no Narayana; names only in 「依据与边界」.
- Check: technique_name, after every fact check, through the existing one
section repair; a writer's own plain gloss next to a name is kept instead.
A section held back for a name still counts toward the draft length, so the
draft stays usable and reaches the repair.
- Computed fallback text uses plain labels and drops the Narayana paragraph.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
- a range chosen in rectification is the range the annual report reads (same source as the personal report and chat)
- a divisional ascendant changing inside the checked range no longer caps the conclusion type at D; the first rated item of that section carries the plain birth-time line
- D2 Hora candidates no longer make the wealth ceiling unreadable
- writing rules: technique, numbers and status go to the folded basis; body text in everyday words, no disclaimers; a reader wording pass after the checks
- the basis folds under the text on the page and follows it in the export
- a planet on its own natal position is a return
- basis length caps 600/800 -> 1000; a rewrite logs section id and check codes
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
A pronoun whose planet was in a removed sentence takes that planet; only a planet
back in its natal sign is given to the reading; the 现在这几年 section leads with
the stage of life and dates only the main dasha; the health reminder is one fixed
program sentence; the synastry appendix no longer doubles 所以 and shows the
technical summary with names, no machine tags, minutes as 分 and 秒.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Astrologer 2026-11-19, answer to question 17. Two flags per condition: witness_is_reference
(the Moon in the 1st from the Moon, true for every chart) and witness_is_debilitated_planet
(Mercury as its own witness, a real position condition that still counts). A planet that hits
only the always-true branch keeps the label 落陷, gets no row or cancel relation and is never
the basis of a Raja Yoga candidate; the literal hit stays in backstage evidence (literal_only).
This is the product policy neecha_bhanga_policy_v1; the raw rule version stays light_on_life_p295.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Astrologer 2026-11-19, layering rule and Q19. The page and export drop the
exploratory block; the appendix and stored rows use her sentences and keep
the four no-period states apart; the follow-up no longer volunteers scan
periods or reads system gaps out as missing data.
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
The main detector forms a cancellation for each debilitated planet with the same name and
rule_id. The detected-list merge, the route sieve and the yoga layer each treated the second
planet as a repeat, so Einstein's Moon kept its status label but lost its row, id and edge.
Per-object rules now dedupe by (rule_id, planet): instance key and evidence id name the planet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Evidence-pack step 2, one-topic vertical slice (owner 2026-11-18; the astrologer's 11-18
review §二.4–二.6, §三 questions 1 and 3, §六 step 2).
- The evidence base gains the one relation the data states with a named rule: the Neecha
Bhanga main detector cancels the debilitation effect of its planet (the placement stays),
and depends on the witnesses it cites. No weakens edge: there is no named rule for one.
- lib/evidence-pack.ts builds the wealth pack from the card's own indexed facts: direct
evidence; limits (cancellations with their witnesses, engine-stated yoga reductions,
coexisting pressure); time (only when the question asks about time, range from the
question); gaps and disagreements. The dependency closure follows every level with a
visited set, cycle record and budget; a stopped walk is a visible gap.
- CONSULT_EVIDENCE_PACK_DOMAINS (default off; wealth on staging only) swaps the wealth key
block for the pack; off, the model view is byte-identical. The time scope comes from the
split question, never from user feedback (test E).
- The backtest harness reads a whole facts packet; timing.horizon is registered as non-fact.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
The card's model view gains evidence_index: the snapshot id and coverage, one id
template per fact path, status codes with a legend and the per-id exceptions.
Positions, houses, roles, afflictions, Arudha and derived houses, points, SAV,
KP, Argala, transits, double transit, Saturn cycles, the annual chart, every
dasha level, judgment and participation rows now resolve; rows that name their
own id keep it. Branch rows carry their #br: id from the facts packet. A
coverage check lists any card leaf that is neither a fact nor marked not
citable; it is empty on all public golden cards.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Per the astrologer's 2026-11-18 review of the evidence-pack draft (§二 1-5, §五.3, §五.5):
logical_key / snapshot-scoped evidence_id / content_hash; birth-branch ids bound to the parent
range and intervals; snapshot over opaque subject, birth input revision, calculation profile,
rule versions, retractions, observation instant and coverage (calculation time recorded only);
execution / formation / interpretation / source status; base-level edges with type, scope,
conditions, method version, basis and review status (engineering dependencies only);
dependency_closure with visited set, cycle report and budget; interpretation_key; a parseable
synthetic fixture and the review's test list.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8
Evidence-pack step 1 (owner 2026-11-18). One ASCII form, ev:<family>:<body>, for D1 and varga
placements, AL/UL derived houses, special points, Shadbala, dasha periods, transits, double
transit and yoga rule instances; an evidence base with links taken only from what the data
states; a snapshot id binding the birth input revision, calculation date and rule versions.
The internal-word check drops ids from the basis section and logs them in the body.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DbBXoyn7Hj3LCW3BUs5bZ8