Loomworks — C5 seed-amendment session handoff

Version. 0.1 Date. 2026-06-04 Status. Handoff. Opens a fresh chat to make the seed-amendment decision that gates C5 (the Domain layer). Report-and-decide session; no build, no CR. Audience. The next Claude.ai chat (scoping/decision role) + Operator.

Plain-language summary

This chat exists to make one foundational decision: does the Loomworks seed's Memory model widen from "the engagement's Memory" (what seed v0.9 commits to today) to N-scope Memory (organization / team / role / domain / jurisdiction scopes above the engagement) — and if it does, what does a Manifestation organize: one engagement's Memory, or the composed N-scope view?

This is the last gate in front of scoping C5. It is Operator-authority, seed-level work — not a code decision. The output is a seed revision (v0.10) or a decision not to widen. It is a Discovery-record moment: the seed deliberately held the engagement-scoped line, so any widening must name that prior position alongside the new one.

Do not draft a C5 scoping note, CR, or any build artifact in this chat. The seed decision comes first. Scoping C5 against a seed that contradicts it is the silent-conflict path the project forbids.

How C5 got here (the trajectory, so it isn't lost)

A prior session (2026-06-04) settled C2's bookkeeping, then worked the question "what builds next." The path:

1. M1 (thin-seed un-weld) is built and merged — the lone committed MUST is done. Everything below is COULD until a concrete need promotes it.
2. C1 (title-review at induction) was considered and set aside. It was the adjacent, finishable candidate — but a CC staleness read showed M1 dissolved its motivation: C1 existed to fix bad auto-titles from thin seeds; M1 made seeds rich, so auto-titles now derive from substantive content. C1 stays COULD, motivation-weakened. (Correction preserved: an earlier turn had recommended C1 as next; the M1 finding reversed that.)
3. C5 (Domain layer) was chosen as the foundational bet — large, parallel-capable, the hard prerequisite for C6 (knowledge elevation), and unlike C1 its value did not erode.
4. A read-only CC alignment audit (engine main HEAD eb572c0) oriented C5 against seed v0.9 and surfaced two gates in front of scoping.
5. Gate 2 (naming collision) was resolved and filed this session — see below.
6. Gate 1 (this seed amendment) is the remaining gate — and the reason this chat exists.

What C5 is (so the decision has context)

C5 the Domain layer turns the seed's single-engagement Memory into N-scope Memory, in two halves:

• N-space Memory (substrate): Memory scopes above the engagement — organization, team, role, domain, jurisdiction, campaign, partner, audience — reached via membership chains (engagement → organization → sub-domain → domain). A new tier above the engagement.
• N-domain composition (use): composing from N scopes at every consequential moment — hierarchical floor (e.g. a jurisdiction sets a floor lower scopes can't violate), orthogonal contribution (scopes add non-conflicting pieces), override-with-attestation (a specific scope overrides a default, visibly). A render draws brand from organization scope, content from engagement scope, disclaimers from jurisdiction scope.

It does not add a fifth room. It extends Memory (room 1) and feeds Manifestation/Shaping/Rendering, which compose from the relevant scopes at production.

C5 unlocks: C6 (knowledge elevation — hard dependency, "nowhere to promote to" without the domain tier), white-label/managed-hosting tenancy (a "structural consequence"), the Shared-Subject Engagement Pattern, and cross-engagement Companion awareness.

Backing docs (all report-only / investigation, none scoped or built):

• investigations/loomworks-memory-space-extensibility-investigation-v0_1.md — the substrate half (N-space Memory). Note: the master build plan's C5 pointer omits this doc; it is half of C5.
• investigations/loomworks-engagement-domain-composition-investigation-v0_1.md — the use half (N-domain composition).
• investigations/loomworks-knowledge-elevation-pathway-investigation-v0_1.md — C6's doc; defines the domain hierarchy C5's store must hold; states the substrate gap.
• architecture/loomworks-architecture-specification-v0_4.md — §N-space Memory, §shared-subject pattern, §N-domain composition (carries the concept, marked "Investigating" / "filed but not yet exercised").

THE DECISION THIS CHAT MUST MAKE

Decision 1 — does the seed's Memory model widen to N-scope?

What the seed commits to today (v0.9, 2026-05-26): Memory is the engagement's accumulated knowledge — singular, engagement-scoped, monolithic. "Memory is the source. Nothing downstream exists without it." The seed says "domain" only as operator expertise and the engagement's subject — never as a Memory scope.

The trap to avoid: the architecture spec v0.4 (2026-05-25) does carry the N-space model — but the seed (2026-05-26) is one day newer and deliberately kept Memory engagement-scoped. So this is not the seed lagging the spec. The seed was written after the spec and drew the line on purpose. Per project rule — seed governs over spec — C5 cannot be scoped as code-only; it needs the seed to widen first. That widening is this Operator's call.

Decision 2 — the load-bearing framing question (must be answered, not deferred)

If Memory becomes N-scope, what does a Manifestation organize?

• The seed assumes one engagement's Memory ("a Manifestation organizes the engagement's Memory").
• C5 implies the composed N-scope view (renders compose across scopes).

This changes what Manifestation is. It is not a detail for the eventual CR — it belongs in the seed amendment itself.

What is NOT in tension (C5 preserves these — confirm, don't relitigate)

Memory-does-not-forget, trajectory/corrections-preserved, automatic provenance, Memory-grows — all hold across scopes. C5 is additive to them.

What is ALREADY settled (do not reopen)

Gate 2 — the "Organization" name collision — RESOLVED and filed (standing-notes/loomworks-standing-note-organization-name-collision-v0_1.md, committed 028f843):

The step-up arc shipped an Organization auth-policy object (engine fe8ee0b — presence window, always-tap, memberships). C5 wants "organization" as a Memory scope (brand/voice/commitments). Resolution: keep them separate; organization Memory is a row in the generic N-space scope table with scope_type = "organization" — no second Organization entity. Unification was considered and deliberately deferred to this seed session (not foreclosed): if this chat concludes an organization is genuinely one entity seen from two layers — auth floor + Memory scope — that is a clean forward decision to make here. But the default is separate.

Open-question set C5 carries (for awareness — not all for this chat)

These come due at C5 scoping time, after the seed decision. Listed so the seed decision can be made knowing what rests on it; do not try to resolve them in this chat.

• Composition (domain-composition §10.4): conflict resolution across orthogonal scopes; composition latency on hot-path checks; composition visibility to Operators ("show me the composition"); adversarial composition (can a malicious Operator bypass constraints?); N-domain composition with the standing Companion.
• Governance (memory-space §6/§9): which scope type ships first (org Memory likely forced before domain Memory by multi-Contributor engagements); who governs team/org/role/domain Memory; conflict rule (specificity-wins, with the flagged exception that a domain regulatory floor must not be overridable by engagement convenience); steward roles; when a domain-steward role enters vocabulary.

Dependency note (not a blocker for the seed decision)

C5's access-gating leans on OVA, which is still a permissive stub (_alpha_authorizer_stub). The docs already mark OVA as anticipated, so this is consistent, not stale — but C5's composition-gated-by-access cannot be real until OVA is. The seed decision and substrate design can proceed; the gating realizes later.

Orientation pair to load at the start of this chat

1. Seed v0.9 — candidate-seeds/loomworks/loomworks-candidate-seed-v0_9.md (the document being amended; read its Memory and Manifestation sections in full).
2. Current-status manifest — highest-numbered in current-status/ (confirm the version; do not assume).
3. This session's two artifacts: the collision standing note (028f843) and the CC alignment audit (the Part 1 / Part 2 read of C5 backing material vs seed v0.9 — it has the exact quotes and the four-point consistency surface). If the audit text isn't carried into the new chat, have CC re-run the Part 2 retrieval.

Everything in loomworks-record is behind Cloudflare Access — CC retrieves.

Expected output of this chat

Either:

• A seed amendment — seed v0.10 widening the Memory section to N-scope and answering the Manifestation-organizes question, with the prior engagement-scoped position named alongside the new one (Discovery-record discipline). Then C5 is scoping-ready and a separate chat does the scoping note → Step 0 → CR.
• Or a decision not to widen (C5 reshaped or deferred), recorded with reasoning.

Either way: no C5 build artifact in this chat. The seed first.

Record state at handoff (verify before treating as current)

Local loomworks-record main is 2 commits ahead of origin (523e72f), unpushed:

• be89193 — master build plan v0.5 (C2 Increment 2 settled, stays COULD)
• 028f843 — Organization collision standing note (gate 2)

Push is an explicit Operator authorization, not yet given. If this chat needs origin to be current, confirm push state with CC first (git -C /Users/dunin7/loomworks-record status) rather than assuming.