Inventory nodespace-docs: what is live, what is superseded, and where each page goes #3728

Re-inventory before beginning the rebuild work. This inventory is a snapshot taken on 2026-10-11 and the docs and code keep moving: pages have been added since (development/windows-testing.md, given a row by hand), and every "Code today" cell was true of main on that date only. Before any node is written from development/docs-inventory.md (the rebuild in #3727 and its sub-issues), repeat the inventory step: re-check each page's status against the code, add rows for pages that did not exist, refresh the counts, and put the changes to the user as a decision where a destination changes. Until then, treat a row as a lead, not as a fact.

Overview

The first step of rebuilding the design record in NodeSpace (#3727): a page-by-page inventory of nodespace-docs that says what is still true, what is dead, and where each live piece belongs in the graph. Nothing is migrated here. This can be done before sync exists.

Problem Statement

Nobody knows how much of the 206 non-archived pages is live. Decisions have been superseded and amended in place, pages describe designs ahead of the code, and several pages restate the same thing. Without an inventory the rebuild cannot be sized or split.

Proposed Solution

A table, one row per page (per section for a long page), with:

• status: live, superseded, duplicate of another page, ahead of the code, or dead;
• for a decision record: the decision as it stands today in one or two sentences, with its amendments folded in, and the records it replaced;
• destination: a decision node, a skill attached to the project, a reference document, a spec for work still ahead, or not migrated;
• what it should link to (the type, the area of code, the decisions it depends on).

Then a short proposal: the collections the reference documents go in, and whether architecture and component references need anything the existing types do not give.

Documentation

• ADR-092 §1 to §4 - the spec, plan, decision and project types the record is rebuilt in
• ADR-094 "The shipped workflow" - standards are skills attached to a project; a node is read with what governs it
• documentation.md - the rule that all design lives in nodespace-docs, which this work replaces

Acceptance Criteria

• Every page under decisions/, architecture/, components/, development/ and operations/ has a row
• strategy/ and archived/ are covered by a rule (kept or not) with the exceptions named
• Each live decision is stated in its current form, and each superseded one names what replaced it
• Pages that are ahead of the code are marked, with what the code does today
• The count of decision nodes, skills and reference documents the rebuild will produce is given
• The proposal on collections and on whether a reference type is needed is put to the user as a decision, and the answer is recorded
• The inventory is repeated against main before any dependent issue begins: each page's status re-checked, rows added for pages that did not exist, counts refreshed, and any changed destination put to the user as a decision

Technical Specifications

Reference Files

• nodespace-docs: decisions/ (91 files), architecture/ (35), components/ (14), development/ (19), operations/ (2), strategy/ (47), archived/ (152)
• packages/core/src/models/core_schemas.rs - the types available

Implementation Details

• Check a claim against the code before calling a page live; a page that reads as current may describe a target.
• The inventory is a working document and lives in nodespace-docs until the rebuild is done.

Related Issues

• Rebuild NodeSpace's own design record in NodeSpace from the current state, not by converting the docs #3727 (parent)

Non-Goals

• Writing any node into the graph
• Editing the docs to fix what the inventory finds