HL — TFW-55 / Phase A: Project North Star Essay¶

Date: 2026-08-26 Author: Coordinator Status: ✅ APPROVED — owner-approved derivation 2026-08-26; master contract remains frozen at a60bc6d Parent HL: HL-TFW-55 Master freeze: a60bc6d — [codex/TFW-55/freeze/coordinator] re-freeze owner-approved canon architecture Research: Iteration 1 · Iteration 2 Authority: derivation-only; the master HL owns Vision, Target State, Phases, DoD, DoF, and Principles

Context¶

Phase A rewrites one production file: .tfw/README.md. It becomes the English Project North Star and shortest complete explanation of TFW's stable meaning. It does not rewrite the operational specification, the root landing pages, or the working Body of Knowledge.

The owner has selected this architecture after research:

Philosophy of Trace
        ↓ grounds
TFW methodology
        ↓ is realized by
workflows and implementations
        ↓ one self-applying reference example
this repository

Methodology is an owner positioning decision, not a category or novelty established by Iteration 2. The public explanation must therefore earn the word through a clear function, minimum composition, exclusions, and observable consequences.

Source Authority for Phase A¶

Source	Phase A role	Not allowed
`Книга TFW/мини-эссе 0.8/` — nine essays	Primary author voice, problem-led order, metaphors, and conceptual language	Copying the series wholesale or preserving an overclaim because it reads well
`Книга TFW/body of knowledge/00. Архитектура TFW.md`	Owner-approved architecture and working definition	Treating the external Obsidian file as repository authority after the rewrite
`Книга TFW/body of knowledge/version 0.1.md`	Completeness, contradiction, hypothesis, and risk checklist	Calling the draft BoK canonical, hiding its open status, or expanding Phase A into a BoK
TFW-55 Iteration 1 and 2 RES	Evidence limits, competing explanations, F1–F4/R1 completeness checks	Exposing research labels as the essay's teaching structure or claiming human-learning evidence
Current `.tfw/README.md` and repository corpus	Material to retain, qualify, subtract, and link	Treating every historical/current mechanism as timeless meaning

The Result the Owner Will Review¶

Definition contract — semantic preview, not mandatory final prose¶

TFW is a methodology for joint work between people and AI, grounded in the Philosophy of Trace. It applies when a person delegates part of intellectual work to agents: the person retains purpose, authority, judgment, acceptance, accountability, and the responsibility to stop; agents perform bounded work; selected durable traces preserve enough material context, decisions, result/current state, and continuation for the work to be inspected and resumed.

The essay should reach this answer through the reader's problem, not open with a compressed definition dump. It must not begin with HL/TS/RF names, installation, or the Full lifecycle.

Problem-led narrative map¶

F1–F4 and R1 remain internal completeness tests. Public prose uses ordinary terms such as Goal, Value, Question, Task, Context, Boundaries, Trace, authority, result/current state, and continuation.

Order	Reader's question	Required answer	Source emphasis	Target words
1	Why does ordinary AI work disappear?	Much intellectual work becomes invisible; an agent can produce an output without preserving why it exists or how to continue	Mini-essays 01–02	150–190
2	Why is chat not enough?	Model memory and chat history are useful raw material, not selected organizational memory or authority	Mini-essay 03	120–160
3	What is a Trace?	A selected, durable, inspectable material record that makes purpose, decisions, state, and continuation visible; not transcript, private reasoning, or automatic truth	Mini-essay 04 + RES	190–240
4	What makes work continuable?	Goal and Value orient; Questions expose uncertainty; Tasks bound work; Context and Boundaries prevent drift; result/current state and next step or close preserve continuity	Mini-essays 05–06	230–280
5	What remains human when agents do the work?	Purpose, legitimate authority, judgment, acceptance, accountability, and stop decisions remain explicit; agents receive bounded work	Mini-essays 02, 05 + RES	170–210
6	How can work “remember itself”?	Durable artifacts answer six operational questions and allow another human or agent to continue; this is neither consciousness nor independent authority	Mini-essay 07 + master §3	150–190
7	What exactly is TFW?	Philosophy of Trace → TFW methodology → realizations/workflows; repository = self-applying reference implementation	Mini-essay 08 + architecture note	190–240
8	Why are Light, Assisted, and Full different?	They are proportional realizations; each adds coordination/assurance when risk requires it, and one observed teaching route—not the definition or a universal maturity law	Mini-essay 08 + RES	110–150
9	What is stable, what is not, and where next?	`NS1` purpose, `NS2` principles, `NS3` non-goals; North Star/spec/corpus/future-BoK authority; links to use and audit	Architecture note + master A2/A4	250–320
	Essay ceiling	Coherent English essay readable in one sitting		≤ 2,000

Project North Star contract¶

The essay contains three visible, citable anchors:

ID	Content	Minimum test
`NS1 — Purpose`	Why TFW exists and what value it protects when intellectual work is delegated	A coordinator can reject work that does not preserve purpose, human authority, inspectability, or continuation
`NS2 — Principles`	A small stable set grounded in the Philosophy of Trace	Each principle changes an observable work choice; no slogan-only list
`NS3 — Non-goals`	What TFW must not become	Explicitly rejects: prompt/chat archive; deterministic generator; replacement for judgment; maximum-documentation bureaucracy; vendor-bound tool; untested capability claim

Phase B will point to these same IDs from all three root READMEs. It must not invent a second North Star.

Authority visible in the essay¶

"Why does TFW exist; what must remain stable?" → Project North Star (.tfw/README + root pointer)
"What concepts and open questions need depth?" → future approved, versioned BoK (separate task)
"How does this implementation work today?"    → conventions · glossary · workflows · Editions
"Why did it become this?"                     → tasks · verified knowledge · Git history

The working Obsidian BoK remains source material. Phase A may explain the future authority slot in one concise sentence, but it must not create, publish, link as canonical, or pre-design the BoK.

Values inherited from the master contract¶

Value carried into the essay	Observable consequence in the text
Questions before premature answers	Open uncertainty is visible before a task is treated as solved
Human purpose remains human	No sentence gives an agent legitimate authority or accountability
Bounded delegation	Agent work has purpose/context, boundaries, and acceptance authority
Trace, not transcript	Selected material continuity is preserved; hidden reasoning is never required
Continuation over isolated output	A result identifies authority, current state, and next step or explicit close
Proportional assurance	Evidence, review, and verified knowledge scale with risk; Full artifacts are not universal minima
Self-application without mysticism	The repository can explain its change history; “self-aware” means inspectable capability only
Subtraction before addition	North Star content is funded inside the unchanged 2,000-word ceiling

Deliverables¶

Rewrite .tfw/README.md under the problem-led narrative and unchanged 2,000-word ceiling.
State one functional definition and the approved Philosophy of Trace → TFW methodology → realizations architecture.
Define selected Trace, continuability, human/agent boundaries, operational self-awareness, and proportional assurance in domain-independent language.
Designate NS1 Purpose, NS2 Principles, and NS3 Non-goals as the .tfw/README.md half of the Project North Star.
State North Star / future approved BoK / living specification / corpus authority without creating a new surface.
Remove or qualify deterministic, code-centric, self-maintaining, same-artifact, provenance-only, absolute memory-ownership, and unbounded agent claims.
Preserve source and subtraction decisions in the normal Phase A evidence/RF trace, not in a public claims register.

Explicitly Not in Phase A¶

root README.md, README.ru.md, and README.kk.md — Phase B after Phase A REVIEW;
the BoK — a separate next task after Phase A REVIEW;
changes to workflows, templates, artifact contracts, Editions, adapters, configuration, or runtime behavior;
a methodical guide, book, course, deck, university package, commercial offer, or brand redesign;
proof of learning efficacy, terminology reception, adoption, market demand, or category novelty.

Phase-Local Risks¶

Risk	Control in TS
The essay becomes a summary of nine source essays	One narrative, ≤2,000 words, one job per section; source wording is material rather than structure to preserve verbatim
`methodology` is presented as research-proven elevation	Function/composition/exclusions lead; category and component novelty are not research claims
The North Star becomes slogans or a new registry	Three citable sections live inside the essay and have observable reject/accept tests
The draft BoK silently becomes authority	It is a checklist only; future approval requires a separate task after Phase A REVIEW
Founder language overclaims reproducibility, expertise, observability, or memory ownership	Required claim-disposition ledger narrows each claim before publication
Full mechanics creep into the opening	Artifact inventories and lifecycle explanation are excluded; living specification owns detail
Editions become pedagogical doctrine	Describe one observed problem-led route and proportional implementations; preserve evidence limits

HL — TFW-55 / Phase A: Project North Star Essay | 2026-08-26