Add lore review council and AI prose critic

This commit is contained in:
Meta Vens 2026-08-22 22:07:31 +02:00
commit e8add7409a
45 changed files with 2495 additions and 18 deletions

View file

@ -1,7 +1,7 @@
---
title: Kalsa Project Memory
status: active
snapshot_at: 2026-08-22T17:26:42Z
snapshot_at: 2026-08-22T19:54:33Z
---
# Kalsa Project Memory
@ -29,6 +29,8 @@ database and does not grant canon status to any setting claim.
- `memory-public-epistemic-ownership` — `Kalsa/Public/` is an inhabitant-facing knowledge commons, not an abbreviated author encyclopedia. Explanatory models require a culture, institution, lineage, text, witness, or named synthesis; translated terms must carry evidence, material reproduction, interests, and characteristic limits rather than smuggling backstage taxonomy into player knowledge.
- `memory-divine-uncertainty` — Gods are powerful, situated witnesses rather than authorial oracles. Divine confidence, memory, doctrine, and testimony remain evidence shaped by binding, rivalry, self-conception, and blind spots.
- `memory-play-canon` — Persistent play creates receipts and canon candidates. Runtime outcomes become shared setting canon only when an owning canonical note explicitly adopts them.
- `memory-independent-lore-review` — Every deepening pass uses seven independent lore critics before repair and after repair: institutional causality; culture and situated knowledge; material life and ecology; history, conflict, and change; Ghostlight play and consequence; AI-isms and prose; and reader experience and navigation. Critics diagnose, the coordinator adjudicates, subject notes own canon, and a no-finding report is valid.
- `memory-ai-ism-judgment` — AI-shaped prose is diagnosed by recurring construction and reading effect, never detector scores, word bans, or authorship claims. Kalsa's known pressure includes correction theatre, symmetrical wonder-then-cost ledgers, list-clause congestion, repeated frames, checksum conclusions, viewpoint choruses, workshop-dialect leakage, and generic smoothing; functional lists, deliberate rhetoric, and precise technical terms remain protectable.
## Open questions

Binary file not shown.

View file

@ -90,13 +90,17 @@ Before editing an institution, region, practice, or cross-cutting system:
2. Inventory seed statements, adopted facts, contradictions, and unknowns.
3. Name the canonical owner and model its inputs, outputs, dependencies,
beneficiaries, costs, failure modes, and neighboring interfaces.
4. Critique generic defaults, unsupported consensus, missing material support,
absent subordinate perspectives, and consequences that stop at the page.
5. Repair the smallest coherent set of owner notes.
6. Falsify the result with one hostile or marginal perspective and one
4. Assemble an immutable review packet and give every seat in
`workshop/review-council/README.md` an independent diagnostic turn.
5. Adjudicate every substantive finding; critics diagnose and the coordinator
decides the bounded repair brief.
6. Repair the smallest coherent set of owner notes.
7. Falsify the result with one hostile or marginal perspective and one
historical stress case.
7. Verify seed integrity, links, publication boundaries, and the Quartz build.
8. Record the pass in the ledger and promote only durable judgment into project
8. Give every council seat a post-repair regression turn; Reader Experience
receives a fresh blind packet.
9. Verify seed integrity, links, publication boundaries, and the Quartz build.
10. Record the pass in the ledger and promote only durable judgment into project
memory.
Passing prose is not enough. A mature institution must explain authority,

View file

@ -20,16 +20,23 @@ A pass must:
costs, failure modes, enforcement limits, and interfaces.
4. **Critique** generic fantasy defaults, missing material constraints, false
consensus, absent subordinate perspectives, and consequences that stop at
the page edge.
5. **Repair** the smallest coherent set of canonical owner notes. Delete or
the page edge. Assemble an immutable packet and collect independent reports
from every seat in the [[../review-council/README|Lore Review Council]].
5. **Adjudicate** every substantive council finding as accepted, modified,
rejected, or deferred. Preserve raw disagreement; do not vote lore into
existence or give reports directly to a repair writer.
6. **Repair** the smallest coherent set of canonical owner notes. Delete or
demote duplicate owners before adding compensating prose.
6. **Connect** the repair to people, places, events, resources, language,
7. **Connect** the repair to people, places, events, resources, language,
magic, and playable pressure.
7. **Falsify** the result through at least one hostile or marginal perspective
8. **Falsify** the result through at least one hostile or marginal perspective
and one historical stress case.
8. **Verify** seed hashes, wikilinks, publication boundaries, the Quartz build,
9. **Regress** the repaired candidate through every council seat. Reader
Experience receives a fresh blind packet; other critics receive their raw
findings and the relevant dispositions.
10. **Verify** seed hashes, wikilinks, publication boundaries, the Quartz build,
affected rendered pages, and the actual git diff.
9. **Record** the result in `ledger.md`, update `queue.md`, and promote only a
11. **Record** the result in `ledger.md`, update `queue.md`, and promote only a
durable operating lesson or decision into `.epiphany/project-memory.md`.
## Completion Standard
@ -48,6 +55,8 @@ not certify depth. The qualitative parity standard lives in `benchmark.md`.
- authority map before and after;
- unresolved questions and explicit non-decisions;
- hostile/marginal and historical stress findings;
- immutable review-packet identity, seven raw reports, finding dispositions,
repair brief, and seven regression reports;
- verification commands and results;
- rendered-page inspection when published content changed.

View file

@ -17,7 +17,9 @@ Review each major institution or system across these dimensions:
8. variation by class, region, lineage, language, and magical capacity;
9. historical crises and path dependence;
10. internal contradiction and active factions;
11. consequences when players help, harm, capture, expose, or reform it;
11. consequences from at least two plausible, target-supported player courses,
including a consequential choice or misunderstanding and a durable owner
or receipt;
12. provenance: adopted fact, attributed belief, inference, or unresolved design.
Use the following review scale as a discussion aid:
@ -76,6 +78,16 @@ Kalsa reaches institutional parity when:
- comparison against representative Aetheria notes finds comparable causal
density even when Kalsa uses fewer or shorter pages.
## Council Evidence
Depth parity is argued through the seven independent remits in
[[../review-council/README|the Lore Review Council]]. No aggregate score or
majority vote substitutes for the reports. Institutional causality, culture and
situated knowledge, material life, historical change, play consequence, prose,
and reader navigation must each receive a turn. Canon fidelity, provenance,
ownership, and spoiler safety constrain every remit and the final repository
verification rather than living in one continuity-shaped critic.
## Aetheria Reference Sample
Observed on 2026-08-22 in `F:\Projects\AetheriaLore\Aetheria\Worldbuilding`:

View file

@ -1,14 +1,45 @@
---
title: Kalsa Canonical Body Map
status: working-map
snapshot_at: 2026-08-22T17:26:42Z
scope: deepening-waves-1-through-16-plus-kalsa-026-kalsa-029-and-kalsa-030-boundaries
snapshot_at: 2026-08-22T19:54:33Z
scope: deepening-waves-1-through-16-plus-kalsa-026-kalsa-029-kalsa-030-and-kalsa-032-review-control
---
# Kalsa Canonical Body Map
This is an operational map of the live authoring machine, not lore and not a replacement for owner notes. It records institutional authority after KALSA-007 through KALSA-022, publication authority through KALSA-026, vault-level spoiler enclosure through KALSA-029, and Public epistemic ownership through KALSA-030. Waves 11–16 are repair overlays on existing subject owners, not new parallel authorities.
## Lore review and repair control
- **Owner:** `workshop/review-council/README.md` owns council membership,
sequencing, packet boundaries, and report gates. The active pass coordinator
owns finding dispositions, the bounded repair brief, integration, and
closeout. Canonical subject notes retain lore authority.
- **Inputs:** an immutable baseline packet, exact seed and canonical sources,
adopted authority maps, Public reader paths, protected strengths, operator
decisions, and one specialist prompt per council seat.
- **Outputs:** seven independent diagnostic reports, one adjudication, one
bounded repair brief, seven regression reports, and a recorded completion
decision.
- **Derived/display-only state:** critic confidence, agreement, pattern counts,
aggregate impressions, and report summaries are evidence. They may not write
canon, publication, or project memory.
- **Forbidden writers:** no critic, majority, detector, vocabulary list,
checklist, or generic Soul label may decide a setting claim or edit an owner.
Repair writers cannot silently average raw reports.
- **Shared paths:** frozen packet -> seven independent diagnoses -> coordinator
dispositions -> named-owner repair -> seven regressions -> ordinary seed,
link, layout, publication, Quartz, rendered-page, and diff verification.
- **Deletion/demotion line:** the former omnibus critique and undifferentiated
Soul checkbox are demoted. Canon fidelity, provenance, ownership, and spoiler
safety remain cross-cutting constraints and final checks, not a continuity
critic seat.
- **Verification layer:** report inventory, every-finding disposition, blind
reader packet where Public is in scope, no unresolved accepted essential
defect, negative authority checks, and the repository verification suite.
- **Unresolved seam:** report assembly and dispatch remain Markdown-first. No
runtime or automation owns them until repetition justifies that machinery.
## Canonical substrate
- **Owner:** `Kalsa/Spoilers/Foundations/` owns world, Ark/Fall, Prometheus, Kaos, Projection/God/thaumavore metaphysics, Gift taxonomy, and language/reckoning inheritance; `Kalsa/Spoilers/Reference/Canon and Provenance.md` owns canon-state vocabulary.

View file

@ -69,3 +69,43 @@ promoted or retired.
- **Verification:** nine immutable seed witnesses; 551 complete-vault wikilinks across 69 notes; 34 Public links across nine reader notes; publication, vault-layout, negative-tool, and exact-output checks; production Quartz build of nine inputs and 33 outputs; rendered inspection of the home, magic, and teaching-source pages.
- **Durable judgment promoted:** Public explanations require a situated inhabitant owner, evidence practice, institutional body, and characteristic limit. A local synonym without those consequences is only the author taxonomy in costume.
- **Open pressure:** KALSA-031 repeats this audit for the remaining traditions one bounded owner at a time; exact Ju'onai descendant-language forms, school antiquity, and regional variations remain unresolved.
## 2026-08-22 — Lore review council and AI-ism calibration
- **Scope:** [[passes/KALSA-032-review-council|KALSA-032]], establishing a
version-controlled seven-seat Lore Review Council for every future deepening
pass.
- **Authority cut:** critics diagnose independently; the coordinator adjudicates
and integrates; subject notes retain canon. A generic Soul checkbox,
consensus, detector, word list, or automatic rewrite cannot stand in for the
named reports.
- **Council:** Institutional Causality; Culture and Situated Knowledge;
Material Life and Ecology; History, Conflict, and Change; Ghostlight Play and
Consequence; AI-Isms and Prose; Reader Experience and Navigation. Every seat
returned a raw diagnosis and a regression on one frozen implementation; the
reader turns used blind Public packets.
- **First verdict:** Institutional Causality blocked completion because the
baseline omitted an in-scope hash, repaired candidates were not frozen, and
findings lacked stable identities. Other critics found informal-knowledge,
material-chain, event-evidence, play-access, universal-action-menu, and
term-origin seams.
- **Repair:** prompts and templates now preserve snapshot identity, coordinator
occupancy, finding IDs, informal reproduction paths, inhabitant-model limits,
material handoffs, evidence scope, event owners, bounded Ghostlight access,
target-earned interventions, play traces, and culturally owned technical
language.
- **Regression:** all accepted findings passed their original seats. Six
grounded critics verified every implementation hash; a fresh reader found no
spoiler or process-language leak and independently confirmed the deferred
Public voice and navigation pressure.
- **Canonical lore changed:** none. KALSA-031 and KALSA-024 carry bounded
cultural and historical pressure; KALSA-033 owns Public entry-route,
dead-end, and narrator-voice repair.
- **Verification:** seven diagnostic and seven regression artifacts; 23 memory
records projected to CultCache; nine seed witnesses; vault enclosure; 551
links across 69 canonical notes; Public-only source and output boundaries;
lore-tool negative tests; Quartz build of nine inputs and 33 outputs.
- **Durable judgment promoted:** review specialties must speak independently,
preserve raw disagreement, and regress the same repaired body. AI-shaped prose
is diagnosed through recurrent construction and reading effect, never
authorship claims, token bans, or detector scores.

View file

@ -3,6 +3,7 @@ pass_id: KALSA-000
status: draft
target: replace-me
canonical_owner: unresolved
coordinator: replace-me
started: YYYY-MM-DD
completed:
---
@ -63,6 +64,32 @@ What real setting and play outcome does this pass serve?
- Consequences that stop at the page edge:
- Duplicate or ambiguous owners:
## Review council
- **Packet identity:**
- **Repaired packet identity:**
- **Packet path:**
- **Protected strengths:**
- **In-scope inhabitant model boundary:** owner or synthesis; evidence and
reproduction; useful domain; blind spot; protected unknown.
- **Ghostlight lens:** point of contact; character-accessible evidence;
target-supported interventions; consequence horizon or play-bearing owner.
- **Event-owner evidence or explicit evidence gap:**
| Critic | Diagnostic report | Verdict | Regression report |
| --- | --- | --- | --- |
| Institutional Causality | | | |
| Culture and Situated Knowledge | | | |
| Material Life and Ecology | | | |
| History, Conflict, and Change | | | |
| Ghostlight Play and Consequence | | | |
| AI-Isms and Prose | | | |
| Reader Experience and Navigation | | | |
- **Synthesis and dispositions:**
- **Accepted repair brief:**
- **Deferred pressure added to queue:**
## Repair
- Canonical files to change:
@ -101,6 +128,10 @@ claimed mechanism?
- [ ] Affected rendered pages were inspected.
- [ ] Git diff changes only intended owners and supporting indexes.
- [ ] Benchmark review completed without treating counts as proof.
- [ ] All seven independent diagnostic reports exist.
- [ ] Every substantive finding has a coordinator disposition.
- [ ] All seven regression reports exist; no essential accepted defect remains.
- [ ] The final closeout names what should now be left alone.
## Decision record

View file

@ -0,0 +1,191 @@
---
pass_id: KALSA-032
status: complete
target: mandatory deepening review council and AI-ism prose critic
canonical_owner: workshop/review-council/README.md
coordinator: Codex /root
started: 2026-08-22
completed: 2026-08-22
---
# KALSA-032 — Deepening Review Council
## Objective
Make every critique/repair pass answer to several independent editorial
faculties before completion. Add one critic dedicated to recurring AI-shaped
prose habits, grounded in the current Vault rather than a generic banned-word
list.
## Source inventory
### External and neighboring evidence
- Wikipedia's `Signs of AI writing` field guide is descriptive rather than
prescriptive. It identifies clusters such as superficial significance,
vague attribution, negative parallelism, rule-of-three lists, inflated
copular substitutes, abstract associations, repeated formatting templates,
and excessive rhetorical symmetry. It warns that signs are not proof and
that detectors and human guesses produce false positives.
- AetheriaLore's live `Vault Cleanup` contract separates Modeling, Imagination,
fresh-reader questioning, source-bounded lore research, Hands, and Soul.
Its `Revision Council Master Prompt`, `Revision Council Specialist Prompts`,
and `Revision Council Passes` provide the reusable independent-report,
coordinator-adjudication, bounded-repair, regression, and fresh-reader shape.
Kalsa does not copy its novel-specific specialties.
### Kalsa evidence
- Public and author prose repeatedly uses corrective antithesis, especially
`not X; Y`, `rather than`, and `can ... and still ...`.
- Public summaries often praise a wonder, immediately balance it with social
costs, and close with a list-shaped moral checksum. The Ti'asantatca passage
in `Public/World/Peoples and Polities.md` is the clearest operator-identified
example.
- Lists of four to eight nouns recur as a signal of comprehensiveness even when
one concrete consequence would carry more force.
- Authoring vocabulary such as `authority`, `bounded`, `material`, `legible`,
`claim`, `record`, `preserve`, and `consequence` is load-bearing in workshop
maps but can flatten culture, setting prose, and character speech into one
institutional essay voice.
- Existing Soul review is strong at truth, ownership, and boundary checks but
has no durable prompt surface for prose uncanniness or reader fatigue.
## Authority map before repair
- **Owner:** the pass coordinator informally decides which reviews to request
and how to interpret them.
- **Inputs:** seed evidence, owner notes, pass records, diff, benchmark, and
occasional ad hoc specialist findings.
- **Outputs:** one synthesis and a final Soul verdict.
- **Derived state:** commentary, agent reports, and grep inventories may reveal
pressure but have no standard artifact or completion gate.
- **Forbidden writers:** reviewers may not write canon, but this is convention
rather than an explicit council contract.
- **Shared paths:** no required path guarantees that every dedicated critic
receives the same candidate or speaks before repair.
- **Cut line:** retire ceremonial `Soul review` checkboxes that can be satisfied
by one undifferentiated opinion.
## Process model
- **Function:** seven independent lore critics inspect one immutable review packet;
the coordinator adjudicates their findings; the same seven receive the
repaired candidate for bounded regression.
- **Authority and enforcement limits:** critics diagnose and recommend. They do
not edit canon, vote truth into existence, or compel a rewrite by majority.
- **Material support:** stable prompts, one shared report contract, raw reports,
synthesis, dispositions, regression reports, and the pass record.
- **Variation:** Reader Experience receives reader-visible prose without
workshop intention or hidden canon. Other critics receive only the context
necessary for their remit.
- **Internal contradiction:** requiring all voices can become ceremony. A
critic may return `no finding`; the coordinator must not manufacture work to
justify the role.
## Repair
- Add `workshop/review-council/` as the process owner.
- Define seven dedicated critics: institutional causality; culture and situated
knowledge; material life and ecology; history, conflict, and change;
Ghostlight play and consequence; AI-isms and prose; and reader experience
and navigation.
- Give the AI-Isms and Prose critic a Kalsa-specific calibration without
converting examples into forbidden strings.
- Require raw independent reports, coordinator dispositions, and post-repair
regression from every critic in each deepening loop.
- Integrate the council into the deepening README, pass template, benchmark,
Body map, project instructions, and project memory.
## Review council
- **Packet identity:** `KALSA-032-baseline-01`; candidate blob hashes are
recorded in the packet.
- **Repaired packet identity:** `KALSA-032-repaired-02`; all six grounded seats
verified its 21 blobs and a final blind reader passed the unchanged Public
surface. See [[../reviews/KALSA-032/final-revalidation]].
- **Packet path:** `workshop/deepening/reviews/KALSA-032/baseline-packet.md`;
Reader Experience used `reader-packet.md`.
- **Protected strengths:** seven distinct lore remits, independent speech,
explicit no-finding reports, blind Public reading, coordinator authority,
canon boundaries, seed specificity, and contextual rather than detector-led
prose criticism.
| Critic | Diagnostic report | Verdict | Regression report |
| --- | --- | --- | --- |
| Institutional Causality | [[../reviews/KALSA-032/diagnosis/Institutional Causality]] | Block | [[../reviews/KALSA-032/regression/Institutional Causality]] — Pass |
| Culture and Situated Knowledge | [[../reviews/KALSA-032/diagnosis/Culture and Situated Knowledge]] | Pass with findings | [[../reviews/KALSA-032/regression/Culture and Situated Knowledge]] — Pass |
| Material Life and Ecology | [[../reviews/KALSA-032/diagnosis/Material Life and Ecology]] | Pass with findings | [[../reviews/KALSA-032/regression/Material Life and Ecology]] — Pass |
| History, Conflict, and Change | [[../reviews/KALSA-032/diagnosis/History Conflict and Change]] | Pass with findings | [[../reviews/KALSA-032/regression/History Conflict and Change]] — Pass |
| Ghostlight Play and Consequence | [[../reviews/KALSA-032/diagnosis/Ghostlight Play and Consequence]] | Pass with findings | [[../reviews/KALSA-032/regression/Ghostlight Play and Consequence]] — Pass |
| AI-Isms and Prose | [[../reviews/KALSA-032/diagnosis/AI-Isms and Prose]] | Pass with findings | [[../reviews/KALSA-032/regression/AI-Isms and Prose]] — Pass |
| Reader Experience and Navigation | [[../reviews/KALSA-032/diagnosis/Reader Experience and Navigation]] | Pass with findings | [[../reviews/KALSA-032/regression/Reader Experience and Navigation]] — Pass with deferred findings |
- **Synthesis and dispositions:** [[../reviews/KALSA-032/synthesis]].
- **Accepted repair brief:** harden snapshot identity, finding IDs,
coordinator provenance, informal knowledge transmission, inhabitant-model
boundaries, material chains, event-owner evidence, Ghostlight access and play
traces, target-supported interventions, and AI-ism term-origin analysis.
- **Deferred pressure added to queue:** KALSA-031 and KALSA-024 were narrowed;
KALSA-033 owns Public entry-route, dead-end, and narrator-voice repair.
## Falsification
### Hostile perspective
A competent human writer uses antithesis, a three-item list, or the word
`bounded` because the sentence needs it. The critic must be able to protect the
passage and explain why local function outweighs statistical resemblance.
### Stress case
Seven critics disagree. The coordinator preserves every raw report, rules on
each substantive finding, and repairs only accepted defects. Absence of
consensus does not create split ownership; absence of a critic report blocks
completion.
## Authority map after repair
- **Owner:** `workshop/review-council/README.md` owns council membership,
sequencing, packet boundaries, and report requirements. The pass coordinator
owns adjudication. Canonical subject notes retain lore authority.
- **Inputs:** one immutable candidate packet, protected strengths, target
audience, exact source authority, and bounded context per critic.
- **Outputs:** seven raw reports, one synthesis with dispositions, seven
regression responses, and a finite repair list.
- **Derived state:** pattern counts, detector scores, consensus, and critic
confidence are evidence only.
- **Forbidden writers:** no critic, report, detector, word list, or majority
may edit canon or decide truth.
- **Shared paths:** packet → seven independent reports → coordinator
synthesis and dispositions → owner repair → seven regressions → final
verification and record.
- **Deleted or demoted authority:** an unchecked or generic `Soul review` label
no longer stands in for the council.
## Verification
- [x] All council prompts exist and name a distinct remit.
- [x] Every critic has a required report and regression contract.
- [x] Reader Experience has a blind packet boundary.
- [x] The AI-ism critic treats patterns as contextual evidence, not proof or
forbidden vocabulary.
- [x] Kalsa-specific recurring patterns appear in its calibration.
- [x] Deepening workflow and pass template require every critic to speak.
- [x] Seed hashes, links, publication boundary, and production build remain
valid.
- [x] Independent council regression finds no process ownership defect.
## Decision record
- **Adopted:** a seven-critic review council with coordinator adjudication.
- **Explicitly not adopted:** detector scores, automatic style gates, majority
voting, automatic critic edits, a dedicated continuity critic, or
novel-specific character, POV, chapter, and scene machinery.
- **Unresolved:** whether later automation should assemble packets and reports;
the process remains Markdown-first until a runtime owner exists.
- **Project-memory judgment promoted or retired:** promoted
`memory-independent-lore-review` and `memory-ai-ism-judgment`; regenerated the
CultCache projection.
- **Next highest-value pressure:** run the council against the Public vault and
repair accepted prose findings in a separate bounded pass.

View file

@ -14,14 +14,16 @@ play value.
| KALSA-006 | First society deepening pilot | One complete society pass will test the rubric, provenance scheme, and gameplay handoff before the loop scales. | complete and scaled | Ten owner clusters, five repair waves, and one direct-entry dungeon now exercise the loop. |
| KALSA-022 | Final ownership repair | Candidate leakage and two split event owners remained after the second Soul pass. | complete | Alliance quarantined; Broken Windtrap and Grey Scald gained factual owners. |
| KALSA-023 | Intuition polity and Alliance counterparty | The Alliance cannot reach canonical parity while one reciprocal owner is a blank source promise. | blocked | Source-grounded Intuition design or operator direction |
| KALSA-024 | Named regional variants | Current machinery is usable; future passes should add concrete houses, lineages, shrines, crews, routes, and settlements only where play creates pressure. | queued | Actual campaign or setting-use demand |
| KALSA-024 | Named regional variants | Current machinery is usable; future passes should add concrete houses, lineages, shrines, crews, routes, settlements, and mixed or cross-boundary lives only where play creates pressure. | queued | Actual campaign or setting-use demand |
| KALSA-025 | Final Soul parity verdict | The first two Soul passes rejected specific defects; the final bounded pass verified their exact removal. | complete | Parity defensible for the adopted institutional set. |
| KALSA-026 | Knowledge and spoiler architecture | The first publication collapsed author truth, reader lore, character knowledge, and GM answers into one surface. | complete | Eight situated reader notes, default-deny Quartz input, generated-route checks, and bounded Soul review passed. |
| KALSA-027 | Kalsa world-compiler integration | Ghostlight needs world truth, retrieval evidence, and epistemic projections with typed affordances and leakage boundaries. | blocked | Ghostlight-owned compiler/runtime implementation and shared schema decision |
| KALSA-028 | Spectacular surface adventure | The reader vault needs a magic-forward high-fantasy play proof alongside the industrial Ashen Intake. | queued | Bounded adventure selection and campaign demand |
| KALSA-029 | Vault spoiler enclosure | The Quartz boundary was safe, but casual Obsidian exploration still exposed author domains and session recents. | complete | Author/GM tree enclosed, root gate neutral, workspace neutral and ignored, and negative checks passed. |
| KALSA-030 | Inhabitant models and public epistemic ownership | The physical spoiler boundary is sound, but author mechanics still need culturally owned names, schools, lineages, disputes, and incomplete models before Public can become an inhabitant-facing knowledge commons. | complete | Ju'onai causal scholarship now owns openings, four rival lineages, attestation and administration remain separate authorities, and Public projects the dispute without exposing the author taxonomy. |
| KALSA-031 | Remaining situated magical traditions | The Ju'onai pass proves the method, but Selza'a, Jamnai, Papsenai, prophetic, Sarxe, divine, and technical explanations still need the same owner/evidence/institution/failure audit where author labels substitute for local thought. | queued | Work one bounded tradition at a time, prioritized by Public exposure and Ghostlight play pressure; do not invent the unresolved Intuition polity through this pass. |
| KALSA-031 | Remaining situated magical traditions | The Ju'onai pass proves the method, but Selza'a, Jamnai, Papsenai, prophetic, Sarxe, divine, and technical explanations still need the same owner/evidence/reproduction/failure audit where author labels substitute for local thought. | queued | Work one bounded tradition at a time, include one culturally owned non-magical relation or practice where evidence supports it, and inspect source-grounded historical formation without inventing an event or the unresolved Intuition polity. |
| KALSA-032 | Lore review council and AI-ism critic | Deepening passes need independent criticism across distinct lore-quality surfaces, including the recurring correction-theatre, moral-ledger, list-clause, and workshop-dialect prose habits found in the current vault. | complete | Seven versioned prompts, independent diagnoses, adjudication, same-snapshot regressions, memory, and verification passed. |
| KALSA-033 | Public entry route and narrator-voice repair | Blind reader and AI-ism calibration found a survey-first entrance, compelling dead-end glimpses, repeated narrator-level institutional cautions, and a praise-then-cost/list-clause cadence that flattens otherwise situated material. | queued | Freeze a Public packet; inspect canonical owners before changing presentation; preserve spoiler friction, Ju'onai's earned register, Low Sere mystery, and the two situated sources. |
Do not fill the queue with every imaginable fantasy topic. Add an item when a
real owner, dependency, contradiction, or play requirement creates pressure.

View file

@ -0,0 +1,97 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
candidate_identity: working-tree-blob-hashes-recorded-below
audience: future Kalsa pass coordinators and critic agents
public_blind_packet: false
---
# KALSA-032 Council Calibration Packet
## Objective and scope
- **Target:** the proposed Kalsa Lore Review Council, its seven remits, and its
integration into every institutional deepening pass.
- **Process owner:** `workshop/review-council/README.md`.
- **Files in scope:** the council README, shared prompt, seven critic prompts,
packet/report/synthesis templates, and the deepening README, pass template,
benchmark, queue, project instructions, and KALSA-032 pass record.
- **Calibration corpus:** `Kalsa/Public/index.md`, all notes under
`Kalsa/Public/World/`, and `Kalsa/Public/Sources/Four Readings of a Broken
Sluice.md`.
- **Repair boundary:** this pass may repair only workshop process, project
instructions, Body map, and project memory. Lore findings become named queue
pressure for a later bounded pass.
## Authority and provenance
- Immutable seed evidence remains under `seed/original/` and is not a process
edit target.
- Canonical subject notes retain lore authority. The council diagnoses; the
pass coordinator adjudicates; Hands repairs only named owners.
- `Kalsa/Public/` owns situated presentation. `Kalsa/Spoilers/` holds author and
GM truth. Workshop documents cannot settle setting facts.
- The protocol shape is adapted from AetheriaLore's Novella Council. Novel-only
roles and a dedicated continuity critic are intentionally omitted.
- Canon fidelity, provenance, ownership, and spoiler safety are cross-cutting
constraints and final verification duties rather than a council seat.
## Protected strengths
- Seven genuinely different lore-quality remits, not seven renamed general
editors.
- Every critic speaks independently before repair and again after repair.
- A no-finding report is valid; ceremony and manufactured findings are not.
- Critics preserve successes, identify collateral risk, and name an edit to
resist.
- Critics cannot write canon or decide by vote.
- Reader Experience uses a blind reader-visible packet when Public is in scope.
- The AI-Isms critic diagnoses recurrence and effect without detector scores,
word bans, authorship claims, or regex tribunals.
## Known concerns
- Distinct specialties can still overlap into seven versions of institutional
analysis.
- Requiring fourteen reports may become ceremony if packet and output contracts
are vague.
- The AI-Isms prompt must catch Kalsa's recurring corrective antithesis,
praise-then-cost ledger, list-clause density, repeated frames, checksum
conclusions, viewpoint chorus, and workshop dialect without banning useful
rhetoric or technical vocabulary.
- Lore-quality review must not drift into novel-specific continuity, scene,
character-arc, or chapter-pacing machinery.
## Candidate identity
The following Git blob hashes freeze the baseline reviewed by the council:
```text
260b2e6cbba921625877228d92258dac7796b879 AGENTS.md
269904dde2d51d376fbfe6ba31276769f7ae8b7c workshop/deepening/README.md
0d280916d3175b07f57abdeb68619d72418b31c7 workshop/deepening/pass-template.md
942c71c050b655c1cb2abefb239764ea29ae8ee4 workshop/deepening/benchmark.md
2840a1b25c87d9290972e0b5132996ff50104a18 workshop/deepening/queue.md
82ceb03ec6cc95025833d80eccbafb24d8417cf4 workshop/deepening/passes/KALSA-032-review-council.md
4745e31d035e0d5bdf39c9f76ea05fb3e848e571 workshop/review-council/README.md
5ed604ff74fa93aa6bfbab1a2c73c7fd91e9ab50 workshop/review-council/shared-prompt.md
9e4c2a6fd7286b0e668b0fb69f473c2307885ac8 workshop/review-council/packet-template.md
622edf3b50ce763a71369274fc672573366d2f20 workshop/review-council/report-template.md
4761b44f333e56f111433bf6a096d100374d7c91 workshop/review-council/synthesis-template.md
d96d41f25259ac367395db8e4787004d71f8768f workshop/review-council/critics/Institutional Causality.md
980caf870b582b79709eacc20890b018a6401f36 workshop/review-council/critics/Culture and Situated Knowledge.md
a063984af257f2e3cc679cf0f7ddbd83194f31d2 workshop/review-council/critics/Material Life and Ecology.md
d33b34e1e13f7147893c02482589db6dd99f67a3 workshop/review-council/critics/History Conflict and Change.md
407cc738bad49fa012452a505cbc55bb8589d64f workshop/review-council/critics/Ghostlight Play and Consequence.md
db381663c0cbc0817417f8b8154bd9f5eb284a86 workshop/review-council/critics/AI-Isms and Prose.md
eeb0b26d2a53407f21766598cd337427dfe54346 workshop/review-council/critics/Reader Experience and Navigation.md
```
## Evidence limitations
- This is a process calibration, not a claim that the whole Public vault has
received a final prose or lore repair.
- The exact AetheriaLore vault-critic prompts were not recovered; only the live
cleanup contract and Novella Council protocol were inspected.
- Current Public findings may identify later pressure but cannot authorize lore
edits in KALSA-032.

View file

@ -0,0 +1,89 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
critic: AI-Isms and Prose
phase: diagnosis
verdict: pass-with-findings
---
# Critic Report
## Concise diagnosis
**Pass with findings. No blocking finding.** The prompt recognizes Kalsa's
actual recurring prose machinery and rejects detector logic, isolated-signal
verdicts, and word bans. One calibration seam remains: distinguish workshop
dialect from culturally owned technical language.
## Protected successes
- The prompt requires clusters, recurrence, context, and reading effect.
- Named patterns match the corpus: Ti'asantatca's praise-then-invoice paragraph,
corrective antithesis, list-clause congestion, overview chorus, and checksum
endings.
- It protects functional inventories, deliberate antithesis,
culture-specific rhetoric, and precise technical terms.
## Findings
### AIP-01
- **Priority:** High-value.
- **Location:** AI-Isms prompt, Workshop dialect leakage bullet and final
protection paragraph.
- **Evidence:** The corpus contains both genuine leakage and situated use of
the same vocabulary. `Public/index.md` closes in workshop-adjacent language
about situated learning and final words. Ju'onai passages use bounded, claim,
record, and institutional distinctions as owned intellectual practice.
`Four Readings` makes that register dramatic, contested, and speaker-owned.
- **Effect:** A critic could strip earned vocabulary from the Ju'onai while
failing to identify narrator-level diffusion as the defect.
- **Repair direction:** Require a term-origin test: narrator default, named
school or office, or particular speaker. Diagnose diffusion across unrelated
voices, not vocabulary alone.
- **Collateral risk:** Do not undo KALSA-030 or simplify the sluice dialogue.
- **Dependency:** Culture and Situated Knowledge.
- **Confidence:** High.
### AIP-02
- **Priority:** Optional.
- **Location:** report template Findings, compared with the AI-Isms prompt's
final paragraph.
- **Evidence:** The specialist prompt requires recurrence evidence and retained
function; the shared fields do not name those obligations explicitly.
- **Effect:** A report could offer isolated stylistic dislikes or deletion
without preserving function.
- **Repair direction:** State in the specialist prompt that recurrence or
contrast belongs under Evidence, and retained function under Collateral risk.
Do not enlarge the shared template unless other seats need it.
- **Collateral risk:** Extra fields can turn the report into ceremony.
- **Dependency:** Process owner.
- **Confidence:** Medium.
## Pattern or dependency map
- **Correction theatre:** Public entrance and Relics and Ruins use recurring
`can ... and still` and `It does not` structures.
- **Symmetrical moral ledger:** Ti'asantatca is clearest; related praise-cost
cadences appear at the Public entrance, Hegemony, and mobility passages.
- **List-clause congestion:** common across Gods, Kalsa, Magic, and Peoples;
several lists remain functional.
- **Repeated frames and viewpoint chorus:** Magic traditions and polity
summaries give unrelated traditions one essay rhythm.
- **Checksum conclusions:** “Kalsa's wonder is not elsewhere,” the final Relics
aphorism, and Magic's “most dangerous sentence” explain prior examples.
- **Likely false positives:** cult-support and delving-compact inventories, and
corrective exchanges in `Four Readings`.
## Tempting edit to resist
Do not ban `not`, `bounded`, `claim`, `record`, or parallel clauses. Do not
“de-AI” `Four Readings of a Broken Sluice`; its antitheses belong to speakers
with incompatible stakes.
## Ranked actions
1. No essential action.
2. Add the term-origin test.
3. Clarify recurrence evidence and retained function inside existing fields.

View file

@ -0,0 +1,125 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
critic: Culture and Situated Knowledge
phase: diagnosis
verdict: pass-with-findings
---
# Critic Report
## Concise diagnosis
**Pass with findings. No blocking finding.** The council explicitly protects
culturally owned disagreement, incomplete but usable theories, opacity, and
non-omniscient public knowledge. Its strongest calibration example—the broken
sluice dialogue—proves that this can work. Two handoff defects could still pull
future passes toward formal institutions and magic-indexed cultures.
## Protected successes
- The critic directly tests whether people exist beyond magic category, polity,
occupation, assigned psychology, and player arrival. It protects opacity and
unresolved belief instead of equating incompleteness with failure.
- The Situated Knowledge Benchmark gives incomplete models a rigorous shape:
evidence, owner, reproduction, useful domain, rival accounts, and
characteristic blind spot. “A local synonym for an author label scores
absent” defends against laundering backstage taxonomy.
- `Four Readings of a Broken Sluice.md` situates knowledge in circulating
client-college copies, trade-speech translation, rival schools,
administrative separation, and a keeper whose material objection survives
the schools' explanations.
- `Magic and Miracles.md` distinguishes what traditions can notice from what
they cannot establish; it grants none the author's final word.
## Findings
### CSK-01
- **Priority:** High-value.
- **Location:** `critics/Culture and Situated Knowledge.md`, paragraph beginning
“For every public explanation”; calibration examples in `Kalsa.md`, Crown
Cities and Roads through twilight.
- **Evidence:** The prompt requires every public explanation to have an
inhabitant owner, then asks for its “teaching body, patronage, instruments or
rites.” The corpus also contains diffuse knowledge: popular stories combine
rival Crown accounts; travellers learn warning marks, threshold speech, host
oaths, and route customs without a single school or patron.
- **Effect:** A literal critic may force household, oral, itinerant, customary,
or anonymous accumulated knowledge into a formal institution-shaped model.
- **Repair direction:** Trace the actual reproduction path: household
transmission, apprenticeship, oral repetition, seasonal gathering, market
circulation, ritual, travel, testimony, or formal teaching. Patronage and
instruments apply only where the model claims them.
- **Collateral risk:** Do not weaken the requirement for a legible inhabitant
source or let “common knowledge” become an omniscient author voice.
- **Dependency:** Institutional Causality where a formal school or office exists.
- **Confidence:** High.
### CSK-02
- **Priority:** High-value.
- **Location:** `packet-template.md`, Authority and provenance;
`pass-template.md`, Institutional model and Review council.
- **Evidence:** The benchmark asks for owner, evidence, reproduction, useful
domain, rivals, blind spot, and unknowns, but the packet template offers only
one line for Public inhabitant presentation. The pass record has no durable
place to carry the tested boundary into adjudication and repair.
- **Effect:** The coordinator may repair a visible gap by completing the theory
from author knowledge. The good criterion exists, but its handoff is lossy.
- **Repair direction:** For in-scope public explanatory models, record
inhabitant owner or synthesis, evidence and reproduction path, useful domain,
characteristic blind spot, and protected unknown. Carry accepted boundaries
into protected material or the repair brief.
- **Collateral risk:** Do not inventory the whole vault, disclose hidden
taxonomy, or turn folk observation into schema ceremony.
- **Dependency:** Reader Experience, provenance, and spoiler verification.
- **Confidence:** High.
### CSK-03
- **Priority:** Optional.
- **Location:** `queue.md`, KALSA-024 and KALSA-031; `Peoples and Polities.md`;
`Magic and Miracles.md`, What different traditions say.
- **Evidence:** Public presentation still introduces most named peoples through
polity and magical capacity. It contains useful counterevidence, but
KALSA-031 queues remaining magical traditions while KALSA-024 defers concrete
social variants until campaign demand.
- **Effect:** Successive passes could deepen magical explanations while leaving
humor, inheritance, ordinary aspiration, translation, class idiom, and mixed
lives as supporting texture around magic.
- **Repair direction:** Amend existing queued pressure: each situated-tradition
pass should test one non-magical relation or practice owned by the same
people, and KALSA-024 should include mixed and cross-boundary lives when play
creates pressure.
- **Collateral risk:** Do not invent the unresolved Intuition polity, demand
symmetrical coverage, or add generic festivals and cuisine.
- **Dependency:** History, Conflict, and Change; Material Life and Ecology.
- **Confidence:** Medium.
## Pattern or dependency map
The weak seam is transmission:
`bounded inhabitant model -> packet summary -> coordinator disposition ->
repair brief`
The packet and pass templates do not yet preserve the model's reproduction path
and protected limit. Queue selection also begins from magical traditions, so
ordinary culture depends on the critic pushing beyond the initial category.
## Tempting edit to resist
Do not require every people to receive matching sections for humor, kinship,
ritual, class, food, dissent, and aspiration. That would replace magic-category
flattening with checklist-category flattening. Protect uneven emphasis, diffuse
knowledge, and the broken-sluice dialogue's refusal to provide a final page.
## Ranked actions
1. No essential action.
2. Broaden formal reproduction language and preserve each in-scope model's
owner, evidence path, useful domain, blind spot, and protected unknown
through packet and repair handoff.
3. Tie KALSA-031 and KALSA-024 together without demanding symmetrical cultural
completion.

View file

@ -0,0 +1,102 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
critic: Ghostlight Play and Consequence
phase: diagnosis
verdict: pass-with-findings
---
# Critic Report
## Concise diagnosis
**Pass with findings.** The council creates real Ghostlight pressure: it asks
what characters can perceive and alter, demands durable consequences, and
protects uncertainty and inhabitant agency. Two seams could still turn that
pressure into generic quest menus: packets lack a bounded player-access lens,
while benchmark and prompt repeat a standard suite of intervention verbs.
## Protected successes
- The prompt tests what players can notice, attempt, misunderstand, change,
lose, carry away, or make worse without author knowledge.
- It protects uncertainty, asymmetric knowledge, unsolved problems, and lives
that do not begin when players arrive.
- `Relics and Ruins.md` demonstrates thresholds, claimants, irreversible acts,
incomplete evidence, refusal, rescue obligations, and durable receipts.
- `Four Readings of a Broken Sluice.md` provides a non-quest-shaped case: a
failed work, partial explanations, missing repair authority, material loss,
and a keeper who makes knowledge answer to a dry field.
## Findings
### GPC-01
- **Priority:** High-value.
- **Location:** packet template Objective and scope; council README Packet Contract.
- **Evidence:** Packets name target, owner, files, audience, and reader path,
but not a plausible character-facing entry, accessible evidence, available
kinds of intervention, or consequence horizon.
- **Effect:** The critic must assume capabilities from author context or produce
free-floating “players could” hooks.
- **Repair direction:** Add a small Ghostlight lens: plausible point of contact,
evidence available without backstage truth, and expected consequence horizon.
Permit an overview to route to a play-bearing owner.
- **Collateral risk:** Do not create adventure briefs, prescribe party
composition, or disclose GM answers.
- **Dependency:** Spoiler and knowledge boundaries; Reader Experience owns
navigation, not character capability.
- **Confidence:** High.
### GPC-02
- **Priority:** High-value.
- **Location:** benchmark qualitative dimension 11; Ghostlight critic paragraph
beginning “Flag passive encyclopedia texture”.
- **Evidence:** The benchmark asks what happens when players help, harm,
capture, expose, or reform every target. The critic adds helping, exposing,
stealing, joining, reforming, and destroying.
- **Effect:** Each target may answer the same verb checklist even when its real
pressure is passage, testimony, maintenance, refusal, rescue, bargaining, or
accepting irreversible cost.
- **Repair direction:** Test only target-supported interventions. Require one
consequential choice or misunderstanding, a materially different response to
another plausible course, and a durable owner or receipt.
- **Collateral risk:** Do not weaken branching consequences or leave
institutions as untouchable encyclopedia objects.
- **Dependency:** Institutional Causality and this seat's player-visible remit.
- **Confidence:** High.
### GPC-03
- **Priority:** Optional.
- **Location:** Ghostlight critic opening paragraph; report template Findings.
- **Evidence:** The remit names routes, actors, clocks, resources, hazards, and
bargains, but does not require one tested interaction.
- **Effect:** Reports may inventory missing ingredients without showing where
play becomes inert.
- **Repair direction:** When evidence permits, trace observable cue -> attempted
action or refusal -> resistance or misreading -> responding owner -> durable
consequence or receipt.
- **Collateral risk:** Do not demand a complete encounter, balanced scenario,
or single solution.
- **Dependency:** Institutional and Material mechanisms.
- **Confidence:** Medium.
## Pattern or dependency map
`orientation -> playable locus -> consequence-bearing interaction`
The council should test that progression rather than demand that every page
perform all three jobs.
## Tempting edit to resist
Do not add Adventure Hooks headings, hook quotas, mandatory clocks, or a full
help/harm/join/destroy matrix to every lore note.
## Ranked actions
1. No essential action.
2. Add the bounded play-access lens and use target-earned interventions.
3. Calibrate findings around one evidence-supported play trace.

View file

@ -0,0 +1,95 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
critic: History, Conflict, and Change
phase: diagnosis
verdict: pass-with-findings
---
# Critic Report
## Concise diagnosis
**Pass with findings.** The remit is distinct and useful: it tests formation,
path dependence, rival memory, and whether old conflicts still shape present
choices. It protects disputed chronology and does not demand one authoritative
public history. One packet-boundary gap prevents the factual-event ownership
check from being fully exercised.
## Protected successes
- The prompt requires one factual event owner while neighbors retain evidence,
memory, interpretation, and consequence.
- `Peoples and Polities.md`, under Sarxe and Saxfoldi, carries path dependence
from conquest and massacre into a mobile political center, distributed
survival, and continuing mobile/settled conflict.
- `Gods and the Dead.md`, under Covenant and submission, makes visually similar
rites carry incompatible histories of consent, inheritance, and conquest.
- `Four Readings of a Broken Sluice.md` preserves transmission, alteration, and
disagreement while keeping the repair dispute alive.
## Findings
### HCC-01
- **Priority:** High-value.
- **Location:** council README Packet Contract; packet template Objective and
scope and Authority and provenance; KALSA-032 calibration corpus.
- **Evidence:** The critic prompt requires checking that adopted event history
has one factual owner. The calibration packet supplies Public presentation
but no canonical event-owner notes for the Sarxe dispersals, divine
conquests, Crown histories, or Divided City conquest.
- **Effect:** The seat can test public memory and present consequences, but
cannot verify singular, duplicated, or absent factual ownership.
- **Repair direction:** When event ownership is under review, supply the
relevant owner and neighboring memory/consequence notes, or explicitly state
that ownership evidence is unavailable and outside the verdict.
- **Collateral risk:** Do not expose hidden canon to the blind reader, require
every pass to invent an event note, or turn factual ownership into one public
interpretation.
- **Dependency:** Process owner, provenance, and spoiler verification.
- **Confidence:** High.
### HCC-02
- **Priority:** High-value, deferred lore pressure.
- **Location:** `Peoples and Polities.md`, Papsenai and Ti'asantatca, The Ju'onai
Hegemony, and The divided city.
- **Evidence:** These sections establish inherited shield service,
conquest-backed administration, levy conflict, overlapping imperial power,
and restricted prophetic knowledge, with little evidence of formation,
crises, failed reforms, or rival memories.
- **Effect:** The conflicts are playable and material, but some institutions
arrive as polished present-tense arrangements; their likely directions of
change are harder to derive.
- **Repair direction:** Queue later bounded owner passes to recover one
source-supported hinge, failed settlement, inherited compromise, or rival
memory. Invent no event merely to satisfy the rubric.
- **Collateral risk:** Do not produce an omniscient timeline, symmetrical
histories, or backstage explanation in compact Public pages.
- **Dependency:** Canonical owners; Culture and Situated Knowledge;
Institutional Causality.
- **Confidence:** Medium-high; hidden owner notes were outside this packet.
## Pattern or dependency map
- Conquest or massacre -> Sarxe memory -> mobile Saxfoldi -> distributed
survival and settled/mobile conflict.
- Chosen, inherited, or imposed covenant -> similar rite -> incompatible
obligations and consent claims.
- Layered occupation and adaptation -> ruin infrastructure -> current water,
burial, residency, and salvage claims.
- Present conflict without supplied formation -> Hegemony, Divided City, and
Ti'asantatca -> strong pressure but weaker trajectory.
## Tempting edit to resist
Do not reconcile Crown Cities, divine covenants, Sarxe dispersals, or broken
sluice traditions into one public chronology. The useful repair is culturally
owned memory tied to consequence, not an authorial master timeline.
## Ranked actions
1. No essential action.
2. Make event-owner evidence or its explicit absence part of the packet.
3. Defer Public historical pressure to later canonical-owner passes.

View file

@ -0,0 +1,142 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
critic: Institutional Causality
phase: diagnosis
verdict: block
---
# Critic Report
## Concise diagnosis
**Block.** The council's authority boundaries and specialist remit are sound,
but the current protocol does not preserve one immutable institutional body
across either review cycle. It also lacks stable finding identities through
adjudication and regression. Those defects make completion reproducible only by
trust in the coordinator's memory.
## Protected successes
- `workshop/review-council/README.md`, Authority Map through Repair Phase,
cleanly separates diagnosis, adjudication, repair, and canonical ownership.
Critics cannot decide lore, and majority agreement remains evidence rather
than authority.
- `critics/Institutional Causality.md` tests authorization, labor,
reproduction, enforcement limits, and failed handoffs without demanding a
universal bureaucracy. The calibration corpus proves the remit is specific:
`Relics and Ruins.md` separates technical competence from ownership, while
`Four Readings of a Broken Sluice.md` asks who owns repair after several
bounded authorities decline it.
- Independent reports, explicit no-finding outcomes, collateral protection,
and coordinator dispositions give defects a plausible route without turning
critics into canon writers.
## Findings
### IC-01
- **Priority:** Essential.
- **Location:** `baseline-packet.md`, Objective and scope and Candidate identity.
- **Evidence:** The KALSA-032 pass record is declared in scope, but
`workshop/deepening/passes/KALSA-032-review-council.md` has no recorded blob
hash. Its baseline hash is `82ceb03ec6cc95025833d80eccbafb24d8417cf4`;
every listed candidate hash otherwise matches.
- **Effect:** Critics can unknowingly review different versions of the process
model and authority map while the packet still claims an immutable shared
baseline.
- **Repair direction:** Add the pass record to the frozen identity, or remove
it from candidate scope and supply a frozen excerpt as context.
- **Collateral risk:** Do not create recursive hashing ceremony around the
packet or later reports.
- **Dependency:** Packet and provenance verification.
- **Confidence:** High; the declared scope and manifest disagree directly.
### IC-02
- **Priority:** Essential.
- **Location:** `workshop/review-council/README.md`, Packet Contract, Regression
Phase, and Completion Gate.
- **Evidence:** The diagnostic candidate must be frozen, but regression critics
merely receive “the repaired candidate.” No repaired snapshot identity or
mutation boundary is required; only Reader Experience receives a newly
assembled packet.
- **Effect:** Sequential regression reports may inspect different repairs. A
later change can invalidate earlier regressions while all seven reports still
satisfy the completion gate.
- **Repair direction:** Freeze one repaired candidate identity before
regression. Derive critic-specific packets, including the blind reader
packet, from that snapshot. Any subsequent repair should invalidate the
affected regressions and begin another bounded cycle.
- **Collateral risk:** Do not leak hidden material into the Reader Experience
packet or require every critic to receive irrelevant files.
- **Dependency:** Reader Experience packet construction and final verification.
- **Confidence:** High; the first snapshot is explicit and the second is absent.
### IC-03
- **Priority:** High-value.
- **Location:** `report-template.md`, Findings; `synthesis-template.md`, Finding
dispositions; `README.md`, Regression Phase.
- **Evidence:** Diagnostic findings receive no required identifier. The
synthesis invents an `ID`, while regression asks in prose which findings were
resolved.
- **Effect:** Split, merged, modified, and deferred findings can lose identity
between critic, coordinator, repair brief, and regression. An essential
defect can appear closed through ambiguous paraphrase.
- **Repair direction:** Require critic-local finding IDs at diagnosis; preserve
those IDs through disposition, repair instructions, regression status, and
queue deferral.
- **Collateral risk:** IDs must remain routing keys, not scores or canon authority.
- **Dependency:** Report and synthesis contracts.
- **Confidence:** High; the handoff fields do not share a stable key.
### IC-04
- **Priority:** High-value.
- **Location:** `workshop/review-council/README.md`, Authority Map and Repair
Phase; `workshop/deepening/pass-template.md` frontmatter and Review council.
- **Evidence:** The coordinator adjudicates every finding and alone integrates
canonical repairs, but the pass contract records no coordinator identity or
explicit assumption of that role.
- **Effect:** The decisive office exists, but its occupant and handoffs are
invisible. A critic or repair agent can silently become adjudicator,
weakening independence and obscuring who exercised integration authority.
- **Repair direction:** Record the coordinator in the pass and synthesis
artifacts. Permit one person or agent to hold multiple roles when necessary,
but require that overlap to be explicit.
- **Collateral risk:** Do not confuse role provenance with a demand for
managerial hierarchy or distinct human staffing.
- **Dependency:** Pass-record provenance.
- **Confidence:** Medium-high; authority is well described, but role occupancy
is not.
## Pattern or dependency map
`frozen baseline -> critic finding ID -> coordinator disposition -> bounded
repair -> frozen repaired snapshot -> regression by finding ID -> closeout`
The candidate defines the actors and most authority boundaries well. Its
missing mechanisms occur at identity-bearing handoffs: baseline scope, report
routing, coordinator occupancy, and repaired-state regression.
The Public corpus demonstrates that the Institutional Causality seat can detect
the intended defects. Ti'asantatca names workers, cohorts, observers, tenders,
and supporting households without yet naming command or succession; the sluice
dialogue exposes several competent authorities unable to own repair. Those are
later lore pressures, not findings KALSA-032 may settle.
## Tempting edit to resist
Do not empower this critic or the coordinator to invent a universal office that
resolves every failed handoff. `Four Readings of a Broken Sluice` gains force
from separated authority and unresolved responsibility. The council should
route that pressure to the owning lore pass, not answer it from workshop prose.
## Ranked actions
1. Freeze the complete diagnostic candidate and one repaired regression snapshot.
2. Add stable finding IDs and explicit coordinator identity across the report,
synthesis, pass, and regression contracts.
3. Leave the seven specialist remits and their canon boundary intact; no
omnibus institutional reviewer is needed.

View file

@ -0,0 +1,87 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
critic: Material Life and Ecology
phase: diagnosis
verdict: pass-with-findings
---
# Critic Report
## Concise diagnosis
Pass with findings. No blocking finding. The remit clearly distinguishes
material mechanism from a generic suffering ledger. One high-value hardening
would prevent inventories of material nouns from masquerading as causal chains.
## Protected successes
- Preserve the explicit rejection of automatic misery ledgers. Requiring a
mechanism, payer, place, response, and resulting choices directly addresses
the calibration concern.
- Preserve attention to maintenance, residues, bodies, care work, seasons, and
uneven access. These distinguish this remit from Institutional Causality.
- Preserve miraculous abundance, spectacle, and delight. Material scrutiny
should explain this magic's consequences, not force mundane scarcity back
into every wonder.
## Findings
### MLE-01
- **Priority:** High-value.
- **Location:** `critics/Material Life and Ecology.md`, first and second quoted
paragraphs.
- **Evidence:** The prompt names strong inputs and rejects abstract misery
accounting, but it does not say that naming workers, tools, inputs, and costs
remains an inventory unless their handoffs form a causal chain. The corpus
contains both noun-rich summaries and compact mechanisms such as incense
revealing airflow and chants coordinating distant controls.
- **Effect:** A checklist-shaped report could approve a material-cost ledger
because enough concrete nouns appear, or criticize only cadence while missing
absent physical handoffs.
- **Repair direction:** Require each substantive finding to trace one evidenced
chain—source/input -> labor or infrastructure -> allocation/use -> failure,
residue, or recovery—and identify the exact broken or unavailable link.
- **Collateral risk:** Do not require every overview to carry a supply-chain
monograph or imply all magic consumes mundane fuel.
- **Dependency:** Institutional Causality for jurisdiction; AI-Isms for cadence.
- **Confidence:** High.
### MLE-02
- **Priority:** Optional.
- **Location:** Material critic prompt as a whole, against a Public-only corpus.
- **Evidence:** Every lore file supplied for calibration is a presentation-only
Public note; hidden owner notes are intentionally absent.
- **Effect:** A critic could overstate “the setting has no mechanism” when the
packet establishes only that Public does not show it.
- **Repair direction:** Label a chain absent, outside the packet, delegated
through a named link, or appropriately compressed for the page's function.
- **Collateral risk:** Do not let “perhaps hidden elsewhere” excuse unsupported
public claims.
- **Dependency:** Packet design, provenance, and coordinator adjudication.
- **Confidence:** High.
## Pattern or dependency map
- `Relics and Ruins` supplies concrete operation and maintenance.
- `Peoples and Polities` mixes real labor organization with a praise-then-cost
ledger.
- `Magic and Miracles` and `Gods and the Dead` sometimes provide material
inventories without a full handoff.
- An explicit chain test would make this gradient reproducible.
## Tempting edit to resist
Do not turn the remit into an exhaustive food-water-fuel checklist for every
navigation or overview page. Preserve compression, spectacle, and concrete
examples; require full mechanisms from proper owners or bounded passes.
## Ranked actions
1. No essential action.
2. Require one traced material chain or exact broken link for each substantive
finding.
3. Optionally require an evidence-scope label so Public compression is not
misreported as absent canon.

View file

@ -0,0 +1,118 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-reader-01
critic: Reader Experience and Navigation
phase: diagnosis
verdict: pass-with-findings
---
# Critic Report
## Concise diagnosis
**Pass with findings. No blocking finding.** The spoiler consent gate is
unmistakable, the public path establishes a distinctive world quickly, and Low
Sere converts the setting into immediate playable pressure. The main weakness
is route hierarchy: the entrance foregrounds broad explanatory pages whose
repeated institutional cautions flatten momentum before stronger place and
source material.
The reader's working picture was a twilight world where settlements survive
through ancient infrastructure, negotiated miracle, inherited expertise, and
overlapping claims. Low Sere, Ti'asantatca's shield workers, the mobile Sarxe
center, and the divided City of Prophets created the strongest desire to follow.
## Protected successes
- `Kalsa/index.md` plainly distinguishes Kalsa for Readers from the Author and
GM Vault without an accidental spoiler lure.
- `Public/Places/Low Sere.md` provides compact spatial orientation, unequal
stakes, named actors, and an immediate adventure without resolving the dispute.
- `At the Basin Table.md` is the most memorable page. The drain key, name
cloth, pale marks, and unexplained knock carry lore through people.
- `Four Readings of a Broken Sluice.md` rewards rereading; the keeper keeps a
difficult scholarly dispute attached to missing seed and responsibility.
## Findings
### REN-01
- **Priority:** High-value, deferred lore pressure.
- **Location:** `Public/index.md`, Begin with the world as its people meet it.
- **Evidence:** Five broad survey pages precede one concrete place and two
situated sources; Low Sere appears sixth.
- **Effect:** A top-to-bottom reader encounters abstraction before meeting
anyone. Attention rises sharply at Low Sere and the Basin Table.
- **Repair direction:** Offer two routes: start with a live problem through Low
Sere and its source, or survey the world through overview notes.
- **Collateral risk:** Do not demote overview material or imply Low Sere
represents every culture.
- **Dependency:** Public navigation owner.
- **Confidence:** High.
### REN-02
- **Priority:** High-value, deferred lore pressure.
- **Location:** `Peoples and Polities.md`, especially Papsenai and
Ti'asantatca, The divided city, and Sarxe and Saxfoldi.
- **Evidence:** Compelling places and conflicts have no reader-visible
continuation for Ti'asantatca, the divided city, or Saxfoldi.
- **Effect:** Curiosity is raised and stranded; some links are painted doors.
- **Repair direction:** Link existing public material where it exists.
Otherwise label glimpses and route to the nearest situated material rather
than manufacture thin pages.
- **Collateral risk:** Do not expose hidden owners or demand link symmetry.
- **Dependency:** Public planning and future place/source owners.
- **Confidence:** High.
### REN-03
- **Priority:** High-value, deferred lore pressure.
- **Location:** recurring endings and framing across Public index, Peoples,
Magic, Gods, and Relics.
- **Evidence:** Variants of “does not grant authority,” “does not erase claims,”
“does not establish meaning,” and useful-yet-dangerous institutions recur in
a uniform explanatory voice.
- **Effect:** The Public surface can sound like one morally pre-digested
institutional handbook despite denying a universal schoolbook.
- **Repair direction:** Let named texts, offices, disputes, objects, and
consequences carry more of the ambiguity; remove narrator conclusions where
evidence already demonstrates them.
- **Collateral risk:** Do not replace careful limits with caricature or erase
protection against essentializing peoples.
- **Dependency:** AI-Isms and Culture remits.
- **Confidence:** Medium-high.
### REN-04
- **Priority:** Optional, deferred lore pressure.
- **Location:** `Public/World/Kalsa.md`, geography and Roads through twilight.
- **Evidence:** Sunward/coldward is clear, but later destinations are not placed
relative to one another beyond scattered cues.
- **Effect:** The climatic axis survives in memory; a travel-scale picture does not.
- **Repair direction:** Add a bounded traveller's route or itinerary rather
than an omniscient atlas.
- **Collateral risk:** Do not settle disputed geography or expose author structure.
- **Dependency:** Geography and Public presentation owners.
- **Confidence:** Medium.
## Pattern or dependency map
`safe root gate -> survey-first Public index -> strong premise plus repeated
abstract cautions -> Low Sere -> highest attention in situated sources`
The Public surface is strongest when an inhabitant, office, object, or disputed
record owns the explanation.
## Tempting edit to resist
- Do not add a glossary; unfamiliar terms accumulate meaning through use.
- Do not explain the Low Sere knock, decide what the dead are, or reveal the
Crown Cities' final nature.
- Do not build an omniscient atlas or one-page-per-people catalogue.
## Ranked actions
1. No essential action in KALSA-032.
2. Queue a bounded Public entry, onward-route, and narrator-voice repair.
3. Consider one inhabitant-owned travel orientation only with owner evidence.

View file

@ -0,0 +1,54 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-repaired-02
supersedes: KALSA-032-repaired-01
candidate_identity: final-implementation-blob-hashes-recorded-below
coordinator: Codex /root
---
# KALSA-032 Final Repaired Packet
`KALSA-032-repaired-01` passed all seven regressions, but its candidate manifest
included queue and memory surfaces before truthful closeout. This final snapshot
changes no council doctrine or prompt. It records the KALSA-032 queue state as
complete, advances Body and memory snapshot timestamps, and includes the synced
CultCache memory projection. All counselors must verify this identity before
the pass is committed.
```text
260b2e6cbba921625877228d92258dac7796b879 AGENTS.md
820547810e5a7d8990e35f151bd8f370f8dcd5ef .epiphany/project-memory.md
0077c96833043f20736343b59ee215ee672b4b6d .epiphany/state/project-memory.cc
269904dde2d51d376fbfe6ba31276769f7ae8b7c workshop/deepening/README.md
b619f5ff1e9c75bfaa101aea44bde3e4a7ed0e8e workshop/deepening/pass-template.md
1ee05a2ae8ebd797e30b1bc8480197cdeb50b74d workshop/deepening/benchmark.md
d498d0338ece56012b26ecf3c64bd180be2e5848 workshop/deepening/body-map.md
403c06189a9acdf92a8c540b7706519e452a01d7 workshop/deepening/queue.md
865d06577182164398856d36a1d0a382e0fa3b8a workshop/review-council/README.md
5ed604ff74fa93aa6bfbab1a2c73c7fd91e9ab50 workshop/review-council/shared-prompt.md
81db83f5254ba2e90f1f77e85ea8bc5ebaa74c98 workshop/review-council/packet-template.md
e92ee2dd2f34e415868e0d14263dab9e51d9d4fb workshop/review-council/report-template.md
e2f6d0cfe6efbb21d750322bbebc6e5dc0f2ca22 workshop/review-council/synthesis-template.md
e142276b13062be7792310c31b8e4a32dab8a05e workshop/review-council/regression-template.md
d96d41f25259ac367395db8e4787004d71f8768f workshop/review-council/critics/Institutional Causality.md
7ba90172261aae5c815a40f0de3af7b5853d62c4 workshop/review-council/critics/Culture and Situated Knowledge.md
10e6079850ce8a3420cf4898ff243e14504da6d3 workshop/review-council/critics/Material Life and Ecology.md
d33b34e1e13f7147893c02482589db6dd99f67a3 workshop/review-council/critics/History Conflict and Change.md
9ec7dd60f6231b5f202ff9e4bf0c4a02c731dfbb workshop/review-council/critics/Ghostlight Play and Consequence.md
043e9313cf3de9b55d1a78d4cfd5bee26d147d0d workshop/review-council/critics/AI-Isms and Prose.md
eeb0b26d2a53407f21766598cd337427dfe54346 workshop/review-council/critics/Reader Experience and Navigation.md
```
The pass, synthesis, ledger, packets, and reports are audit evidence and remain
outside the implementation identity so their closeout links can be appended
without pretending to mutate the reviewed machine.
## Bounded revalidation
- Verify all twenty-one hashes.
- Confirm the critic's prompt and accepted repairs are unchanged from
`KALSA-032-repaired-01`.
- Confirm queue closeout and memory projection introduce no defect within the
critic's remit.
- Return a brief verdict for `KALSA-032-repaired-02` and name what remains
deferred or must be left alone.

View file

@ -0,0 +1,18 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-reader-repaired-02
process_identity: KALSA-032-repaired-02
audience: curious Ghostlight player exploring a high-fantasy setting vault
public_blind_packet: true
---
# KALSA-032 Final Fresh Reader Packet
Begin at `Kalsa/index.md`, then follow only links available through
`Kalsa/Public/`. Do not open `Kalsa/Spoilers/`, `seed/`, `workshop/`,
`.epiphany/`, generated output, or any critic report.
The reader source remains unchanged. Confirm that queue and memory closeout did
not leak into the player surface, that navigation remains Public-only, and that
previously observed reader pressure remains a bounded later repair rather than
an essential KALSA-032 defect. Do not edit lore.

View file

@ -0,0 +1,49 @@
---
pass_id: KALSA-032
repaired_packet_id: KALSA-032-repaired-02
status: passed
---
# KALSA-032 Final Revalidation
KALSA-032-repaired-01 passed all seven regression turns. Closeout then changed
only queue status, Body and memory snapshot time, and the generated CultCache
projection. Because those surfaces appeared in the first manifest, the
coordinator froze `KALSA-032-repaired-02` and revalidated the final body instead
of letting a stale hash wear a completion label.
## Results
| Critic | Identity checked | Verdict | Essential defect |
| --- | --- | --- | --- |
| Institutional Causality | all 21 blobs | Pass | none |
| Culture and Situated Knowledge | all 21 blobs | Pass | none |
| Material Life and Ecology | all 21 blobs | Pass | none |
| History, Conflict, and Change | all 21 blobs | Pass | none |
| Ghostlight Play and Consequence | all 21 blobs | Pass | none |
| AI-Isms and Prose | all 21 blobs | Pass | none |
| Reader Experience and Navigation | blind Public packet | Pass | none |
The six grounded critics confirmed that their prompts and accepted repairs were
unchanged, the queue closed truthfully, memory and its CultCache projection
remained non-canonical, and the Body map preserved subject-note authority. The
fresh reader found no spoiler, queue, memory, council, or closeout leak; every
followed reader link remained in `Kalsa/Public/`.
## Leave alone
- the seven-seat boundary and same-snapshot rule;
- stable finding IDs and explicit coordinator occupancy;
- informal knowledge reproduction and protected incomplete models;
- concise material-chain and Ghostlight play traces;
- target-supported interventions rather than universal hook menus;
- AI-ism term-origin analysis, false-positive protection, and the ban on
detector or token-list verdicts;
- the consent gate, difficult names, Low Sere route, situated source dialogues,
and unresolved testimony.
## Deferred pressure
KALSA-024 and KALSA-031 retain bounded cultural and historical work. KALSA-033
owns Public entry hierarchy, onward routes, situated orientation, and recurring
narrator-level moral/checksum cadence. None is an essential KALSA-032 defect.

View file

@ -0,0 +1,18 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-reader-01
candidate_identity: current-public-source-at-baseline-01
audience: curious Ghostlight player exploring a high-fantasy setting vault
public_blind_packet: true
---
# KALSA-032 Blind Reader Packet
Begin at `Kalsa/index.md`, then follow only links available through
`Kalsa/Public/`. Read the current Public notes as a casual explorer. Do not open
`Kalsa/Spoilers/`, `seed/`, `workshop/`, `.epiphany/`, generated Quartz output,
or other council reports.
Evaluate the entry path and reader-visible world using the Reader Experience
and Navigation prompt. This is calibration evidence for a review process. Do
not edit lore.

View file

@ -0,0 +1,43 @@
---
pass_id: KALSA-032
baseline_packet_id: KALSA-032-baseline-01
repaired_packet_id: KALSA-032-repaired-01
critic: AI-Isms and Prose
phase: regression
verdict: pass
---
# Critic Regression Report
## Finding status
| Finding ID | Disposition | Status | Evidence |
| --- | --- | --- | --- |
| AIP-01 | Accept | Resolved | The critic requires term-origin analysis and diagnoses diffusion rather than tokens, protecting earned Ju'onai vocabulary. |
| AIP-02 | Accept | Resolved | The prompt routes recurrence/contrast to Evidence and retained function to Collateral risk without enlarging the shared template. |
All twenty repaired candidate hashes matched `KALSA-032-repaired-01`.
## New defects within remit
- None.
## Protected strengths
- Cluster, recurrence, context, and reading effect remain the basis of diagnosis.
- Detector claims, word bans, automatic rewriting, and regex gates remain excluded.
- Functional inventories, deliberate antithesis, culture-specific rhetoric, and
earned technical terms remain protected.
## Essential further change
- None.
## Leave alone
- Leave the Kalsa-specific pattern inventory and shared report fields intact.
- Add no token lists, detector scores, or further vocabulary warnings.
## Verdict
**Pass.** AIP-01 and AIP-02 are resolved with no new prose-review defect.

View file

@ -0,0 +1,45 @@
---
pass_id: KALSA-032
baseline_packet_id: KALSA-032-baseline-01
repaired_packet_id: KALSA-032-repaired-01
critic: Culture and Situated Knowledge
phase: regression
verdict: pass
---
# Critic Regression Report
## Finding status
| Finding ID | Disposition | Status | Evidence |
| --- | --- | --- | --- |
| CSK-01 | Accept | Resolved | The critic now requires the actual household, oral, customary, itinerant, ritual, testimonial, or formal reproduction path. |
| CSK-02 | Accept | Resolved | Packet and pass templates preserve owner/synthesis, evidence/reproduction, useful domain, blind spot, and protected unknown. |
| CSK-03 | Defer | Resolved as deferred routing | KALSA-031 now tests one supported non-magical relation or practice; KALSA-024 includes mixed or cross-boundary lives under actual play pressure. |
All relevant candidate hashes matched `KALSA-032-repaired-01`.
## New defects within remit
- None. The repairs impose neither symmetrical coverage nor formal institutions
on diffuse knowledge.
## Protected strengths
- Culturally owned disagreement, incomplete useful models, opacity, protected
unknowns, uneven emphasis, and source-grounded deferral survived.
## Essential further change
- None.
## Leave alone
- Leave the reproduction examples as possibilities rather than a checklist.
- Leave KALSA-031 and KALSA-024 bounded.
- Leave Public lore untouched in KALSA-032.
## Verdict
**Pass.** Accepted findings are resolved and deferred pressure is routed without
invention.

View file

@ -0,0 +1,45 @@
---
pass_id: KALSA-032
baseline_packet_id: KALSA-032-baseline-01
repaired_packet_id: KALSA-032-repaired-01
critic: Ghostlight Play and Consequence
phase: regression
verdict: pass
---
# Critic Regression Report
## Finding status
| Finding ID | Disposition | Status | Evidence |
| --- | --- | --- | --- |
| GPC-01 | Accept | Resolved | Packet and pass templates carry point of contact, character-accessible evidence, supported intervention, and consequence horizon/play-bearing owner. |
| GPC-02 | Accept | Resolved | Benchmark and critic use target-supported courses, differing responses, and durable owners/receipts rather than a universal action menu. |
| GPC-03 | Accept | Resolved | The prompt traces a play path when evidence permits and allows overviews to route to a play-bearing owner. |
All twenty repaired candidate hashes matched `KALSA-032-repaired-01`.
## New defects within remit
- None.
## Protected strengths
- Character action remains bounded by available evidence.
- Uncertainty, asymmetric knowledge, spectacle, unsolved problems, and
inhabitant agency remain protected.
- Overview pages need not become encounters.
## Essential further change
- None.
## Leave alone
- Keep the Ghostlight lens compact.
- Keep the evidence-permitting trace and overview exception.
- Add no hooks quota, mandatory clocks, or intervention matrix.
## Verdict
**Pass.** GPC-01 through GPC-03 are resolved without a play-review regression.

View file

@ -0,0 +1,44 @@
---
pass_id: KALSA-032
baseline_packet_id: KALSA-032-baseline-01
repaired_packet_id: KALSA-032-repaired-01
critic: History, Conflict, and Change
phase: regression
verdict: pass
---
# Critic Regression Report
## Finding status
| Finding ID | Disposition | Status | Evidence |
| --- | --- | --- | --- |
| HCC-01 | Accept | Resolved | Packet and pass templates conditionally require factual event owners plus neighboring memory/consequence notes, or an explicit inability to judge. |
| HCC-02 | Defer | Resolved as bounded deferral | KALSA-031 routes source-grounded formation without invented events; KALSA-033 requires owner inspection before Public changes. |
All twenty repaired candidate hashes matched `KALSA-032-repaired-01`.
## New defects within remit
- None.
## Protected strengths
- Myths, disputed chronologies, selective archives, and unresolved
contradictions remain protected.
- Event evidence remains conditional and does not create one public history.
- Public lore stayed unchanged.
## Essential further change
- None.
## Leave alone
- Leave the remit and conditional event-owner field concise.
- Do not invent origin events or expand KALSA-032 into lore repair.
## Verdict
**Pass.** HCC-01 is resolved, HCC-02 is coherently deferred, and no historical
ownership regression appeared.

View file

@ -0,0 +1,47 @@
---
pass_id: KALSA-032
baseline_packet_id: KALSA-032-baseline-01
repaired_packet_id: KALSA-032-repaired-01
critic: Institutional Causality
phase: regression
verdict: pass
---
# Critic Regression Report
## Finding status
| Finding ID | Disposition | Status | Evidence |
| --- | --- | --- | --- |
| IC-01 | Accept | Resolved | The baseline packet now records the omitted pass-record hash captured in the original report. |
| IC-02 | Accept | Resolved | The council README requires one frozen repaired identity, derived packets, same-identity regressions, and invalidation after mutation. All twenty declared blobs matched. |
| IC-03 | Accept | Resolved | Report, README, synthesis, and regression contracts preserve stable critic-local IDs. |
| IC-04 | Accept | Resolved | `Codex /root` is recorded in packet, synthesis, and pass; reusable templates require explicit occupancy and overlap. |
## New defects within remit
- None. Identity and handoff mechanisms were added without granting critics,
reports, or majorities canon authority.
## Protected strengths
- Diagnosis, adjudication, repair, and canonical ownership remain separate.
- Failed handoffs and enforcement limits remain reviewable without demanding
centralized institutions.
- Reader blindness, bounded packets, no-finding reports, and the prohibition on
automatic critic edits survived.
## Essential further change
- None.
## Leave alone
- Leave the seven remits, coordinator adjudication, subject-note authority, and
permission for informal or plural institutions unchanged.
- Add no universal office, aggregate score, or automatic enforcement layer.
- Do not reopen lore in this process pass.
## Verdict
**Pass.** IC-01 through IC-04 are resolved on the verified repaired snapshot.

View file

@ -0,0 +1,42 @@
---
pass_id: KALSA-032
baseline_packet_id: KALSA-032-baseline-01
repaired_packet_id: KALSA-032-repaired-01
critic: Material Life and Ecology
phase: regression
verdict: pass
---
# Critic Regression Report
## Finding status
| Finding ID | Disposition | Status | Evidence |
| --- | --- | --- | --- |
| MLE-01 | Accept | Resolved | The critic now traces source/input through labor/infrastructure and allocation/use into failure, residue, or recovery, naming the broken handoff. |
| MLE-02 | Accept | Resolved | The critic distinguishes absent canon, outside-packet evidence, named delegation, and appropriate compression. |
All twenty candidate hashes matched `KALSA-032-repaired-01`.
## New defects within remit
- None. The chain and evidence-scope requirements remain concise.
## Protected strengths
- The automatic misery-ledger test survived.
- Maintenance, residues, bodies, care work, seasons, uneven access, abundance,
spectacle, and delight remain protected.
## Essential further change
- None.
## Leave alone
- Leave the repaired prompt alone; further mandatory fields would add ceremony.
- Leave Public calibration lore untouched in this pass.
## Verdict
**Pass.** MLE-01 and MLE-02 are resolved on the verified snapshot.

View file

@ -0,0 +1,104 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-reader-repaired-01
critic: Reader Experience and Navigation
phase: regression
verdict: pass-with-findings
---
# Fresh Reader Report
## Concise diagnosis
**Pass with findings.** The consent gate is clear, every followed Public link
remained inside `Kalsa/Public/`, and no review language or hidden mechanism
appeared. The strongest route remains Public index -> Low Sere -> At the Basin
Table. The deferred pressure is real: wider-world curiosity often lacks an
onward route, and repeated cautions about authority and incomplete knowledge
become instructional across overview pages.
## Protected successes
- The root spoiler boundary is explicit and the Public route is preferred.
- The Public entrance establishes the unmoving sun, stormshield labor,
unreliable gods, relic infrastructure, and dangerous dead without resolving
their causes.
- Low Sere turns those concerns into forty-one hearths, wet rooms, a stolen
tool, provisional authority, dead workers, and one black door.
- Both situated sources preserve memorable objects, disagreement, consequence,
and mystery without glossary exposition.
## Findings
### REN-01
- **Priority:** High-value, deferred.
- **Location:** Public index navigation and Peoples and Polities onward links.
- **Evidence:** Low Sere and the Ju'onai dialogue have situated follow-through;
Ti'asantatca, Jamnai, the divided city, Selza'a country, and Saxfoldi do not.
- **Effect:** Curiosity repeatedly rises and has nowhere to travel.
- **Repair direction:** Mark Low Sere as the available first playable thread;
add one contrasting situated branch only when an owner can support it.
- **Collateral risk:** Do not manufacture equal coverage or expose material
before it has an inhabitant owner.
- **Confidence:** High.
### REN-02
- **Priority:** High-value, deferred.
- **Location:** repeated qualifications across Public overview notes.
- **Evidence:** Several pages restate that explanations do not own truth,
discovery does not erase claims, evidence does not establish meaning,
competence does not grant authority, and classifications do not determine
people.
- **Effect:** Later instances become skimmable and expose project doctrine as a
pre-digested lesson.
- **Repair direction:** Keep the principle but let concrete offices, disputes,
failed decisions, and sources carry it.
- **Collateral risk:** Do not erase Kalsa's distinction between power,
knowledge, and rightful authority.
- **Confidence:** High.
### REN-03
- **Priority:** Optional, deferred.
- **Location:** Ju'onai section in Peoples and Polities.
- **Evidence:** Dense terminology appears before the dialogue that makes it
dramatic; the link is only at the footer.
- **Effect:** Readers may skim before reaching the best situated explanation.
- **Repair direction:** Add one early, relevant link to the dialogue.
- **Collateral risk:** Do not turn prose into an inline reference interface.
- **Confidence:** High.
### REN-04
- **Priority:** Optional, deferred.
- **Location:** Public YAML properties.
- **Evidence:** Governance metadata may be visible in some raw-vault property
settings, though no process text appears in prose.
- **Effect:** Possible immersion break, not a spoiler leak.
- **Repair direction:** Prefer presentation-layer suppression over removing
source metadata.
- **Collateral risk:** Do not weaken ownership and publication checks.
- **Confidence:** Medium.
## Pattern or dependency map
- Safe root and Public-only links held.
- Overview pages form an abstract loop.
- Low Sere and the two sources create the strongest situated branches.
- Attention falls during repeated qualification and rises with objects, voices,
and immediate consequence.
## Tempting edit to resist
Do not add a glossary, universal magic explanation, complete polity matrix, or
equal page per people. Do not resolve the Crown Cities, divine perception,
returning dead, coldward judgment, or the Low Sere knock.
## Ranked actions
1. No essential action; consent and public-only navigation held.
2. Preserve the deferred KALSA-033 route and narrator-voice pressure.
3. Consider one earlier Ju'onai source link and presentation-layer property
suppression during that bounded pass.

View file

@ -0,0 +1,81 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-repaired-01
baseline_packet_id: KALSA-032-baseline-01
candidate_identity: repaired-implementation-blob-hashes-recorded-below
coordinator: Codex /root
audience: Kalsa lore-review council
public_blind_packet: false
---
# KALSA-032 Repaired Council Packet
## Repair scope
This snapshot contains the process implementation after coordinator
adjudication in `synthesis.md`. Critics receive this packet, their own raw
diagnostic report, the synthesis, and only the candidate files needed for their
remit. They do not receive other raw reports.
KALSA-032 changed no canonical or Public lore. Reader-visible findings were
deferred to KALSA-033. The repaired implementation must therefore be judged on
whether it routes future lore work coherently, preserves authority, and resolves
the accepted process defects.
## Repaired identity
```text
260b2e6cbba921625877228d92258dac7796b879 AGENTS.md
7f0478cc0da4693ade768cd3bba769e5b09f5b5b .epiphany/project-memory.md
269904dde2d51d376fbfe6ba31276769f7ae8b7c workshop/deepening/README.md
b619f5ff1e9c75bfaa101aea44bde3e4a7ed0e8e workshop/deepening/pass-template.md
1ee05a2ae8ebd797e30b1bc8480197cdeb50b74d workshop/deepening/benchmark.md
2dcd7223588314fbd09049afd6aaf8ebdf9e1f58 workshop/deepening/body-map.md
88bca7eee3afcbe19a07edb4d956e0f368809788 workshop/deepening/queue.md
865d06577182164398856d36a1d0a382e0fa3b8a workshop/review-council/README.md
5ed604ff74fa93aa6bfbab1a2c73c7fd91e9ab50 workshop/review-council/shared-prompt.md
81db83f5254ba2e90f1f77e85ea8bc5ebaa74c98 workshop/review-council/packet-template.md
e92ee2dd2f34e415868e0d14263dab9e51d9d4fb workshop/review-council/report-template.md
e2f6d0cfe6efbb21d750322bbebc6e5dc0f2ca22 workshop/review-council/synthesis-template.md
e142276b13062be7792310c31b8e4a32dab8a05e workshop/review-council/regression-template.md
d96d41f25259ac367395db8e4787004d71f8768f workshop/review-council/critics/Institutional Causality.md
7ba90172261aae5c815a40f0de3af7b5853d62c4 workshop/review-council/critics/Culture and Situated Knowledge.md
10e6079850ce8a3420cf4898ff243e14504da6d3 workshop/review-council/critics/Material Life and Ecology.md
d33b34e1e13f7147893c02482589db6dd99f67a3 workshop/review-council/critics/History Conflict and Change.md
9ec7dd60f6231b5f202ff9e4bf0c4a02c731dfbb workshop/review-council/critics/Ghostlight Play and Consequence.md
043e9313cf3de9b55d1a78d4cfd5bee26d147d0d workshop/review-council/critics/AI-Isms and Prose.md
eeb0b26d2a53407f21766598cd337427dfe54346 workshop/review-council/critics/Reader Experience and Navigation.md
```
The KALSA-032 pass record and synthesis are evolving audit artifacts. Their
coordinator identity, report inventory, dispositions, protected ledger, and
repair brief were complete before this freeze. Appending regression links and
closeout does not mutate the reviewed implementation.
## Accepted repairs represented
- complete baseline scope and one immutable repaired snapshot;
- stable finding IDs and explicit coordinator identity;
- actual formal or informal knowledge-reproduction paths;
- protected boundaries for in-scope inhabitant models;
- traced material chains and evidence-scope labels;
- event-owner evidence or an explicit inability to judge it;
- bounded Ghostlight access, target-earned interventions, and a play trace;
- term-origin analysis for workshop-dialect findings;
- regression template and same-snapshot completion gate;
- bounded deferral of Public entry, link, historical, and narrator-voice
pressure to KALSA-031, KALSA-024, and KALSA-033.
## Protected strengths
Use the protected-material ledger in `synthesis.md`. No regression may demand a
continuity critic, novel-specific machinery, detector or regex gate, lore edits
inside this pass, automatic critic edits, majority rule, or symmetric content
coverage.
## Regression rule
Return `regression-template.md`. Address the IDs from your own report. Report a
new defect only when the repaired implementation introduced or retained one in
your remit. Preferences do not reopen the repair. Name what should now be left
alone.

View file

@ -0,0 +1,19 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-reader-repaired-01
baseline_packet_id: KALSA-032-reader-01
candidate_identity: unchanged-public-source-plus-kalsa-032-repaired-01-process
audience: curious Ghostlight player exploring a high-fantasy setting vault
public_blind_packet: true
---
# KALSA-032 Fresh Reader Regression Packet
Begin at `Kalsa/index.md`, then follow only links available through
`Kalsa/Public/`. Do not open `Kalsa/Spoilers/`, `seed/`, `workshop/`,
`.epiphany/`, generated output, or any critic report.
The reader-visible source is intentionally unchanged in this process pass. Read
it fresh and return the Reader Experience and Navigation report contract. The
purpose is to confirm that the deferred pressure remains accurately bounded and
that no process change leaked into the player surface. Do not edit lore.

View file

@ -0,0 +1,101 @@
---
pass_id: KALSA-032
packet_id: KALSA-032-baseline-01
status: adjudicated
coordinator: Codex /root
---
# KALSA-032 Council Synthesis and Adjudication
## Report inventory
| Critic | Diagnostic report | Verdict | Regression report |
| --- | --- | --- | --- |
| Institutional Causality | [[diagnosis/Institutional Causality]] | Block | [[regression/Institutional Causality]] — Pass |
| Culture and Situated Knowledge | [[diagnosis/Culture and Situated Knowledge]] | Pass with findings | [[regression/Culture and Situated Knowledge]] — Pass |
| Material Life and Ecology | [[diagnosis/Material Life and Ecology]] | Pass with findings | [[regression/Material Life and Ecology]] — Pass |
| History, Conflict, and Change | [[diagnosis/History Conflict and Change]] | Pass with findings | [[regression/History Conflict and Change]] — Pass |
| Ghostlight Play and Consequence | [[diagnosis/Ghostlight Play and Consequence]] | Pass with findings | [[regression/Ghostlight Play and Consequence]] — Pass |
| AI-Isms and Prose | [[diagnosis/AI-Isms and Prose]] | Pass with findings | [[regression/AI-Isms and Prose]] — Pass |
| Reader Experience and Navigation | [[diagnosis/Reader Experience and Navigation]] | Pass with findings | [[regression/Reader Experience and Navigation]] — Pass with deferred findings |
All seven reports were produced independently. No critic read another report
before coordinator adjudication. Reader Experience received
`KALSA-032-reader-01`, not the grounded baseline packet.
## Protected-material ledger
- Critics diagnose; coordinator adjudicates; subject notes own canon.
- Every seat receives an independent diagnostic and regression turn. An honest
no-finding result is valid.
- The council has seven setting-facing remits rather than manuscript-specific
continuity, character, chapter, scene, or pacing roles.
- Reader review remains blind when Public material is in scope.
- AI-ism review diagnoses recurrent construction and reading effect without
detector claims, banned vocabulary, or automatic rewriting.
- Ju'onai technical language is culturally owned and must not be stripped merely
because authoring language uses similar words.
- Miraculous abundance, mystery, functional inventories, uneven cultural
emphasis, and target-specific forms of play remain protectable.
- KALSA-032 changes process only. Current Public lore findings cannot authorize
opportunistic lore edits.
## Finding dispositions
| ID | Critic | Disposition | Evidence and reason | Exact repair instruction |
| --- | --- | --- | --- | --- |
| IC-01 | Institutional Causality | Accept | The baseline declared the pass record in scope but omitted its hash. | Add the baseline pass-record hash without recursively hashing the packet. |
| IC-02 | Institutional Causality | Accept | Seven regressions could otherwise inspect seven different repaired bodies. | Freeze one repaired snapshot; derive all packets from it; invalidate affected regressions after later mutation. |
| IC-03 | Institutional Causality | Accept | Finding identity was lost between report, synthesis, repair, regression, and deferral. | Require stable critic-local IDs throughout the path. |
| IC-04 | Institutional Causality | Accept | The coordinator office had authority but no recorded occupant. | Record coordinator identity in packet, pass, and synthesis; disclose role overlap. |
| CSK-01 | Culture and Situated Knowledge | Accept | Public knowledge may reproduce through households, travel, custom, or oral transmission rather than a formal school. | Replace institution-biased reproduction language with the actual applicable transmission path. |
| CSK-02 | Culture and Situated Knowledge | Accept | The packet and pass could lose a model's protected blind spot during repair handoff. | Record owner/synthesis, evidence/reproduction, useful domain, blind spot, and protected unknown when an inhabitant model is in scope. |
| CSK-03 | Culture and Situated Knowledge | Defer | Current queue pressure begins from magical traditions and may neglect life beyond magic, but KALSA-032 cannot write culture. | Amend KALSA-031 and KALSA-024 to test supported non-magical and cross-boundary life without symmetry quotas. |
| MLE-01 | Material Life and Ecology | Accept | Concrete nouns can still form a decorative cost inventory without a material handoff. | Require one evidenced source/input-to-use-to-failure/residue/recovery chain for substantive findings. |
| MLE-02 | Material Life and Ecology | Accept | Public compression could be mistaken for absent canon. | Require an evidence-scope label: absent, outside packet, delegated, or appropriately compressed. |
| HCC-01 | History, Conflict, and Change | Accept | Event ownership cannot be judged without the owner and neighboring memory notes. | Supply event-owner evidence when in scope or state that the packet cannot support the verdict. |
| HCC-02 | History, Conflict, and Change | Defer | Several Public conflicts lack visible historical formation, but hidden owners were not supplied. | Route source-grounded formation pressure into KALSA-031/KALSA-033; invent no event. |
| GPC-01 | Ghostlight Play and Consequence | Accept | The packet lacked a bounded character-access lens. | Add point of contact, accessible evidence, supported intervention, and consequence horizon/play-bearing owner. |
| GPC-02 | Ghostlight Play and Consequence | Accept | Repeated help/harm/capture/reform verbs could impose the same quest menu on every target. | Replace the universal verbs with two target-supported courses, different response, and durable owner/receipt. |
| GPC-03 | Ghostlight Play and Consequence | Accept | Ingredient lists can avoid showing where interaction becomes inert. | Ask for one evidence-supported play trace when the packet permits it. |
| AIP-01 | AI-Isms and Prose | Accept | The same abstract vocabulary can be narrator leakage or earned Ju'onai register. | Require term-origin analysis and diagnose diffusion across unrelated voices, not tokens. |
| AIP-02 | AI-Isms and Prose | Accept | Recurrence and retained function were not clearly routed into shared fields. | Put recurrence/contrast in Evidence and retained function in Collateral risk. |
| REN-01 | Reader Experience and Navigation | Defer | The safe entrance is survey-first and delays the strongest concrete route. | KALSA-033 tests separate live-problem and world-survey paths. |
| REN-02 | Reader Experience and Navigation | Defer | Ti'asantatca, the divided city, and Saxfoldi raise curiosity without reachable continuation. | KALSA-033 links existing Public material or labels/reroutes glimpses; it creates no symmetry stubs. |
| REN-03 | Reader Experience and Navigation | Defer | Repeated narrator-level caution creates one pre-digested institutional voice. | KALSA-033 repairs accepted prose clusters through situated evidence while protecting Ju'onai's register. |
| REN-04 | Reader Experience and Navigation | Defer | Travel-scale orientation is weak but may be intentionally unsettled. | KALSA-033 may test one inhabitant-owned itinerary only after owner evidence. |
## Repair brief
- **Owners allowed to change:** `workshop/review-council/`, deepening process
docs, KALSA-032 evidence, `AGENTS.md`, Body map, queue, and project memory.
- **Authority deleted or demoted first:** an unfrozen repaired candidate,
paraphrase-only finding tracking, silent coordinator occupancy, formal-school
bias, material noun inventories, and universal player-action menus.
- **Accepted changes:** implement every accepted instruction above.
- **Explicit non-changes:** no Kalsa lore or Public presentation changes; no
continuity critic; no automated detector, regex style gate, aggregate score,
majority vote, or automatic critic edit.
- **Deferred queue pressure:** KALSA-031 and KALSA-024 receive bounded cultural
and historical constraints; KALSA-033 owns Public route and narrator-voice
repair.
- **Verification:** freeze one repaired packet, run all seven regressions, then
execute the repository verification suite.
## Regression closeout
- **Repaired packet identity:** `KALSA-032-repaired-02`; after the first
snapshot passed, truthful queue and memory closeout required a final freeze.
All six grounded critics independently verified all twenty-one recorded blob
hashes. Reader Experience used `KALSA-032-reader-repaired-02` against
unchanged Public source. See [[final-revalidation]].
- **Essential findings resolved:** IC-01 and IC-02. Every accepted process
finding passed its original seat's regression.
- **Essential findings remaining:** none.
- **Regressions accepted for repair:** none. The fresh reader confirmed the
existing deferred Public pressure and found no spoiler or process-language leak.
- **What must now be left alone:** the seven-seat boundary, compact packet and
report forms, contextual AI-ism pattern inventory, culturally owned Ju'onai
register, mystery, non-symmetric coverage, and all Kalsa lore in this pass.
- **Final council decision:** Pass. KALSA-032 may close after repository and
memory verification succeeds.

View file

@ -0,0 +1,157 @@
# Kalsa Lore Review Council
This directory owns the reusable critic prompts and council protocol for Kalsa
deepening passes. It does not own lore. A critic diagnoses one quality surface;
the pass coordinator adjudicates findings; canonical subject notes retain every
setting claim.
The protocol adapts the independent-report and regression shape of
AetheriaLore's Novella Council to a setting vault. Manuscript-only specialties
are omitted. Kalsa instead reviews whether its lore can generate institutions,
cultures, material pressures, history, adventures, and an inviting
inhabitant-facing reading experience without collapsing into authorial jargon
or polished machine prose.
## Council Seats
Every deepening pass gives each seat one diagnostic turn and one post-repair
regression turn:
1. [[critics/Institutional Causality|Institutional Causality]]
2. [[critics/Culture and Situated Knowledge|Culture and Situated Knowledge]]
3. [[critics/Material Life and Ecology|Material Life and Ecology]]
4. [[critics/History Conflict and Change|History, Conflict, and Change]]
5. [[critics/Ghostlight Play and Consequence|Ghostlight Play and Consequence]]
6. [[critics/AI-Isms and Prose|AI-Isms and Prose]]
7. [[critics/Reader Experience and Navigation|Reader Experience and Navigation]]
A seat may return `No blocking finding`. Silence is not a report, and a critic
must not manufacture a defect to justify its seat. Replace or retire a seat if
several passes show that its remit is redundant; do not keep a pageant of agent
names.
## Authority Map
- **Owner:** this file owns council membership, sequence, packet boundaries,
and completion gates. Individual prompt files own only their diagnostic
remit.
- **Inputs:** an immutable candidate packet assembled from the pass, exact
source and owner notes, intended audience, protected strengths, known
concerns, and the candidate diff or baseline.
- **Outputs:** seven raw diagnostic reports, coordinator dispositions, a finite
repair brief, seven regression reports, and a final council decision.
- **Derived state:** confidence labels, pattern inventories, counts, agreement,
and report summaries are evidence. They do not decide lore or publication.
- **Forbidden writers:** critics, majorities, detector scores, checklists, and
report prose may not edit canon, promote a candidate, or invent a missing
answer. Workshop files may not become a second lore database.
- **Shared path:** baseline packet -> independent diagnostic turns ->
coordinator adjudication -> bounded owner repair -> independent regression
turns -> coordinator closeout -> ordinary repository verification.
- **Cut line:** a generic `Soul review`, one omnibus opinion, or an aggregate
score cannot stand in for the seven named reports. Continuity is not a
dedicated seat; source fidelity, provenance, ownership, and spoiler safety
constrain every seat and remain part of final verification.
## Packet Contract
Create the packet before critics begin, using [[packet-template]]. Freeze its
identity with a commit, tag, or recorded file hashes. Diagnostic critics do not
see one another's reports.
Record the coordinator in the packet, pass, and synthesis. One person or agent
may hold more than one role when necessary, but the overlap must be explicit;
no critic or repair writer silently inherits adjudication authority.
Six grounded critics receive only the canonical and workshop sources necessary
for their remit. Reader Experience and Navigation receives the reader-visible
entry path, target-audience description, and any navigation surface under test.
It does not receive hidden canon, authorial intention, known concerns, or other
reports. When a pass changes only author material, give that critic the
plausible direct-entry path an author or GM would actually use and say that the
packet is not spoiler-blind.
The packet must distinguish:
- immutable seed evidence;
- adopted canon and its owner;
- provisional design, rumor, contested belief, and unknowns;
- Public inhabitant presentation;
- protected peculiarities and explicit operator decisions;
- the exact scope critics may inspect.
## Diagnostic Phase
Prepend [[shared-prompt]] to the appropriate critic prompt. Each critic returns
[[report-template]] as a separate artifact. Reports may cite the same defect,
but they must arrive independently.
The coordinator then records a disposition for every substantive finding:
- `Accept` — the evidence identifies a defect in scope;
- `Modify` — the defect is real but the proposed direction crosses an owner or
threatens protected material;
- `Reject` — the finding is unsupported, outside remit, or would make Kalsa
more generic;
- `Defer` — the pressure is real but belongs to a named later pass or needs
operator input.
Agreement increases attention, not authority. A lone well-supported finding
can block a pass. Seven correlated preferences cannot vote a seed fact out of
existence.
Every diagnostic finding receives a stable critic-local ID. Preserve that ID
through disposition, repair instruction, regression, and queue deferral. The ID
is a routing key, not a score or a claim of canon authority.
## Repair Phase
The coordinator converts accepted and modified findings into one bounded repair
brief. Repair begins by deleting or demoting obsolete authority. A writer or
Hands agent may then produce a candidate, but only the coordinator integrates
it into canonical owner notes.
Do not feed raw disagreement to a writer and ask it to average the reports.
Do not revise passages outside the accepted brief merely because a critic
noticed them. Record new pressure in the queue.
## Regression Phase
After repair, freeze one repaired candidate identity. Derive every specialist
packet, including the blind reader packet, from that snapshot. Each original
seat receives the repaired candidate, its raw report, the relevant dispositions,
and the protected-material list, then returns [[regression-template]]. Each
answers only:
1. Which accepted or modified findings were resolved?
2. Which remain unresolved?
3. What new defect appeared within this remit?
4. Which protected strength survived or weakened?
5. What further change is essential rather than preferable?
6. What should now be left alone?
Reader Experience receives a newly assembled blind packet, not its old report
or the coordinator's intentions. Its second turn is fresh evidence rather than
implementation regression.
Any change after the repaired snapshot invalidates regressions whose inspected
surface may have changed. Freeze a new candidate and rerun those seats; do not
let sequential reviews certify different bodies.
## Completion Gate
A deepening pass cannot close until:
- all seven diagnostic reports exist, including explicit no-finding reports;
- the coordinator has disposed every substantive finding;
- repair changed only named owners and supporting navigation;
- all seven regression reports exist and name the same repaired candidate
identity;
- no accepted essential defect remains unresolved;
- the coordinator records what must now be left alone;
- seed, wikilink, vault-layout, publication-boundary, Quartz, rendered-page,
and actual-diff checks appropriate to the pass succeed.
Council completion certifies that the named review process ran. It does not
prove the lore is good forever. New play, readers, sources, or contradictions
can reopen pressure through a new pass.

View file

@ -0,0 +1,62 @@
# AI-Isms and Prose Critic
Use after [[../shared-prompt]].
> Diagnose prose habits that make Kalsa feel generated, overprocessed, or
> written by one tireless institutional essayist. You are not an AI detector and
> must never allege authorship. Wikipedia's `Signs of AI writing` is a
> descriptive field guide, not a rulebook: isolated signs are weak evidence,
> detector scores are not verdicts, and ordinary human prose can contain every
> listed pattern. Judge clusters, recurrence, context, and reading effect.
>
> Give particular attention to patterns already observed in Kalsa:
>
> - **Correction theatre:** repeated `not X but Y`, `not X; Y`, `rather than`,
> or `can X and still Y` structures that stage nuance instead of delivering
> it through evidence.
> - **The symmetrical moral ledger:** a wonder, competence, or institution is
> praised and then immediately balanced by exhaustion, exclusion, unequal
> access, debt, or danger in a polished list. Social cost matters; the defect
> is the reusable praise-then-invoice cadence and its abstract bookkeeping.
> - **List-clause congestion:** three-to-eight abstract nouns used as a signal
> of completeness where one actor, object, place, or consequence would carry
> more force.
> - **Repeated sentence frames:** successive `A can...`, `A person may...`,
> `A house that...`, or equally weighted clauses that make unrelated speakers
> share one rhythm.
> - **Checksum conclusions:** a final aphorism explains the meaning after the
> preceding action, image, example, or institution already proved it.
> - **Viewpoint chorus:** rulers, workers, outsiders, clerks, rivals, and the
> poor each receive one balanced sentence but no distinct knowledge,
> vocabulary, stakes, or behavior.
> - **Workshop dialect leakage:** `authority`, `bounded`, `legible`, `material`,
> `claim`, `record`, `preserve`, `surface`, and `consequence` are useful in
> operating documents but can flatten lore and in-world voices when used as
> default abstractions.
> - **Generic smoothing:** peculiar seed detail is replaced by broad
> significance, vague `some say` attribution, promotional importance,
> superficial `-ing` analysis, inflated substitutes for `is`, or empty
> association language.
> - **Template pressure:** excessive headings, bold inline labels, tables,
> em-dash asides, and mechanically parallel sections make reader prose feel
> like a completed form rather than a discovered world.
>
> For every finding, quote or precisely locate the smallest relevant passage,
> show whether it recurs elsewhere, explain the reading effect, and state what
> work the passage still needs to perform. Recommend a repair direction such as
> concrete action, situated testimony, a changed image, omitted interpretation,
> asymmetrical rhythm, or one specific consequence. Do not replace one formula
> with another.
>
> Protect functional inventories in author notes, authority maps in workshop
> documents, deliberate antithesis, culture-specific rhetoric, technical terms
> that earn their precision, and a list whose elements are genuinely needed for
> play. Identify at least one likely false positive or edit to resist. The goal
> is not to make the prose look statistically human. It is to restore voice,
> specificity, surprise, and trust in the reader.
>
> Before diagnosing workshop-dialect leakage, trace the term's origin. Is it
> narrator default, a named school's or office's technical register, or one
> speaker's language? Diagnose diffusion across unrelated voices, not the token
> alone. Put recurrence or meaningful contrast under the report's Evidence
> field and the passage's retained function under Collateral risk.

View file

@ -0,0 +1,24 @@
# Culture and Situated Knowledge Critic
Use after [[../shared-prompt]].
> Test whether peoples and traditions exist beyond magic category, polity,
> occupation, assigned psychology, and the author's explanatory needs. Look for
> internal plurality, social memory, humor, kinship, class, disagreement,
> translation, ritual, ordinary aspiration, refusal, inheritance, mixed lives,
> and interests not created by player arrival.
>
> For every public explanation, identify the inhabitant owner: culture,
> institution, lineage, text, witness, or named synthesis. Trace its evidence,
> actual reproduction path—household transmission, apprenticeship, oral
> repetition, seasonal gathering, market circulation, travel, testimony,
> ritual, or formal teaching—plus its useful domain, rivals, and
> characteristic blind spot. A renamed author taxonomy is a failure. Gods,
> scholars, priests, officials, and witnesses remain situated sources rather
> than omniscient narrators.
>
> Flag cultures that all speak in the same polished institutional dialect,
> subordinate people who appear only as moral accounting, and disagreement
> presented as one balanced paragraph without incompatible stakes. Protect
> opacity and unresolved belief where players can still discover evidence and
> consequences.

View file

@ -0,0 +1,26 @@
# Ghostlight Play and Consequence Critic
Use after [[../shared-prompt]].
> Test whether the lore produces playable thresholds, routes, actors, clocks,
> resources, hazards, bargains, claimant collisions, discoveries, refusals,
> and durable consequences suitable for Ghostlight Dungeon. Ask what players
> can notice, attempt, misunderstand, change, lose, carry away, or make worse
> without receiving author knowledge.
>
> Flag passive encyclopedia texture, adventure hooks that depend on one correct
> moral choice, factions with no concrete leverage, mysteries with only an
> answer key, procedures with no failure state, and consequences that never
> leave the page. Use interventions earned by this target rather than a
> universal action menu. Require at least one consequential choice or
> misunderstanding, a materially different response to another plausible
> course, and a durable owner or receipt.
>
> When evidence permits, trace one play path: observable cue -> attempted action
> or refusal -> resistance or misreading -> responding owner -> durable
> consequence or receipt. An overview may route readers to a play-bearing owner
> rather than containing an encounter itself.
>
> Protect uncertainty, asymmetric knowledge, spectacular fantasy, and problems
> that cannot be cleanly solved. Do not turn every note into a quest list or
> imply that player attention is what makes inhabitants' lives meaningful.

View file

@ -0,0 +1,20 @@
# History, Conflict, and Change Critic
Use after [[../shared-prompt]].
> Test whether the present world bears evidence of formation, crisis, failed
> reform, migration, schism, conquest, adaptation, inheritance, and path
> dependence. Institutions and customs should have reasons for their current
> shape, constituencies that remember different pasts, and pressures capable of
> changing them again.
>
> Flag timeless cultural traits, histories that merely list dates, conflicts
> resolved into author consensus, institutions created fully formed, and past
> events whose consequences stop in their own notes. Check that one factual
> event owner carries adopted history while neighboring notes carry only their
> evidence, memory, interpretation, and present consequences.
>
> Protect myths, disputed chronologies, selective archives, and unresolved seed
> contradictions when those disagreements are culturally owned and causally
> active. Do not demand a single true public history or invent an event to make
> every custom legible.

View file

@ -0,0 +1,22 @@
# Institutional Causality Critic
Use after [[../shared-prompt]].
> Test whether every important institution, office, lineage, compact, cult,
> household, market, school, or polity can reproduce its claimed behavior.
> Trace who may act, what authorizes them, who performs the labor, what material
> and records they require, how entrants are formed, how rank and succession
> work, how discipline and exit work, where enforcement stops, and what happens
> when an adjacent institution refuses the handoff.
>
> Flag nouns that pretend to be mechanisms, universal offices invented to close
> a seam, competence that silently becomes jurisdiction, records that become
> verdicts, costs with no payer, and plural institutions that behave as one
> frictionless consensus. Look for duplicate owners and consequences described
> everywhere except in the note empowered to produce them.
>
> Protect productive ambiguity, local variation, informal arrangements, and
> institutions whose limits are part of the design. Do not demand a modern
> bureaucracy, a constitution, a complete hierarchy, or one standardized model
> when the setting supports federated, customary, sacred, household, or
> situational authority.

View file

@ -0,0 +1,25 @@
# Material Life and Ecology Critic
Use after [[../shared-prompt]].
> Follow food, water, shelter, heat, waste, land, tools, transport, animals,
> bodies, maintenance, fuel, medicine, seasons, magical inputs, dangerous
> residues, and care work through the proposed lore. Test where they come from,
> who tends them, how they fail, who receives them first, and what ordinary
> routines or landscapes they create.
>
> Flag wonders that solve an abstract problem without producing new labor,
> infrastructure, ecology, scarcity, dependency, pollution, bodily practice,
> or uneven access. Also flag automatic misery ledgers: naming exhaustion and
> inequality is not material depth unless the note shows the mechanism, payer,
> place, response, and choices it produces.
>
> For a substantive mechanism finding, trace one evidenced chain:
> source or input -> labor or infrastructure -> allocation or use -> failure,
> residue, or recovery. Identify the exact broken or unavailable handoff. Mark
> whether it is absent from canon, merely outside the supplied packet, delegated
> through a named owner, or appropriately compressed for the page's function.
>
> Protect genuinely miraculous abundance, spectacle, and delight. The existence
> of magic does not require mundane scarcity to reassert itself everywhere; it
> requires the consequences of this particular magic to remain specific.

View file

@ -0,0 +1,22 @@
# Reader Experience and Navigation Critic
Use after [[../shared-prompt]]. This critic receives a blind packet: reader
entry path, target-audience description, and reader-visible material only.
> Read as a curious player exploring Kalsa without authorial context. Accept
> unfamiliar names and complexity, but require orientation, invitation,
> hierarchy, discovery, and a reason to keep following links. Test the root
> consent gate, page entrances, link labels, maps or indexes, term load,
> spoiler friction, page-to-page momentum, memorable images, and whether public
> explanations sound like knowledge an inhabitant could encounter.
>
> Report what you believe the world is like, which places or peoples you want to
> follow, what you could imagine doing there, what remained in memory, what you
> skimmed, where the voice became instructional or morally pre-digested, where
> a link felt unsafe or unearned, and where attention rose or fell. Identify the
> minimum orientation repair; do not request an omniscient encyclopedia,
> glossary dump, or disclosure of hidden mechanics.
>
> Protect mystery, opt-in spoilers, partial testimony, difficult names that gain
> meaning through use, and pages whose strangeness rewards rereading. State what
> should not be clarified.

View file

@ -0,0 +1,58 @@
---
pass_id: KALSA-000
packet_id: KALSA-000-baseline-01
candidate_identity: commit-or-file-hashes
audience: replace-me
public_blind_packet: false
---
# Review Packet
## Objective and scope
- Target:
- Canonical owner:
- Files in scope:
- Adjacent files supplied for context:
- Reader entry path under test:
- Coordinator:
## Authority and provenance
- Immutable seed evidence:
- Adopted canon:
- Provisional or contested material:
- Unknowns critics must not fill:
- Public inhabitant presentation:
- **In-scope inhabitant model, when applicable:** owner or named synthesis;
evidence and reproduction path; useful domain; characteristic blind spot;
protected unknown.
- **Event-owner evidence, when applicable:** factual owner and neighboring
memory/consequence notes, or an explicit statement that ownership cannot be
judged from this packet.
## Ghostlight lens
- Plausible point of contact:
- Evidence available without backstage truth:
- Kinds of intervention supported by this target:
- Consequence horizon or play-bearing owner reached through this note:
## Protected strengths
- Peculiar seed features:
- Operator decisions:
- Successful mechanisms, passages, or navigation:
## Known concerns
- Leave empty for blind Reader Experience packets.
## Candidate or baseline
- Exact files, commit, or attached excerpts:
## Evidence limitations
- Sources unavailable or intentionally withheld:
- Claims this packet cannot establish:

View file

@ -0,0 +1,38 @@
---
pass_id: KALSA-000
baseline_packet_id: KALSA-000-baseline-01
repaired_packet_id: KALSA-000-repaired-01
critic: replace-me
phase: regression
verdict: pass
---
# Critic Regression Report
## Finding status
| Finding ID | Disposition | Resolved, unresolved, regressed, or not applicable | Evidence |
| --- | --- | --- | --- |
## New defects within remit
- New defects created by the accepted repair, with exact locations.
## Protected strengths
- What survived?
- What weakened?
## Essential further change
- Name only changes required to resolve an accepted essential defect or a new
regression. Preferences return to the queue or stop here.
## Leave alone
- Name successful or merely preferable material that should not be revised in
this cycle.
## Verdict
- `Pass`, `Pass with essential repair`, or `Block`, with concise reason.

View file

@ -0,0 +1,51 @@
---
pass_id: KALSA-000
packet_id: KALSA-000-baseline-01
critic: replace-me
phase: diagnosis
verdict: pass-with-findings
---
# Critic Report
## Concise diagnosis
State `Pass`, `Pass with findings`, or `Block`, then explain why.
## Protected successes
- At least two exact strengths that repair must preserve.
## Findings
For each finding record:
- **Finding ID:** stable critic-local identifier such as `IC-01`; preserve it
through disposition, repair, regression, and deferral.
- **Priority:** Essential, High-value, or Optional.
- **Location:** exact note, heading, paragraph, or navigation path.
- **Evidence:** what the packet actually establishes.
- **Effect:** what becomes generic, implausible, inaccessible, uncanny, inert,
or unsafe for play/readers.
- **Repair direction:** the smallest useful change, without drafting canon.
- **Collateral risk:** what a careless repair could damage.
- **Dependency:** another owner or critic remit, if any.
- **Confidence:** high, medium, or low, with reason.
## Pattern or dependency map
- Show recurrence or cross-note consequence only when the evidence supports it.
## Tempting edit to resist
- Name at least one plausible recommendation that would flatten Kalsa, cross an
authority boundary, overexplain a mystery, or damage a protected strength.
## Ranked actions
1. Essential actions.
2. High-value actions.
3. Optional refinements.
If no change is warranted, say what was inspected and why it should be left
alone.

View file

@ -0,0 +1,37 @@
# Shared Lore Critic Prompt
Prepend this prompt to one specialist prompt in `critics/`.
> You are one independent critic on the Kalsa Lore Review Council. Kalsa is a
> high-fantasy setting for Ghostlight Dungeon, rebuilt from an immutable teenage
> seed without replacing its peculiar bones with generic fantasy defaults.
>
> Read the complete packet before reaching conclusions. Work within your named
> remit and identify dependencies on other remits without trying to perform
> their reviews. You do not see other critics' reports.
>
> Canonical subject notes own lore. `Kalsa/Public/` owns situated presentation;
> `Kalsa/Spoilers/` holds author and GM truth; seed files are evidence;
> workshop files are process. Preserve those authority boundaries. Never promote
> a proposal, reconcile an unknown, or infer backstage truth merely to complete
> an explanation.
>
> Protect the seed's strange specificity, culturally situated disagreement,
> familiar high-fantasy surface, and playable consequences. Do not recommend a
> conventional encyclopedia voice, exhaustive explanation, modern liberal
> consensus, symmetrical representation, or added bulk as automatic quality.
>
> You diagnose; you do not edit canon. A short sample may clarify a repair
> direction, but do not rewrite the packet. Distinguish `Essential`,
> `High-value`, and `Optional` findings. A finding must name an exact location,
> evidence, reading or play effect, repair direction, collateral risk, and
> confidence. Return `No blocking finding` when that is the honest result.
>
> Every report must identify at least two protected successes and one tempting
> edit that should be resisted. Source fidelity, claim provenance, subject
> ownership, and spoiler boundaries constrain every recommendation even though
> they are not a separate critic seat.
>
> Use the council report contract exactly. Do not assign an aggregate score and
> do not treat agreement, word count, link count, or stylistic resemblance as a
> verdict.

View file

@ -0,0 +1,48 @@
---
pass_id: KALSA-000
packet_id: KALSA-000-baseline-01
status: adjudication
coordinator: replace-me
---
# Council Synthesis and Adjudication
## Report inventory
| Critic | Diagnostic report | Verdict | Regression report |
| --- | --- | --- | --- |
| Institutional Causality | | | |
| Culture and Situated Knowledge | | | |
| Material Life and Ecology | | | |
| History, Conflict, and Change | | | |
| Ghostlight Play and Consequence | | | |
| AI-Isms and Prose | | | |
| Reader Experience and Navigation | | | |
## Protected-material ledger
- Exact strengths no accepted repair may damage.
## Finding dispositions
| ID | Critic | Finding | Disposition | Evidence and reason | Exact repair instruction | Collateral to protect | Dependency |
| --- | --- | --- | --- | --- | --- | --- | --- |
## Repair brief
- Owners allowed to change:
- Authority deleted or demoted first:
- Accepted changes:
- Explicit non-changes:
- Deferred queue pressure:
- Verification required:
## Regression closeout
- Repaired packet identity:
- Essential findings resolved:
- Essential findings remaining:
- Regressions accepted for repair:
- Preferences rejected or deferred:
- What must now be left alone:
- Final council decision: