facets · docs
‹ Index
DocumentLIVE

Terminology Concordance — Klir · Bunge · Mobus

Status: LIVE. The reference grid, compiled 2026-07-18 from primary sources and primary-read notes; every cell carries a VERIFIED mark or an explicit attribution ruling (the §14 ledger closed same day), and no cell is filled from general knowledge. This is the SSOT the SL lexicon table (spec.md §3) summarizes; it also serves the per-lens UI copy (palette labels, describe() typesetting) and the K≅2 convergence program — a filled-in concordance is a compact convergence exhibit.

How to read it. For each kernel distinction, the row gives each tradition's term and its grounding. Convergence rows (all traditions share a word or an isomorphic concept) are evidence for the common core; divergence cells (a tradition lacks the distinction) are equally informative — they are exactly what the corresponding lens drops, and they explain the lens's rendering choices. VERIFIED = the term was found in a primary text on disk, or in a primary-read note that itself cites the book by page/line. UNGROUNDED = not found in an available source this pass; listed honestly in §14 rather than filled in.

Two verification targets, marked separately (council outside-pass, 2026-07-18): source-verified — the tradition's own text says this (the default meaning of VERIFIED above); repo-verified — the instrument implements this. They fail independently: a cell can be repo-verified yet source-UNGROUNDED (row 1's Bunge M was, until §14.1 closed) or source-verified yet unimplemented. Where the split matters, the row says which target a mark hits; a bare VERIFIED is source-verified.

Citation roots (sibling repos):

  • GSR/ = halcyonic-projects/active/general-systems-reasoner/
  • VAULT/ = the Obsidian vault, operations/systems-science/
  • SSF/ = halcyonic-projects/active/systems-science-foundations/
  • bare paths = this repo

The grid (summary; per-row notes below)

# Distinction Klir Bunge Mobus
1 The system S = (T, R) σ = ⟨C, E, S⟩ (1979); CESM µ(σ) (2004) 7-tuple ⟨C,N,G,B,T,H,Δt⟩; SOI S₀,₀
2 Part inside thing (element of T) component (C) component (C)
3 Thing outside environment (observer-side, not a coordinate) environment (E) environment = sources & sinks
4 Connection relation (R) connection / coupling / bond (▷) interaction / flow (N, G edges)
5 Bond vs mere — (one undifferentiated R) bondage B vs B̄ — (no such split)
6 Connection kind — (thinghood, excluded) flows of energy, matter, fields; informational substances: material, energy, message
7 Direction directed vs neutral systems (observer) asymmetry of action ▷ (agent/patient) inherent flow direction (k→o)
8 Boundary observer-drawn distinction topological boundary (1992), link-type-relative B = ⟨P, I⟩
9 Interface — interface points = i/o terminals ∈ boundary interfaces I (2nd coordinate of B)
10 Work-process primitives — — atomic process taxonomy (combine, split, buffer, …)
11 System types two-axis §2.4 (restrict T / restrict R) kingdoms + five genera (Post. 6.4) concrete/abstract + texture axes
12 Milieu vs objects — — (E is discrete things only) discrete Src/Snk + channels/fields

Bold marks the term SL adopted (or, in row 1, the formal object each lens typesets). Rows 2, 3, 6, 9, 11 are the strong convergence rows; rows 5, 7 (Klir), 10, 12 are single-tradition contributions — each is exactly one lens's distinctive vocabulary.

1. The system itself

  • Klir — S = (T, R). VERIFIED. "S = (T, R) — where T is 'a set of things distinguished within S' and R 'a relation … defined on T'" (GSR/klir_markdowns/Klir-Systemhood-vs-Thinghood-(2001).md:19).
  • Bunge — σ = ⟨C, E, S⟩. VERIFIED. "the ordered triple σ = ⟨C, E, S⟩ is (or represents) a system over T iff C and E are mutually disjoint subsets of T … and S is a nonempty set of relations on the union of C and E" (SSF/docs/reference/mobus-bunge-system-definitions-reference.md:185; minimal model at GSR/bunge_markdowns/Bunge-Concrete-System-Definition-CES-(1979).md:77-79). The fourth CESM coordinate M (Mechanism) is now VERIFIED verbatim (2004, not the 1979 CES paper): "The simplest sketch or model of a concrete system σ is the list of its composition, environment, structure, and mechanism, or µ(σ) = ⟨C(σ), E(σ), S(σ), M(σ)⟩ … M(σ) stands for the mechanisms, or characteristic processes of σ" (GSR/bunge_markdowns/Bunge-How-Does-It-Work-(2004).md:289-294); "M(σ) is empty for conceptual systems" (:300). So CES (1979) and CESM (2004) are both Bunge's, a quarter-century apart — the kernel's CES-with-mechanism-note is faithful to the 1979 formal object while acknowledging the 2004 extension.
  • Mobus — the 7-tuple, SOI. VERIFIED, with the provenance nuance this repo already carries: the book prints a 7-tuple ⟨C,N,G,B,T,H,Δt⟩ (VAULT/mobus/4-a-model-of-system.md:196-199); the 8-tuple with E first-class is the Lean formalization's deliberate improvement (SSF/Systems/Mobus/Tuple.lean), credited as such in anything published. "S₀,₀ is the designated SOI" (SSF/docs/reference/mobus-bunge-system-definitions-reference.md:133).

Convergence reading: three formal objects, one referent — this is the K≅2 claim itself, and the three describe() typesettings render exactly these three signatures.

The named referent (#84, shipped 2026-07-18): SL now names the SOI — system "Name" [: Kingdom/Genus] → CanvasModel.name → the root system's info.name. The name is the one place all three traditions already agreed without a shared word: Mobus's designated SOI S₀,₀, Bunge's σ, Klir's S each denote this system, and the paragraph's proper name ("Process M", "Steel-Plant") is what the denotation binds to.

2. A thing inside the system — the strongest shared word

  • Klir — "thing." VERIFIED. T = "a set of things distinguished within S"; thinghood resides in T (Klir-Systemhood-vs-Thinghood-(2001).md:19).
  • Bunge — "component," composition C. VERIFIED. "the A-composition of σ at a given time t is the set of its A-parts at t" (Bunge-Concrete-System-Definition-CES-(1979).md:55-56).
  • Mobus — "component," C. VERIFIED. "C is a set of components and their 'type' … may have partial inclusion" (mobus-bunge-system-definitions-reference.md:137).

Convergence reading: Bunge and Mobus share the word and the letter — SL's component is a two-tradition word. Klir's divergence ("thing") is principled, not lexical accident: his T carries thinghood, which he excludes from systemhood — which is why the Klir lens relabels what the other two share.

3. A thing outside the system

  • Klir — "environment," observer-side. VERIFIED, not a coordinate. "an agent that is not part of the system … referred to as an environment of the system; it includes in many cases the investigator" (VAULT/klir/klir-facets.md:3648-3656). The bare S=(T,R) has no environment slot; environment enters through input variables.
  • Bunge — "environment," E. VERIFIED. E_A = the things outside C that act on or are acted on by a component (Bunge-Concrete-System-Definition-CES-(1979).md:58-59).
  • Mobus — "environment" = sources and sinks. VERIFIED. "The environment contains all of the sources of inputs … and the sinks for outputs … all of the channels (or fields) through which flows occur" (mobus-bunge-system-definitions-reference.md:40).

Convergence reading: environment is a Bunge–Mobus shared word (with Bunge's E the formal coordinate SL's Role::Environment mirrors); Mobus contributes the source/sink refinement SL uses as author-intent words, with identity edge-derived (spec.md §5.2).

4. A connection between parts

  • Klir — "relation," the seat of systemhood. VERIFIED. Binary R ⊆ T×T (GSR/klir_markdowns/Klir-Relations-(2001).md:5-7); Klir explicitly subsumes "interaction, interconnection, coupling, linkage, cohesion, constraint, interdependence, function, organization, structure…" under mathematical relation (:63).
  • Bunge — "connected (or coupled, or linked, or bonded)." VERIFIED. "two things are connected … if at least one of them acts on the other"; action a▷b "modifies the latter's behavior line, or trajectory, or history" (Bunge-Concrete-System-Definition-CES-(1979).md:27-29).
  • Mobus — "interactions"/flows. VERIFIED. "N is generally a flow network through which real substances are moving from one node (component) to the next with causal influence" (mobus-bunge-system-definitions-reference.md:157-161); boundary flows form the bipartite graph G (:167).

Convergence reading: one concept, three vocabularies at three levels of commitment — Klir's bare relation, Bunge's action-backed connection, Mobus's substance-carrying flow. SL's flow keyword is Mobus's word; the kernel's Relation is Klir's object; the bond semantics are Bunge's. One line of SL touches all three, which is the unified-language thesis in miniature.

5. Bond vs mere relation — Bunge's alone

  • Bunge — VERIFIED, verbatim. "the total set of relations among the components … may be decomposed into its bondage B_A and the set B̄_A of nonbonding relations"; "Unlike a mere relation, a connection makes some difference to its relata"; systemhood requires B_A ≠ ∅ (Bunge-Concrete-System-Definition-CES-(1979).md:23, 33, 62-69).
  • Klir — absent, VERIFIED. All of systemhood collapses into one undifferentiated R (Klir-Relations-(2001).md:63); the Klir lens drops bond-vs-mere (docs/theory-fidelity.md:20).
  • Mobus — absent, VERIFIED by scoped sweep (§14.5). Chapters 3–4 swept for bond/bonding, coupling, connection/relation type, mere aggregate, glue, "makes a difference": every bond hit is physical/chemical/social bonding in the ontogenesis narrative, never a structural typology partitioning relations. His nearest device — boundaries implied where interrelations are "much stronger" among internal components (VAULT/mobus/3-system-ontology.md:1182) — is an interaction-strength/endo-exo criterion, not Bunge's difference-making partition.

SL's mere is therefore a single-tradition contribution, and the aggregate/system verdict it feeds (Bunge Def 1.1 in the UI) is the most lens-specific rule in the tool.

6. The kind of a connection — SL's kind words are Bunge's, verbatim

  • Bunge — VERIFIED, verbatim. "Dynamic connections are often called flows — of energy, matter, or fields. If a physical flow happens to carry information, the connection is called informational … the physical/informational distinction is one of emphasis, not a dichotomy, for every information flow rides on some energy flow" (Bunge-Concrete-System-Definition-CES-(1979).md:92). SL's energy | matter | field | informational is this sentence as an enum (crates/bert-canvas/src/canvas.rs:52-62).
  • Mobus — material / energy / message. VERIFIED. "All processes transform low-quality material, energy, or messages into high-quality versions of the same" (dynamics-research/read-mobus.md:120-124); messages copy rather than conserve (read-mobus.md:305-324). The 1:1 mapping Kind→Substance is code (canvas.rs:64-70): Matter→Material, Informational→Message, Energy/Field→Energy.
  • Klir — absent, VERIFIED. Kind is a thinghood property, excluded from R (docs/theory-fidelity.md:20).

Convergence reading: a genuine two-tradition near-isomorphism (4 Bunge kinds ↔ 3 Mobus substances, collapse documented in code), with Klir's principled abstention explaining why the Klir lens shows untyped edges.

7. Direction

  • Klir — "directed" vs "neutral" systems. VERIFIED, verbatim. "Systems whose variables are classified into input and output variables are called directed systems; those for which no such classification is given are called neutral systems" (VAULT/klir/klir-facets.md:3657-3659). An observer's commitment, added to the source system — exactly why SL carries it as the @directed annotation (view state), not a structure word.
  • Bunge — asymmetry of ▷. VERIFIED. Agent/patient: "If a thing acts upon another, and the latter does not react back, the former is called the agent and the latter the patient" (Bunge-Concrete-System-Definition-CES-(1979).md:29).
  • Mobus — inherent flow direction. VERIFIED. "the direction is assumed from k to o" (mobus-bunge-system-definitions-reference.md:159).

Divergence reading — the most instructive row: the same arrow is observer-optional for Klir, action-asymmetry for Bunge, and substance-transport for Mobus. SL's -> (structural, always present) plus @directed (view state, Klir's toggle) encodes the difference precisely.

8. The boundary — Bunge formalized it too (correction to earlier framing)

  • Mobus — B = ⟨P, I⟩. VERIFIED. "B_{i,l}=⟨P_{i,l},I_{i,l}⟩ where P is the set of properties and … I_{i,l} is the set of interfaces. The exact form of P is still an object of research" (mobus-bunge-system-definitions-reference.md:175-177). The property words are now VERIFIED verbatim: "as porosity (0 being completely non-porous) and 'perceptive fuzziness,' meaning the degree to which it is easily perceived" (VAULT/mobus/4-a-model-of-system.md:376).
  • Bunge — a topological boundary, VERIFIED verbatim (1992). "A boundary component of a system is … one every neighborhood of which contains at least one system component and at least one thing in the environment"; boundaries are link-type-relative ("different link types induce different boundaries") and universal ("every concrete system, except for the universe as a whole, has at least one boundary") (GSR/bunge_markdowns/Bunge-System-Boundary-(1992).md:47-51). Earlier repo framing ("Bunge's boundary is informal") is corrected by this source.
  • Klir — an observer-drawn distinction. VERIFIED. "A distinction splits the world into two parts … 'environment' and … system of interest" (VAULT/klir/klir-facets.md:1442-1446, 9851).

Convergence reading: stronger than previously credited — Bunge's boundary-as-set-of-boundary-components independently anticipates the boundary-identity result the lenses already render (boundary(Bunge) = interfaces(Mobus) as a component set), and his link-type relativity resonates with kind-typed membranes. A genuine two-tradition convergence with Klir's observer reading as the Core-level floor.

9. Interface

  • Mobus — "interfaces," I. VERIFIED as the second coordinate of B (mobus-bunge-system-definitions-reference.md:177). Ruling on I ⊆ C (§14.3, resolved 2026-07-18): the book's claim is that interfaces are components of the boundary subsystem and systems in their own right — "the set of interfaces embedded in the boundary are components in the boundary subsystem and are themselves subsystems. That is, every r_{i,l} ∈ I_{i,l} is an S itself" (VAULT/mobus/4-a-model-of-system.md:386; also :334). The flat subset I ⊆ C over the system's own component set is the Lean formalization's choice (SSF/Systems/Mobus/Tuple.lean, iface_sub) — a faithful simplification (the boundary subsystem's components are components), credited as the Lean's, same pattern as the 8th tuple slot. Cite the book for interfaces-as-(boundary-)components, the Lean for the flat subset constraint.
  • Bunge — "interface points." VERIFIED. "it has some boundary or interface points … the input and output terminals of a system are members of its boundary" (Bunge-System-Boundary-(1992).md:168, 236-237).
  • Klir — absent (§14).

Convergence reading: interface turns out to be a shared Bunge–Mobus word too — Bunge's i/o terminals as boundary members is structurally the same claim as Mobus's I inside B, and both agree with the Lean's identification of interfaces with (boundary) components.

10. Work-process primitives — Mobus's alone

  • Mobus — VERIFIED. The atomic work-process taxonomy and the simplest-process stopping rule (dynamics-research/read-mobus.md:129-135, citing VAULT/mobus/4-a-model-of-system.md:250-251 and 3-system-ontology.md:1195); general process definition: "A process is a system that performs a transformation on its inputs to produce outputs that are different in form, quantity, or organization" (read-mobus.md:76-77). Shipped as ProcessPrimitive (crates/bert-core/src/lib.rs:1263).
  • Bunge — absent (process = trajectory in state space, no primitive taxonomy; dynamics-research/read-bunge.md:29-30). Klir — absent (behavior functions, not functional primitives; read-klir.md:158-172).

11. System types

  • Bunge — kingdoms + genera, VERIFIED verbatim. "These are the only system kingdoms we recognize: conceptual and concrete"; "there are five system genera: physical, chemical, biological, social, and technical" (Postulate 6.4; SSF/docs/reference/system-type-typologies.md:15, 23).
  • Klir — the §2.4 two-axis scheme. VERIFIED. Classify "by restricting T to certain kinds of things; by restricting R to certain kinds of relations"; the type-(a) list "physical, chemical, biological, economic, social, etc." (system-type-typologies.md:66-71). The Bunge-genera ≅ Klir-type-(a) convergence table already exists at system-type-typologies.md:84-92 — this row defers to it.
  • Mobus — concrete/abstract + texture axes. VERIFIED. (system-type-typologies.md:45, 53-56).

12. Milieu vs discrete objects — Mobus's alone

  • Mobus — VERIFIED for the distinction; letters RULED Lean-side (§14.4, resolved 2026-07-18). Environment = discrete sources/sinks plus "all of the channels (or fields) through which flows occur" (mobus-bunge-system-definitions-reference.md:40). The letter-names O (objects) and M (milieu) do not appear in the book as environment letters — chapters 3–4 denote the environment G = ⟨Src, Snk⟩ throughout (VAULT/mobus/4-a-model-of-system.md:285, 301, 304; the E/M letters near Fig. 4.7 label internal processes, not environment sets). The O/M lettering is the Lean layer's convention (SSF/Systems/Mobus/Interface.lean:7), credited as such.
  • Bunge — no such split, VERIFIED. E is discrete acting things only; "field" appears as a flow substance, not an environment partition (Bunge-Concrete-System-Definition-CES-(1979).md:58-59, 92). Klir — absent (klir-facets.md:4153).

13. Existing partial mappings (what this doc federates, not duplicates)

  • SSF/docs/reference/system-type-typologies.md §4 — the row-11 Bunge≅Klir convergence table (two traditions, one row). Deferred to above.
  • SSF/docs/reference/mobus-bunge-system-definitions-reference.md §8 — Mobus-tuple ↔ Bunge-CES definitional side-by-side (two traditions).
  • docs/theory-fidelity.md — per-tradition keep/drop and the three canonical signatures (framing, not a term grid).
  • SSF/docs/reference/mobus-bunge-mathematical-reformulation-guide.md — Mobus→Bunge reformulation (not incorporated this pass).

This document is the first full three-way, twelve-row grid; those four remain the deeper dives per pair.

14. Verification ledger (all cells resolved 2026-07-18)

The five UNGROUNDED cells of the first pass, each now closed as VERIFIED-with-citation or as an explicit attribution ruling. The distinction between the two outcomes is deliberate: a ruling is not a failure — it records that the constraint is the formalization's contribution, credited as such.

  1. Bunge's fourth CESM coordinate M (Mechanism) — VERIFIED verbatim. µ(σ) = ⟨C(σ), E(σ), S(σ), M(σ)⟩ with M the "mechanisms, or characteristic processes" (GSR/bunge_markdowns/Bunge-How-Does-It-Work-(2004).md:289-294, 311-316). Source is the 2004 paper, not the 1979 CES definition — row 1 now carries the two-paper provenance.
  2. Mobus porosity / perceptive fuzziness — VERIFIED verbatim (VAULT/mobus/4-a-model-of-system.md:376; further porosity usages :664, :677, :736, :738). Row 8 updated.
  3. Mobus I ⊆ C — RULED. Book states interfaces are components of the boundary subsystem and systems in their own right (4-a-model-of-system.md:386); the flat subset over the system's own C is the Lean's formalization choice (Tuple.lean, iface_sub), credited as such. Row 9 updated; spec.md §3 wording aligned.
  4. Mobus environment letters O / M — RULED. Not the book's (environment is G = ⟨Src, Snk⟩; 4-a-model-of-system.md:285-304); Lean-side convention (Interface.lean:7). Row 12 updated.
  5. Mobus bond-vs-nonbond counterpart — absence VERIFIED, scoped. Sweep of VAULT/mobus/3-system-ontology.md + 4-a-model-of-system.md for: bond, bonding, coupling, connection type, relation type, structural relation, mere aggregate, aggregate vs system, glue, non-bond, "makes a difference," "hold together," internal/external relation. No structural bonding/nonbonding partition exists; nearest device is the interaction-strength boundary criterion (3-system-ontology.md:1182), a location criterion rather than a difference-making one. Row 5 updated.

Bibliographic independence (the empty citation edge). Mobus does not cite Bunge: a case-insensitive sweep for "Bunge" across all 16 chapter bodies of Systems Science: Theory, Analysis, Modeling, and Design in VAULT/mobus/ returns zero hits, against a text that cites heavily inline (Checkland, Shannon & Weaver, Smith & Morowitz, Mobus 2019 ×10). Caveat, stated honestly: the end-of-book reference list is not on disk (the vault holds chapter bodies only; the full PDF is not in Zotero), so the finding covers the chapter bodies, not the printed bibliography. Why it matters: rows 2, 8, and 9 show Bunge and Mobus converging on component, a formalized boundary, and interface — and the empty citation edge is what makes that convergence evidential rather than genealogical. Independent arrivals at the same distinctions are K≅2 evidence; borrowed vocabulary would not be.

15. Consumers

  • SL lexicon (spec.md §3) — summarizes the "Contributed by" column from this grid.
  • Per-lens UI copy — palette labels and describe() typesetting should match column vocabulary per lens (Klir says thing/relation, Bunge says component/bond, Mobus says component/flow): this grid is the check. The audit against the web app is §15.1.
  • K≅2 convergence program — rows 2, 3, 6, 8, 9, 11 are citable convergence exhibits; rows 5, 7, 10, 12 are the honest divergence ledger that keeps the convergence claim falsifiable.

15.1 Per-lens UI-copy audit (2026-07-20)

The web app's user-facing lens strings, audited against the settled terms above, per lens. Where a string is authored (palette, node inspector, edge-editor titles) a mismatch would be fixed here; none were — every authored label already speaks its lens's column. Where a mismatch lived in a code string (describe() typesetting or the shared edge-editor enum), it is recorded below; all three were fixed by #80's build, which consumed this stabilized section rather than the doc re-deriving it.

Matched — authored labels are settled-term-faithful.

Surface (web/src/…) Klir Bunge Mobus
palette place / connect (canvas/lenses/registry.ts) thing · relation component · environment thing · bond ⁄ mere relation component · environment object · typed flow · interface · primitives
node inspector (canvas/NodePopover.tsx:56) thing component · environment thing component · environment object
edge-editor title (canvas/EdgePopover.tsx) relation rₙ ⊆ T×T · neutral / directed bond ⁄ mere relation · connection kind flow · substance
formal face (FormalPanel.tsx) |T| things (thinghood) · |R| relations (systemhood) 𝒞 · ℰ · bonds (endo/exo) + mere · boundary (1992) C · N · E=⟨O,M⟩ · G · B=⟨P,I⟩ · porosity / fuzziness

Each cell traces to a row: thing/relation (rows 2, 4) and Klir neutral/directed (row 7); component/bond, environment thing, and boundary-1992 (rows 2, 3, 5, 8); component/flow, environment object, B=⟨P,I⟩, interfaces, and primitives (rows 2, 3, 9, 10). The Bunge/Mobus environment split — Bunge says "thing," Mobus says "object" — is itself row-12-faithful (milieu-vs-object is Mobus's alone).

Mismatched — code strings, since fixed. The authored labels above were correct; these three lived in describe() typesetting and the shared edge enum, so they belonged to #80's pair-names-with-math build rather than a doc fix. All three shipped 2026-07-20 with #80 and are kept here as the record of the audit, not as open items — the line numbers below are the pre-fix ones:

  1. The Mobus "substance" picker offers Bunge's kind enum, not Mobus's substances. The Mobus edge editor's substance select (web/src/canvas/EdgePopover.tsx:13,277) lists the raw Kind enum {unspecified, energy, matter, field, informational}. Per row 6, the Mobus lens's substance words are material · energy · message, and the collapse already exists in Rust (kind_to_substance, crates/bert-canvas/src/canvas.rs:64-68: Matter→Material, Informational→Message, Energy/Field→Energy). The Mobus picker should present the substance names through that map. The same enum under the Bunge lens (connection kind, EdgePopover.tsx:209) is correct as-is — row 6 has energy/matter/field/informational as Bunge verbatim.
  2. The Bunge formal object labels the four-coordinate object with the symbol σ. FormalPanel.tsx:73 and kernel/context.ts:48 print the Bunge object as a four-coordinate σ (composition, environment, structure, mechanism). Per row 1 / §14.1, σ is the 1979 CES triple; the four-coordinate object is µ(σ) (Bunge 2004, the mechanism_note provenance). The formal face should carry µ(σ) for the four-tuple, or keep σ to the triple and hold M as the mechanism note.
  3. The Mobus formal object prints the eight-tuple without crediting E as the Lean's improvement. The Mobus formal face (FormalPanel.tsx:106, kernel/context.ts:49) prints ⟨C, N, E, G, B, T, H, Δt⟩ as "S". Per row 1, the book prints the seven-tuple; E-as-first-class is the Lean formalization's deliberate improvement (SSF/Systems/Mobus/Tuple.lean), to be credited as such on any published surface. The formal face is a published surface, so it should carry that credit (a source-note or the "Lean improvement" label), not present the eight-tuple as the book's own.

A fourth, lower-priority note: kernel/context.ts (renderElements) labels every element [thing:id] / [relation:id] regardless of lens. This is a lens-invariant serialization prefix for the LLM context, not lens chrome, and was flagged so #80 could rule on it explicitly. It did: ruled LLM-context-only and left unchanged.