Status: Position, hero-emphasis, glyph (A), font swap (all-Plex), and
headline/tagline all Decided (2026-06-15, owner-confirmed this session).
Remaining open: render glyph A into site-kit; develop the L2 brand layer
(see Position → Horizon). · Owner: PM/Brand · Companions:
doctrine.md (position source), THEMING.md
(sharing mechanism), packages/site-kit (implementation),
web/information-architecture.md (site copy).
The bridge between building the brand and managing it. Read the Invariants before approving any new surface; a live choice that contradicts one with no overriding entry in Decisions is drift.
- For: developers and the agent fleets they run, working across many small, separately-installed tools.
- Job: turn a swarm of separate, stateless tools and agents into one coherent, identity-stable body of work — where the joins between tools are the value.
- Feeling (the ONE thing): trustworthy coherence — "these separate things are actually one fabric, and it will not lie to me about where the seams are."
- Rejects: the dark-mode-gloss, all-green-dashboard, "trust us, it passed" SaaS dev-tool aesthetic; the monolith-with-plugins; ad-hoc bilateral integration; hype adjectives. Not for anyone who wants a single big product that hides its seams.
The hero is the weave, not the tools (owner-confirmed 2026-06-15, tracing to PDR-0023 "the federation is the product; the seams are the crown jewels"). The tools are the threads; the cloth is the product. This supersedes the founding 2026-06-05 framing ("five sharp tools behind one brand"), which the original brand was authored against — see Decision D1.
The name already encodes the position: in weaving the warp is the fixed threads held under tension (each tool, standing alone); the weft is the single thread drawn across them that binds them into fabric (the seam / the SEI spine / the glue). Weft is the binding cross-thread. The name is settled and load-bearing.
The owner's framing: L1 is the toolset (the members — the code-and-debug tools). The realization that became L2 is that agents were never excited by the tools themselves — they were excited by the identity (SEI) pulling every tool's output into one common operating picture. Weft is the suite's shot at the L2 environment: a surface where an agent fleet builds its own scripted dashboards and reporting on the SEI spine + the toolset.
This confirms the hero-flip past doubt (the weave/identity, not the tools, is what has pull) and adds one beat the suite-level brand above does not yet carry: generativity / agency. L2 is not "we wove the cloth for you" — it is "here is the loom; you weave your own view." The metaphor extends cleanly and should be made explicit when the L2 layer is developed:
| Loom term | Weft meaning | Layer |
|---|---|---|
| warp (threads under tension) | the tools, each standing alone | L1 |
| weft (the binding cross-thread) | the SEI identity-join / the seam | the spine |
| loom (the apparatus you weave on) | the L2 environment the fleet builds on | L2 |
| the cloth | the common operating picture woven from SEI-keyed output | the product |
"Common operating picture" is a candidate hero phrase for L2 specifically. The L2 brand layer (its own headline/tagline tier and whether the loom surfaces as a named brand element) is not yet derived — see the open question logged for the next session. The customer of L2 is the fleet (PDR-0024).
-
Personality / values: honest, grounded, engineered-not-glossy, warm, precise. Shows what is absent, not just what is green — and discloses the edges of what it knows (what was excluded, withheld, unresolved). No hype, no emoji (a rule, not a default).
-
Anchor color + its role: indigo
#35487Bis "this is us" — it is Loomweave's thread (identity/the spine) and the federation primary, because the spine is the hero. Warm linen#F5F0E5+ ink is the ground. The brand is one warm light theme — "natural dyes on unbleached cloth" — light-only, no dark mode, no toggle. This is the single most distinctive, load-bearing decision in the brand (deliberate anti-convergence; see Decision D5). -
Per-member thread color: each member owns exactly one dye-pigment thread (indigo / brass / madder / walnut / woad / copper / ash for Lacuna). The federation is the palette. Threads are canon in site-kit
tokens/colors.css. -
Logo DNA: a weave-mark — light warp threads (the tools) crossed by one bold indigo weft thread (the seam, the hero), with the over-under occlusion visible (the join is inspectable — the honesty mechanic, drawn). See Decision D2.
-
Mono is first-class. CLI output, JSON envelopes, SEI tokens and tool names render in mono; the brand speaks the medium its users live in.
-
The promise: every result carries its
cause + reason_class + fix; the brand never shows you an all-green dashboard when it isn't — and it tells you the edges of what it can see. (The honesty invariant — doctrine.md §10, theweft-reasoncontract.)Why it is a hard gate, not a nicety (the keystone): in one fabric, every tool speaks on behalf of every other — so a single confident-wrong answer undermines the whole weave at once. Trust is pooled; one bad thread is visible across the entire cloth. That is why honesty is a federation-level invariant and a hard admission gate (not a per-tool virtue), and why the L2 platform requires honesty to propagate through composition. (The 2026-06-15 Lacuna cold-eval — a confident-empty answer from one tool — is what prompted the lift to L2 and its honesty rules.)
Two field-validated sub-principles (2026-06-15 Lacuna cold-eval):
- The caveat is the headline, not a footnote. A true-but-unqualified number ("141 dead") still misleads; the limit leads. This is why the lead-summary and bounded-by-default conventions are brand-load-bearing, not just ergonomic.
- An honest empty is loud, not clean. A "nothing found" that could be a real nothing or a failed lookup must say which — never a clean-looking empty set a hurried reader mistakes for "all clear." Silent redaction/nulling is an honesty failure even when the intent (don't leak) is right.
- Accent = the member's thread color. A member site/CLI leans on its own pigment; the hub stays indigo. State is shown by a 4px left rule (thread or semantic color), not fills.
- Density / motion: dense, mono-forward, flat. No shadows-as-depth, no
gradients; depth is the warm surface ramp. Radii
3 / 6 / 8. - Voice calibration: plain, declarative, lowercase-mono for any technical token; honesty stated as a limit, not a boast. On-voice exemplars already in the wild — Wardline: "silence is not a clean bill." Legis: "a discipline, not a hardened security boundary." Federation-level lines: "No all-green dashboards. The seams are inspectable." · "An honest empty is loud, not clean." Field-validated — a 2026-06-15 cold-eval evaluator, unprompted: "the honesty about what they can't see is the reason I trust what they do return."
Each names what was chosen, why (from position), and the reversal trigger.
-
D1 — Hero emphasis: the weave, not the tools. (Decided 2026-06-15.) The brand leads with the seam/identity/glue; tools are proof-points. Supersedes the 2026-06-05 tools-led framing. Reverse if the federation ever ships as a single installable product (the "one big product" the position rejects) — at which point the tools-as-components framing would be a lie.
-
D2 — Logo: "the weft pass" (direction A). (Decided 2026-06-15.) The shipped
weft-glyph.svgis a uniform grid (fine under the old tools-led reading; it has no hero thread). Replaced by direction A — one bold indigo weft woven through light warp threads, occlusion visible — in web/brand-proposals/glyph-A-weft-pass.svg. C "crossing" is retained as the compact/favicon lockup. Reverse if A reads as a dash/plus below ~24px in real use — promote C to the small-size mark (already the plan). The livesite-kitweft-glyph.svgis not yet re-rendered (pending this session's render step). -
D3 — Tagline + headline: weave-led. (Decided 2026-06-15.) Retired "Small, sharp tools — woven on one identity" (tools-led). Suite/hub headline: "The tools are the threads. The weave is the product." One-line tagline: "One identity, woven through every tool." Signature honesty line: "No all-green dashboards. The seams are inspectable." Reverse if D1 is reversed. (An L2-specific headline tier is still to be derived — Horizon.)
-
D4 — Typography: drop Space Grotesk; the working fonts are the brand fonts. (Decided 2026-06-15 — owner delegated the type call.) Keep IBM Plex Mono (data/CLI/SEI — load-bearing, on-position) and IBM Plex Sans (text/UI — honest humanist engineering, same superfamily = coherence for free). Retire Space Grotesk for display: it is a heavily-used startup-geometric face (mild category convergence) and its cold geometry fights the warm natural-dye palette. Set headlines in Plex Sans at display weights. Distinctiveness is deliberately carried by color + the weave-mark + the honesty-state UI, not by a display face; type buys trust/legibility/coherence. Reverse if all-Plex headlines test as too utilitarian for brand moments — the warm-distinctive alternate is Fraunces (optical-size soft serif, matches the craft/dye story) used only at display. Wordmark: Plex Sans medium lowercase now; eventual custom-drawn
weftwhoset-crossbar is the weft thread (ties wordmark to glyph). -
D5 — Keep the color anchor (the user is happy with it, and it is correct). Warm linen + ink, light-only, one pigment per member. It already does the position's heaviest lifting (anti-convergence, federation-as-palette, honesty-as-flatness). Do not flex. Reverse only if the suite adopts a surface where a warm light theme is unusable (e.g. a required dark embed) — and then as a scoped local inversion, never a second page theme.
Honesty constraint on this whole section. PDR-0024 is CANDIDATE, gated to a falsifier-test / dogfood proof before promotion; public L2 positioning is owner-reserved (PDR-0023 §5). A brand built on the honesty invariant must be honest about its own maturity — so this copy stays horizon/forward-framed and is not shipped present-tense until PDR-0024 is Adopted. These are derived candidates, not locked canon.
Position (L2). For: an agent fleet working the same repos concurrently — not one agent with a toolbox (the unit the L1 brand still implies). Job: give the fleet (a) one honest common operating picture of the joined flood and (b) a deconfliction surface so agents don't collide — both built-on-able. The ONE thing: "we are one crew operating from one honest picture — and I can build the view I need on it." Rejects: per-agent silos; synthesis that's greppable-away (adds nothing a competent agent couldn't derive); confident-wrong synthesis fed by a lying flood.
The metaphor extends — two planes at the loom (the loom = the L2 environment, per the Horizon map):
| Loom act | L2 plane | Weft meaning |
|---|---|---|
| reading the cloth | sense-making | the common operating picture — the "so what" query over the honest flood |
| no two shuttles in one shed | coordination | deconfliction — the tag-out board; "two agents never edit the same tree blind" |
| the fleet weaves its own views | generativity | agents author their own macros + checks over blessed primitives |
Proposed copy (candidates, not locked):
- L2 hero noun —
common operating picture. Recommend adopt. A borrowed C2 term for the single shared, trusted, authoritative picture a crew operates from. It is the natural noun for the suite's deepest framing — Weft is deconfliction-first, not security, and a common operating picture is the deconfliction artifact. The phrase is canon-aligned, not bolted on. - Platform / generativity line — "The hub owns the alphabet. Your agents write the words." Recommend adopt (it is already the PDR-0024 platform invariant, verbatim). Captures platform + governance guardrail + the build-your-own promise in one line.
- L2 honesty differentiator (the answer to "why not just grep") —
"Honesty propagates through composition: a macro is only as honest as its
weakest call — and the reason rides all the way up." This is the moat against
the greppable-away falsifier: custom synthesis you can trust because the
weft-reasoncarrier propagates. - Coordination line — "Two agents never edit the same tree blind." (Deconfliction in its purest form — PDR-0024.)
- L2 headline (candidate) — "One honest operating picture for the whole fleet."
L2-OPEN-1 (owner ruling) — is "the loom" a named brand element?
Recommendation: no. Keep the loom as the lowercase explanatory metaphor
and Weft as the only brand name; lock common operating picture as the
L2 hero noun instead. The suite already carries many proper nouns (six members +
Lacuna + the SEI + the seams); minting "the Loom" risks "is it a member?"
confusion and over-claims a CANDIDATE bet with a separate product name. Reverse
if L2 ever ships as a distinct hosted surface that genuinely needs its own
handle. (Logged for owner ruling — this is an invariant/lineage decision.)
L2-OPEN-2 (gates promotion of this entire layer) — "platform" collides with the federation axiom. doctrine.md §1 (current canon, updated 2026-06-15) states Weft is "not a platform, not a shared runtime... nothing called 'Weft' to install, deploy, or keep running." The L2 frame is platform-shaped: the loom you build on, a hosted tag-out board (in filigree), an L2 synthesis MCP. So the brand must not call Weft a "platform" or "environment" present-tense until this is reconciled — and naming the collision rather than papering over it is the honest move (it is the deepest reason PDR-0024 is CANDIDATE). The reconciliation to watch: L2-as-convention ("hub owns the alphabet, agents write the words" = blessed primitives + agent composition, documentation-shaped → axiom-compatible) vs. L2-as-runtime (a board/MCP that must be running → strains "nothing to keep running"). Which one L2 becomes is an owner/strategy call, not a brand call — but the brand layer above cannot promote past Candidate until it lands.