DUNIN7 · Loomworks · Scoping Note · Plain-English Version
Current-engagement spine
Giving the Companion a memory of "where you are."
Version. v0.2 (plain-English companion) Date. 2026-06-26 For. The Operator. The two choices below are now made; this version records them. Reading level. Plain English. A technical twin of this note holds the exact field names, file locations, and line numbers. What changed from v0.1. You chose 1B (all three ways of saying "open X" work, including "open 7") and the defaults for what happens when your spot is unclear. Both are now recorded as decided.
Technical twin: loomworks-current-engagement-spine-scoping-note-v0_2.html / .md — same decisions, written for the build session.
The whole thing in a paragraph
Right now, every time you talk to the Companion, it has to be told which engagement you mean — every single time. It has no memory of "the one I'm working in right now." This change gives it that memory. When you say "open Goosey," the Companion remembers you're in Goosey. Then you can just say "commit held 14" and it knows you mean Goosey's held item, without you naming Goosey again.
Why this one first: it's the foundation the whole voice idea sits on. Talking out loud only feels natural if you don't have to re-state where you are with every sentence. This gives the Companion that sense of place.
What problem this fixes
Think of the Companion today like someone with no short-term memory of the room they're standing in. Every sentence you say, you have to start with "in Goosey…" or it's lost.
Today (annoying)
You: "In Goosey, commit held item 14."
You: "In Goosey, show me what's in Shaping."
You: "In Goosey, what's left to do?" You repeat "in Goosey" every time, or nothing works.
After this change (natural)
You: "Open Goosey."
You: "Commit held item 14." (it knows: Goosey)
You: "Show me what's in Shaping." (still Goosey)
You: "What's left to do?" (still Goosey)
The fix is small: give each person a single note that says "the engagement you're currently in." The Companion writes that note when you open something, and reads it whenever you give an instruction without naming where.
How small the change really is
The good news from checking the live system: the place to put this already half-exists. Each person already has a small record that holds things like the Companion's name and which engagement is their Personal one. We're adding one more line to that same record — "current engagement" — right next to a line that already works exactly the same way ("Personal engagement"). It's a copy of something proven, not a new invention.
Picture it
Your personal record already says:
• Companion's name: "…"
• Your Personal engagement: "…"
We're adding one line:
• Where you are right now: "…"
And the spot where the Companion would read that line is already open on every conversation — it's reading the two lines above it already. So reading the third costs nothing extra.
What we build, in plain terms
One new line on each person's record: "where you are right now." Empty to start; points at an engagement once you open one.
The Companion fills it in when you say "open X."
The Companion reads it when you give an instruction without naming an engagement, so "commit held 14" lands in the right place.
You can ask "where am I?" and it tells you the engagement you're in — by name, by your own number for it, and by its E-label.
What we are deliberately NOT building
Worth being clear, because it keeps this small:
No "multiple places at once." One current spot per person. If you later need to juggle several open contexts at the same time, that's a bigger thing for another day.
No "the second one" yet. Being able to say "open the second one" after a list — that's a separate later step. It builds on top of this one, which is part of why this goes first.
Nothing breaks for existing callers. Anything that already names its engagement keeps working untouched. The new memory only kicks in when you don't name one.
Two things — now decided
These were the two open choices. You've made both. They're recorded here as decided.
Decision 1 — Which ways of saying "open X" should work? → 1B
There are three ways you might name an engagement out loud:
By name — "open Goosey." Works today.
By its E-label — "open E0007." Works today.
By your own number for it — "open 7." Does NOT work today. Right now that number only expands a row in a list; it can't actually take you there.
So the choice is: ship with the two that already work, or also build the third now?
Decided — "1B"All three work, including "open 7." So you can name an engagement by its name, by its E-label, or by your own number for it — whichever comes out of your mouth. Honest note on cost: "open 7" by-number doesn't work anywhere today, so we're building that lookup fresh as part of this. It's a bit more than the smaller option would've been, but it makes all three naming styles feel equal — and it folds into this same piece of work.
Decision 2 — What happens when things are unclear? → defaults
Two small sub-questions:
When does your current spot get cleared? When you leave? Only when you explicitly say "close"? Or does it just stay until you open something else?
If the Companion has no current spot AND you didn't name an engagement, what should it do? Guess your Personal space? Or stop and ask?
Decided — the defaultsYour spot stays put until you open something else or say "close." You stay where you were until you choose to move.
If it doesn't know where you are, it asks rather than guesses. Guessing risks putting a commit in the wrong engagement — worse than a quick "which one?"
Why "ask, don't guess" matters
You finish in Goosey, the spot is cleared, and you say "commit held 14" out of the blue. Guessing might file that commit into your Personal space — wrong place, quietly. Asking — "which engagement?" — costs you one breath and never misfiles.
Does this clash with anything settled?
No. The Loomworks seed (the foundation document) doesn't speak to "where am I" memory at all — that lives a level below it, in the technical specification, which is the right home for it. The one place this touches settled ground is how an engagement is named (by name, by your number, by E-label), and that was already decided in the engagement-identity record earlier this month. This note follows that decision; it doesn't reopen it.
One honesty note: I wrote this without the very latest copy of the seed in front of me (I had an older copy). Because this topic sits below the seed no matter which version, the conclusion holds — but whoever writes the detailed build plan should give the current seed a quick last look to confirm nothing new contradicts it.
What happens next
The two choices are made — 1B and the defaults (recorded above).
The note gets sent up to the shared record — only when you say so. (It's saved and committed already; it is NOT sent yet, by your instruction.)
A fresh session writes the detailed build plan from this note. The groundwork's already done, so it starts from solid footing.