TS — TFW-55 / Phase B: Multilingual Public Entry¶

Date: 2026-08-26 Author: Coordinator Status: 🟡 TS_DRAFT — approved by the owner 2026-08-26; ready for /tfw-handoff Parent HL: HL-TFW-55 Phase HL: Phase B derivation Phase A source: RF · REVIEW · Project North Star Master freeze: a60bc6d

1. Objective¶

Turn the repository root into a concise multilingual doorway without creating a second philosophy. English README.md remains the semantic source and the only Task Board; new README.ru.md and README.kk.md provide natural Russian and Kazakh localizations rather than literal translations. Each version must explain enough to understand TFW, choose an Edition, begin work, and find authoritative meaning/history in no more than 800 public-onboarding words.

2. Scope¶

In Scope¶

Replace the 1,485-word English public section above ## Task Board with a ≤800-word doorway derived from the reviewed Project North Star.
Create root-level Russian and Kazakh localizations, each independently composed in natural target-language prose, ≤800 words, and without Task Boards.
Make the same working English · Русский · Қазақша switch the first visible block in all three files.
Preserve one definition, proportional Editions selection, compact Quick Start, license/repository links, and understand/use/audit navigation in every language.
Make NS1, NS2, and NS3 directly discoverable from the English root and semantically represented in both localizations.
Remove duplicated philosophy, comparison, audience-card, FAQ, lifecycle, inventory, adapter, and update prose when authoritative linked material already owns it.
Run separate Russian and Kazakh naturalness/calque critiques and preserve their complete structured reports as dedicated evidence files, with the parity matrix, word/link checks, and subtraction ledger in EV/RF.

Out of Scope¶

Any edit to .tfw/README.md, Project North Star semantics, conventions, glossary, workflows, templates, Editions behavior/content, adapters, configuration, or runtime.
A full Russian/Kazakh translation of the Project North Star, specification, Task Board, task traces, or knowledge base.
BoK creation, book/guide/course content, university packaging, market/legal research, visual rebrand, documentation-site redesign, or launch work.
Proof of human comprehension, learning, adoption, or localization reception outside the bounded language critiques.

3. Principles Check¶

#	Principle from master HL §7	Enforced by	Gate
P1	One North Star, several depths	AC-1, AC-4, AC-7	Doorways link to North Star/mechanics/corpus and do not restate them as competing authority
P2	Philosophy before machinery	AC-1–AC-3	Problem and definition precede Editions and Quick Start in every language
P3	Human purpose remains human	AC-4	Six semantic invariants preserve authority/accountability boundaries
P4	Trace, not transcript	AC-4	Localization matrix rejects chat/archive/determinism drift
P5	Self-awareness must be operational	AC-4	Doorways omit the term or preserve its bounded capability; no anthropomorphic shorthand
P6	Subtract before adding	AC-8	Fixed word ceilings, combined ceiling, and content disposition ledger
P7	Provenance before polish	AC-4, AC-9	English source declaration, parity matrix, language reports, and claim dispositions
P8	Experience reveals the system	AC-6	Editions are proportional and solve visible work/risk needs, not a maturity badge
P9	Refutation before canonicalization	AC-4, AC-9	Language reviewers actively search for missing/extra promises and source drift

4. Affected Files¶

File	Action	Description
`README.md`	MODIFY	English doorway above the existing Task Board; Task Board remains English-only
`README.ru.md`	CREATE	Derived Russian public doorway; no Task Board
`README.kk.md`	CREATE	Derived Kazakh public doorway; no Task Board
`phase-b/evidence/EV__phase-b__multilingual_public_entry.md`	CREATE	Per-AC evidence, parity/terminology matrices, language critiques, word/link/board/subtraction checks
`phase-b/evidence/LANG_REVIEW__phase-b__ru.md`	CREATE	Separate final Russian naturalness, calque, independence, and semantic-equivalence report
`phase-b/evidence/LANG_REVIEW__phase-b__kk.md`	CREATE	Separate final Kazakh naturalness, calque, independence, and semantic-equivalence report

Budget: 6 delivery/evidence files; 5 new; 1 production modification; estimated ≤900 changed lines. The mandatory ONB and RF add two normal phase traces, for 8 execution-stage files total (7 new, 1 modified). Within configured limits: 30 files, 15 new files, 3,000 LOC, 30 modified files.

5. Acceptance Criteria¶

AC-1: English doorway replaces duplicate exposition¶

English README.md becomes a small entry point to the reviewed North Star rather than a second essay.

[ ] The public section begins with the language switch/source notice, existing brand identity, a concrete continuity problem, and one functional TFW definition before machinery.
[ ] The definition preserves Philosophy of Trace → TFW methodology → proportional realizations and the human/agent authority boundary.
[ ] Direct links expose NS1 — Purpose, NS2 — Principles, and NS3 — Non-goals in .tfw/README.md.
[ ] Detailed audience cards, FAQ, comparisons, lifecycle/status explanation, file inventories, adapter tables, and update walkthrough are removed or represented only by authoritative links.
[ ] No deterministic, automatically self-documenting, same-artifact, raw-chat-memory, agent-authority, or component-novelty claim remains.

Gate: Manual semantic/role inspection plus required old-content disposition ledger in EV.

Evidence: N/A — this is a repository-local document transformation with no live-environment behavior claim.

AC-2: Russian doorway is a natural, usable localization [depends: AC-1]¶

README.ru.md communicates the English doorway naturally in Russian without becoming an independent source of TFW meaning or sounding like English syntax rendered word for word.

[ ] It carries every required semantic, Editions, Quick Start, navigation, source, repository, and license unit from the English doorway.
[ ] It declares that English README.md and the English Project North Star are the semantic source.
[ ] TFW, Trace-First Workflow, Edition names, commands, anchors, and paths remain unchanged; localized concepts follow the terminology contract.
[ ] It introduces no new promise, authority rule, capability, product scope, or interpretation.
[ ] Its paragraph order and phrasing are chosen for Russian clarity; it contains no material calque, mechanical mirroring, or translation smell and may reuse natural Russian formulations derived from .tfw/README.md semantics.
[ ] It contains no Task Board, task-status table, or translated operational project state.
[ ] It is valid UTF-8 with no replacement characters or mojibake.

Gate: Semantic-unit parity matrix plus fresh isolated Russian critique under AC-9.

Evidence: N/A — language quality is captured by bounded document critique in EV, not a live user study.

AC-3: Kazakh doorway is a natural, usable localization [depends: AC-1]¶

README.kk.md communicates the English doorway naturally in Kazakh without literal-calque drift or independent authority.

[ ] It carries every required semantic, Editions, Quick Start, navigation, source, repository, and license unit from the English doorway.
[ ] It declares that English README.md and the English Project North Star are the semantic source.
[ ] TFW, Trace-First Workflow, Edition names, commands, anchors, and paths remain unchanged; localized concepts follow the terminology contract.
[ ] It introduces no new promise, authority rule, capability, product scope, or interpretation.
[ ] Its paragraph order and phrasing are chosen for Kazakh clarity; it contains no material calque, mechanical mirroring, or translation smell and may reuse natural Kazakh formulations derived from .tfw/README.md semantics.
[ ] It contains no Task Board, task-status table, or translated operational project state.
[ ] It is valid UTF-8 with no replacement characters or mojibake.

Gate: Semantic-unit parity matrix, terminology/back-translation inspection, and fresh isolated Kazakh critique under AC-9.

Evidence: N/A — language quality is captured by bounded document critique in EV, not a live user study.

AC-4: Semantic parity and authority remain explicit [depends: AC-1, AC-2, AC-3]¶

All three doorways preserve the same six invariants while allowing natural language-specific structure.

[ ] EV maps each invariant to exact sections in English, Russian, and Kazakh.
[ ] Human purpose, legitimate authority, judgment, acceptance, accountability, and stop responsibility remain human/institutional in all versions.
[ ] Trace remains selected durable continuity, not transcript, hidden reasoning, automatic truth, or deterministic reproduction.
[ ] If “self-aware project” appears, it is operationally bounded; omission from the short doorway is allowed.
[ ] English is identified as the semantic source; localizations explicitly yield on conflict.
[ ] The root documents point to the existing North Star and do not create another canon, BoK, or governance surface.
[ ] Semantic units are equivalent, but paragraph counts, sentence order, and local phrasing are not required to mirror English.

Gate: Three-column invariant/authority matrix in EV; zero missing units and zero contradictory/extra claims.

Evidence: N/A — parity is directly inspectable in versioned files.

A visitor can change language without searching, and every repository-local destination resolves.

[ ] English · Русский · Қазақша is the first visible block in all three files with correct relative targets.
[ ] All three files link to Project North Star, Editions, Quick Start, Task Board/history route, repository, and license as applicable.
[ ] Russian and Kazakh audit links lead to the English Task Board/corpus rather than duplicating operational state.
[ ] Every local file/directory link and every required #ns1/#ns2/#ns3 anchor resolves.
[ ] Commands, paths, URLs, and Markdown link destinations are byte-identical across languages where they represent the same destination.

Gate: Automated local link/anchor/target comparison with results recorded in EV.

Evidence: N/A — link resolution is a reproducible local check.

AC-6: Editions and Quick Start remain factual and usable¶

Simplification must not make current use harder to discover or reintroduce unsupported capability claims.

[ ] Light points to editions/01-light/, Assisted to editions/02-assisted/, and Full to .tfw//the Full workflow route.
[ ] Each Edition is explained by work complexity/risk and availability, not prestige or universal maturity.
[ ] Assisted does not promise durable automatic hook dispatch or measured superiority over Light.
[ ] Each language gives usable first actions for a new project, an existing project, and an already-configured project, or one equally functional compact sequence.
[ ] /tfw-plan and other exact commands/paths are never translated or altered.

Gate: Check every Edition/path/command against current editions/README.md, .tfw/quickstart.md, and D57–D60; run a cold navigation inspection from each doorway.

Evidence: N/A — no installation is performed; current documentation facts and navigation are verified locally.

Every doorway answers where to go next without reproducing the destination.

[ ] Understand: direct Project North Star/definition/purpose route.
[ ] Use: Editions selection and Quick Start route.
[ ] Audit: English Task Board, task traces/knowledge, or equivalent history/evidence route.
[ ] Repository, author/attribution where retained, version where retained, and MIT license remain discoverable.
[ ] The English root visibly completes the root + .tfw/README.md Project North Star designation; Russian/Kazakh identify themselves as derived doorways.

Gate: Three-route walkthrough for each language in EV, with exact click sequence and destination.

Evidence: N/A — navigation paths are repository-local and directly inspectable.

AC-8: Budgets, subtraction, and Task Board integrity hold¶

The new entry is smaller, not a multilingual duplication of the old information architecture.

[ ] English public content before ## Task Board is ≤800 whitespace-delimited words; ## Task Board and the entire section after it are excluded from this metric.
[ ] Entire README.ru.md and README.kk.md are each ≤800 whitespace-delimited words.
[ ] English public content plus .tfw/README.md is ≤2,600 words; measured before/final/net counts are recorded.
[ ] English README.md contains exactly one ## Task Board; Russian and Kazakh contain zero Task Board headings/rows.
[ ] From the approved Phase B TS baseline, all English Task Board lines except the normal TFW-55 status row remain byte-identical.
[ ] EV records removed, linked, retained, localized, and newly added content classes; added language/source text is funded by subtraction, while Task Board length has no effect on the onboarding budget or content cuts.

Gate: Reproducible raw-Markdown word counts, heading/row counts, board-tail comparison excluding only the TFW-55 row, and subtraction ledger in EV.

Evidence: N/A — size and board integrity are deterministic file checks.

AC-9: Independent language critiques close material defects [depends: AC-2, AC-3, AC-4]¶

Russian and Kazakh receive separate fresh critiques before RF; the executor may not self-certify both localizations.

[ ] Each critique receives only its frozen semantic packet, the English doorway, target localization, and fixed rubric—no founder intent or preferred verdict.
[ ] Each reports invariant coverage, drift, missing/extra promise, authority, terminology, target-language naturalness, English-syntax calques, translation smell, independent-localization quality, command/path integrity, and severity.
[ ] The Russian reviewer provides an English back-translation of the definition; the Kazakh reviewer provides both English and Russian back-translations of the definition for triangulation.
[ ] Every high-severity semantic, factual, authority, navigation, or serious-idiom defect is corrected and rechecked before RF.
[ ] Pure stylistic preference without material meaning/usability impact is recorded as non-blocking and does not trigger repeated revision.
[ ] evidence/LANG_REVIEW__phase-b__ru.md and evidence/LANG_REVIEW__phase-b__kk.md preserve the complete final structured reports; EV links them, records executor dispositions, and confirms zero unresolved high-severity defects.

Gate: Validate reviewer isolation packet, required report fields, disposition table, and final severity counts in EV.

Evidence: N/A — these are bounded language/document critiques, not evidence of population-level reception.

AC-10: Provenance and scope are reviewable¶

The normal Phase B trace explains the rewrite without creating a public localization registry.

[ ] EV maps AC-1–AC-10 to exact files/sections and verification results.
[ ] RF records semantic source, terminology decisions, old-content dispositions, word counts, link results, board integrity, language reports, and unresolved non-blocking risks.
[ ] No production file outside the three root READMEs is changed.
[ ] .tfw/README.md remains byte-identical to the Phase A reviewed blob 71a4d725cff7d0d7508403589195e9f87a0fc49a.
[ ] No BoK, translation subsystem, glossary/specification change, public roadmap, or canon manifest is created.
[ ] EV and RF contain no placeholders and agree on every count/verdict before handoff to review.

Gate: Path-scope diff, reviewed-blob comparison, placeholder scan, and EV↔RF consistency check.

Evidence: N/A — provenance and scope are established by repository traces and hashes.

Evidence Artifact¶

File	Description
`phase-b/evidence/EV__phase-b__multilingual_public_entry.md`	Required structured evidence: per-AC verdicts, invariant/terminology matrices, Russian/Kazakh critique reports and dispositions, word/link/anchor/board checks, navigation routes, and subtraction ledger
`phase-b/evidence/LANG_REVIEW__phase-b__ru.md`	Separate Russian naturalness/calque/independence and semantic-equivalence report against the final file
`phase-b/evidence/LANG_REVIEW__phase-b__kk.md`	Separate Kazakh naturalness/calque/independence and semantic-equivalence report against the final file

6. Technical Guidance¶

Source order¶

Executor reads only the sources needed for the doorway, in this order:

Master HL §§1, 3–7 and Amendments A2–A5.
Phase B HL.
Reviewed [.tfw/README.md](../../../concepts/philosophy.md), especially the definition, proportional realizations, NS1–NS3, and authority routing.
Phase A RF and REVIEW for evidence limits and non-blocking root-alignment findings.
Current README.md public section for subtraction and working links.
editions/README.md, .tfw/quickstart.md, LICENSE, and current edition paths only to verify usage facts.

Semantic packet¶

The fixed semantic packet used for both localizations and language critiques contains:

the final English doorway;
the relevant definition, Trace, human-authority, proportional-realization, and NS1–NS3 passages from the approved .tfw/README.md, which may supply natural formulations without becoming a second semantic source;
the six semantic invariants in the Phase HL;
the terminology contract;
the required navigation/Quick Start units;
explicit exclusions: no extra promise, authority, maturity claim, automation claim, or Task Board.

Implementation guidance¶

Rewrite the English public section as a unit; do not preserve its current heading structure merely because it exists.
Treat the current Task Board marker as a hard boundary. Capture the approved TS-baseline board tail before editing; later comparisons may ignore only the TFW-55 row used for normal status transitions.
Raw word-count convention: count whitespace-delimited non-empty tokens in Markdown source. English count stops immediately before ## Task Board, excluding the heading and every Task Board line; Russian/Kazakh count the whole file; .tfw/README.md counts the whole file. Never trade away required onboarding meaning because the operational board is long.
Keep the existing logo and calm protocol-grade identity where they fit inside the budget; do not create or edit visual assets.
Preserve brand/code identifiers exactly: Trace-First Workflow, TFW, Light/Assisted/Full, /tfw-*, paths, URLs, NS1–NS3.
Write and validate all three READMEs as strict UTF-8; fail on replacement characters or common mojibake sequences.
Localization topology should differ when natural target-language composition benefits. Semantic-unit parity is mandatory; sentence-by-sentence, paragraph-by-paragraph, and syntax mirroring are not.
Natural formulations and explanatory moves from the approved .tfw/README.md are allowed and encouraged when they improve Russian or Kazakh without changing the English public contract.
Use two fresh isolated Codex tasks or equivalent bounded reviewers for the language critiques. They critique; they do not edit production files or issue the formal TFW REVIEW verdict.
A language concern blocks only when it materially changes meaning, authority, factual accuracy, navigation, or ordinary comprehension. Record tasteful alternatives without cycling.
Do not edit a linked source to make a doorway sentence true. Correct the doorway or record a future finding.

7. Definition of Failure¶

❌ Any doorway exceeds 800 words or combined English explanatory content exceeds 2,600.
❌ Russian/Kazakh adds meaning, authority, promises, capabilities, or operational state absent from English.
❌ English loses the Task Board, another file gains a Task Board, or a non-TFW-55 Task Board line changes.
❌ Language switch, required local links, NS1–NS3, Editions paths, commands, or license navigation break.
❌ The doorway again claims deterministic reproduction, automatic self-documentation, same artifacts/lifecycle, agent authority/accountability, universal pedagogy, or untested capability.
❌ Russian or Kazakh has an unresolved high-severity language/meaning defect, reads as materially calqued/mechanically mirrored English, or lacks its separate final critique report.
❌ Quick Start becomes too abstract to give a new, existing, or configured project a usable next action.
❌ .tfw/README.md, framework mechanics, Editions content, visual identity assets, BoK, book/course/guide, or documentation site changes enter the phase.
❌ EV/RF omits a required matrix/report/count/disposition, contains placeholders, or reports inconsistent results.

8. Phase Risks¶

Risk	Mitigation
English compression becomes generic marketing	Six invariant checks, North Star direct links, and removal/retention ledger
Russian source familiarity silently overrides English	Explicit source declaration and three-column parity matrix
Kazakh reviewer produces fluent but semantically shifted text	Definition back-translation, terminology probes, and authority/promise rubric
Reviewer cycles on stylistic taste	Materiality rule and severity classification; only high-severity defects block
Task Board changes during concurrent work	Compare all board lines except the TFW-55 workflow row rather than relying on one whole-file hash
Public usage facts go stale during rewrite	Read current Editions/Quick Start immediately before drafting and verify paths/commands in RF

9. Cross-Phase Modifications¶

File	Also involved in	Coordination note
`README.md`	Phase A control trace	Phase A changed only the TFW-55 Task Board row; Phase B owns public prose and must preserve the rest of the board
`.tfw/README.md`	Phase A production result	Read-only semantic source in Phase B; reviewed blob must remain unchanged

TS — TFW-55 / Phase B: Multilingual Public Entry | 2026-08-26